Enterprise Information Systems Concepts, Methodologies, Tools and Applications

Enterprise Information
Systems:
Concepts, Methodologies,
Tools and Applications
Information Resources Management Association

USA
Volume I
Business science reference

Hershey • New York
Director of Editorial Content: Kristin Klinger
Director of Book Publications: Julia Mosemann
Acquisitions Editor: Lindsay Johnston
Development Editor: Devvin Earnest
Publishing Assistant: Deanna Jo Zombro
Typesetters: Michael Brehm, Casey Conapitski, Keith Glazewski, Natalie Pronio,
Milan Vracarich, JR., Deanna Zombro
Production Editor: Jamie Snavely
Cover Design: Lisa Tosheff
Published in the United States of America by

Information Science Reference (an imprint of IGI Global)
701 E. Chocolate Avenue
Hershey PA 17033
Tel: 717-533-8845
Fax: 717-533-8661
E-mail: cust@igi-global.com
Web site: http://www.igi-global.com/reference
and in the United Kingdom by

Information Science Reference (an imprint of IGI Global)
3 Henrietta Street
Covent Garden
London WC2E 8LU
Tel: 44 20 7240 0856
Fax: 44 20 7379 0609
Web site: http://www.eurospanbookstore.com
Copyright © 2011 by IGI Global. All rights reserved. No part of this publication may be reproduced, stored or distributed in any form or by
any means, electronic or mechanical, including photocopying, without written permission from the publisher.
Product or company names used in this set are for identification purposes only. Inclusion of the names of the products or companies
does not indicate a claim of ownership by IGI Global of the trademark or registered trademark.
Enterprise information systems : concepts, methodologies, tools and applications / Information Resources Man-
agement Association, editor. p. cm.
Includes bibliographical references and index.
Summary: "This three-volume collection provides a complete assessment of the latest developments in enterprise
information systems research, including development, design, and emerging methodologies"--Provided by publisher.
ISBN 978-1-61692-852-0 (hardcover) -- ISBN 978-1-61692-853-7 (ebook)
1. Management information systems. 2. Information technology--Management. 3. Electronic commerce. 4.
Business enterprises--Computer networks.
I. Information Resources Management Association.
HD30.213.E582 2011
658.4'038011--dc22
2010032232
British Cataloguing in Publication Data

A Cataloguing in Publication record for this book is available from the British Library.
All work contributed to this book set is original material. The views expressed in this book are those of the authors, but not necessarily of
the publisher.
Editor-in-Chief
Mehdi Khosrow-Pour, DBA
Editor-in-Chief
Contemporary Research in Information Science and Technology, Book Series
Associate Editors
Steve Clarke
University of Hull, UK
Murray E. Jennex
San Diego State University, USA
Annie Becker
Florida Institute of Technology USA
Ari-Veikko Anttiroiko
University of Tampere, Finland
Editorial Advisory Board

Sherif Kamel
American University in Cairo, Egypt
In Lee
Western Illinois University, USA
Jerzy Kisielnicki
Warsaw University, Poland
Keng Siau
University of Nebraska-Lincoln, USA
Amar Gupta
Arizona University, USA
Craig van Slyke

University of Central Florida, USA
John Wang
Montclair State University, USA
Vishanth Weerakkody
Brunel University, UK
Additional Research Collections found in the
“Contemporary Research in Information Science and Technology”
Book Series
Data Mining and Warehousing: Concepts, Methodologies, Tools, and Applications
John Wang, Montclair University, USA • 6-volume set • ISBN 978-1-60566-056-1
Electronic Business: Concepts, Methodologies, Tools, and Applications

In Lee, Western Illinois University • 4-volume set • ISBN 978-1-59904-943-4
Electronic Commerce: Concepts, Methodologies, Tools, and Applications

S. Ann Becker, Florida Institute of Technology, USA • 4-volume set • ISBN 978-1-59904-943-4
Electronic Government: Concepts, Methodologies, Tools, and Applications

Ari-Veikko Anttiroiko, University of Tampere, Finland • 6-volume set • ISBN 978-1-59904-947-2
Knowledge Management: Concepts, Methodologies, Tools, and Applications

Murray E. Jennex, San Diego State University, USA • 6-volume set • ISBN 978-1-59904-933-5
Information Communication Technologies: Concepts, Methodologies, Tools, and Applications

Craig Van Slyke, University of Central Florida, USA • 6-volume set • ISBN 978-1-59904-949-6
Intelligent Information Technologies: Concepts, Methodologies, Tools, and Applications

Vijayan Sugumaran, Oakland University, USA • 4-volume set • ISBN 978-1-59904-941-0
Information Security and Ethics: Concepts, Methodologies, Tools, and Applications

Hamid Nemati, The University of North Carolina at Greensboro, USA • 6-volume set • ISBN 978-1-59904-937-3
Medical Informatics: Concepts, Methodologies, Tools, and Applications

Joseph Tan, Wayne State University, USA • 4-volume set • ISBN 978-1-60566-050-9
Mobile Computing: Concepts, Methodologies, Tools, and Applications

David Taniar, Monash University, Australia • 6-volume set • ISBN 978-1-60566-054-7
Multimedia Technologies: Concepts, Methodologies, Tools, and Applications

Syed Mahbubur Rahman, Minnesota State University, Mankato, USA • 3-volume set • ISBN 978-1-60566-054-7
Virtual Technologies: Concepts, Methodologies, Tools, and Applications

Jerzy Kisielnicki, Warsaw University, Poland • 3-volume set • ISBN 978-1-59904-955-7
Free institution-wide online access with the purchase of a print collection!
information science reference

Hershey • New York
Order online at www.igi-global.com or call 717-533-8845 ext.100
Mon–Fri 8:30am–5:00 pm (est) or fax 24 hours a day 717-533-7115
List of Contributors
Aalmink, Jan \ Carl von Ossietzky University Oldenburg, Germany .............................................. 134
Aguirre, Jose-Luis \ Tecnologico de Monterrey, Mexico ............................................................... 1836
Ajaefobi, Joseph \ Loughborough University, UK ........................................................................... 370
Aklouf, Youcef \ University of Science and Technology, Algeria ..................................................... 329
Aktas, Mehmet S. \ TUBITAK (Turkish National Science Foundation), Turkey .................................. 1
Ali, A. K. Hairul Nizam Pengiran Haji \ Staffordshire University, UK ....................................... 1162
Al-Ibraheem, Nawaf \ KNET, Kuwait .............................................................................................. 752
Alizai, Fahd \ Victoria University, Australia .................................................................................... 487
Alonso, Juan Ignacio Guerrero \ University of Seville, Spain ........................................................ 472
Altintas, N. Ilker \ Cybersoft Information Technologies, Turkey ................................................... 1422
Anbuudayasankar, S. P. \ Amrita School of Engineering, India ................................................... 1537
Antlova, Klara \ Technical University, Czech Republic ................................................................. 1573
Argyropoulou, Maria \ Brunel University, UK .............................................................................. 1447
Arora, Hina \ Arizona State University, USA ................................................................................... 584
Asprey, Len \ Practical Information Management Solutions Pty Ltd, Australia ............................ 1470
Atkins, Anthony S. \ Staffordshire University, UK ......................................................................... 1162
Averweg, Udo \ Information Services, eThekwini Municipality & University of KwaZulu-Natal,
South Africa . 706
Balasundaram, S.R. \ National Institute of Technology, Tiruchirappalli, India .............................. 279
Barik, Mridul Sankar \ Jadapur University, India .......................................................................... 154
Barroso, João \ Instituto Superior de Engenharia do Porto, Portugal .......................................... 1769
Batenburg, Ronald \ Utrecht University, The Netherlands ............................................................ 1349
Ben-Abdallah, Hanene \ Mir@cl Laboratory, Faculté des Sciences Economiques et de Gestion,
Tunisia ......... 427
Boehm, Barry \ University of Southern California, USA ................................................................. 986
Bokma, Albert \ University of Sunderland, UK ................................................................................ 550
Bondarouk, Tanya \ University of Twente, The Netherlands ......................................................... 1379
Boonstra, Albert \ University of Groningen, The Netherlands ...................................................... 1480
Boudreau, Marie-Claude \ University of Georgia, USA ............................................................... 1496
Brahe, Steen \ Danske Bank, Denmark ............................................................................................. 835
Brainin, Esther \ Ruppin Academic Center, Israel ......................................................................... 1295
Brena, Ramón \ Tecnologico de Monterrey, Mexico ...................................................................... 1836
Bruno, Giorgio \ Politecnico di Torino, Italy ................................................................................. 1247
Brydon, Michael \ Simon Fraser University, Canada .................................................................... 1399
Buccella, Agustina \ Universidad Nacional del Comahue, Argentina ............................................. 207
Burbach, Ralf \ Institute of Technology Carlow, Ireland ............................................................... 1370
Burgard, Martin \ Saarland University, Germany ......................................................................... 1013
Burgess, Stephen \ Victoria University, Australia ............................................................................ 487
Bussler, Christoph \ Merced Systems Inc., USA ............................................................................... 128
Čančer, Vesna \ University of Maribor, Slovenia ............................................................................ 1871
Cechich, Alejandra \ Universidad Nacional del Comahue, Argentina ............................................ 207
Cetin, Semih \ Cybersoft Information Technologies, Turkey .......................................................... 1422
Chang, Cheng-Chun \ National Chung Cheng University, Taiwan ................................................. 452
Chang, Shuchih Ernest \ National Chung Hsing University, Taiwan ............................................. 714
Chen, Jinjun \ Swinburne University of Technology, Australia ..................................................... 1081
Chen, Ning \ Xi’an Polytechnic University, China ........................................................................... 544
Chen, Qiyang \ Montclair State University, USA ............................................................................. 533
Chen, Ruey-Shun \ China University of Technology, Taiwan ........................................................ 1605
Cheng, Chen-Yang \ Penn State University, USA .............................................................................. 86
Chia Cua, Francisco \ Otago Polytechnic, New Zealand ................................................................ 346
Chiu, Dickson K.W. \ Dickson Computer Systems, Hong Kong .................................................... 1902
Choudhury, Islam \ London Metropolitan University, UK ............................................................ 1646
Cua, Francisco Chia \ University of Otago, New Zealand ............................................................ 1663
Dan, Wang \ Harbin Institute of Technology, China ....................................................................... 1021
Daneva, Maya \ University of Twente, the Netherlands ................................................................. 1941
Dave, Sumita \ Shri Shankaracharya Institute of Management & Technology, India .................... 1183
Davis, Ashley \ University of Georgia, USA ..................................................................................... 776
de Carvalho Costa, Rogério Luís \ University of Coimbra, Portugal ............................................ 901
de Carvalho, Rogerio Atem \ Federal Center for Technological Education of Campos,
Brazil ............. 99
de Cesare, Sergio \ Brunel University, UK ..................................................................................... 1646
Di Florido, Emily \ Brunel University, UK ..................................................................................... 1646
Díaz, Angel \ Instituto de Empresa Business School, Spain .............................................................. 648
Drias, Habiba \ University of Science and Technology, Algeria ...................................................... 329
Dundon, Tony \ National University of Ireland-Galway, Ireland ................................................... 1370
Dunn, Cheryl L. \ Grand Valley State University, USA .................................................................... 189
Eder, Johann \ University of Klagenfurt, Austria ............................................................................. 566
Eggert, Sandy \ University of Potsdam, Germany .................................................................. 946, 1265
Engbers, Sander \ COGAS BV. Business Unit Infra & Networkmanagement, The
Netherlands ............................................................................................................................... 1379
Erwin, Geoff \ Cape Peninsula University of Technology, South Africa .......................................... 706
Faulkner, Stéphane \ University of Namur Rempart de la Vierge, Belgium .................................. 1141
Feki, Jamel \ Mir@cl Laboratory, Faculté des Sciences Economiques et de Gestion,
Tunisia ......... 427
Ferran, Carlos \ The Pennsylvania State University, USA ............................................................. 1816
Foster, Steve \ University of Hertfordshire and NorthgateArinso, UK ............................................. 250
Furtado, Pedro \ University of Coimbra, Portugal .......................................................................... 901
Ganesh, K. \ Global Business Services – Global Delivery, IBM India Private Limited,
India ........... 1537
Garito, Marco \ Digital Business, Italy ............................................................................................ 823
Garrett, Tony C. \ Korea University, Republic of Korea ........................................................ 346, 1663
Gerard, Gregory J. \ Florida State University, USA ....................................................................... 189
Gerritsen, Bart H. M. \ TNO Netherlands Organization for Applied Scientific Research, The
Netherlands . 921
Ghanbary, Abbass \ MethodScience.com & University of Western Sydney, Australia .................... 668
Ghani, Shehzad Khalid \ Prince Sultan University, Saudi Arabia ................................................ 1960
Gibson, Candace J. \ University of Western Ontario,Canada ......................................................... 292
Goicoechea, Iñigo Monedero \ University of Seville, Spain ............................................................ 472
Gómez, Jorge Marx \ Carl von Ossietzky University Oldenburg, Germany ........................... 134, 508
González-Benito, Óscar \ University of Salamanca, Spain ........................................................... 1730
Grabski, Severin V. \ Michigan State University, USA .................................................................... 189
Green, Rolf \ OneView Pty Ltd, Australia ....................................................................................... 1470
Grieger, Martin \ Accenture, Germany ............................................................................................ 638
Gronau, Norbert \ University of Potsdam, Germany ............................................................. 946, 1265
Gulla, Jon Atle \ The Norwegian University of Science and Technology, Norway .......................... 866
Gunasekaran, Angappa \ University of Massachusetts—Dartmouth, USA ...................................... 21
Gurau, Calin \ GSCM – Montpellier Business School, France ...................................................... 1327
Hachaichi, Yasser \ Mir@cl Laboratory, Faculté des Sciences Economiques et de Gestion,
Tunisia ......... 427
Hajnal, Ákos \ Computer and Automation Research Institute, Hungary ......................................... 972
Halgeri, Pritish \ Kansas State University, USA ............................................................................. 1121
Hartmann, Evi \ SMI Supply Management Institute, Germany ....................................................... 638
Hauc, Anton \ University of Maribor, Slovenia .............................................................................. 1871
Heemstra, Fred \ Open University Nederland and KWD Result management, The
Netherlands ............................................................................................................................... 1847
Helms, Marilyn M. \ Dalton State College, USA ........................................................................... 1605
Hilpert, Ditmar \ Reutlingen University, Germany ........................................................................ 1924
Holmström, Jonny \ Umeå University, Sweden ............................................................................. 1496
Holsapple, Clyde W. \ University of Kentucky, USA ...................................................................... 1099
Hu, Haiyang \ Zhejiang Gongshang University, China .................................................................. 1902
Hu, Hua \ Zhejiang Gongshang University, China ......................................................................... 1902
Huang, Shi-Ming \ National Chung Cheng University, Taiwan ....................................................... 452
Hung, Patrick C. K. \ University of Ontario Institute of Technology, Canada .............................. 1902
Hwang, Mark I. \ Central Michigan University, USA .................................................................... 1657
Ignatiadis, Ioannis \ University of Bath, UK .................................................................................. 1209
Ingvaldsen, Jon Espen \ The Norwegian University of Science and Technology, Norway .............. 866
Ioannou, George \ Athens University of Economics and Business, Greece ................................... 1447
Janssens, Guy \ Open University Nederland, The Netherlands ..................................................... 1847
Jih, Wen-Jang (Kenny) \ Middle Tennessee State University, USA ............................................... 1605
Jiménez-Zarco, Ana Isabel \ Open University of Catalonia, Spain .............................................. 1730
Kabene, Stefane M. \ University of Western Ontario, Canada ........................................................ 292
Kelzenberg, Kai \ RWTH Aachen University, Germany ..................................................................... 68
Kerr, Don \ University of the Sunshine Coast, Australia ................................................................ 1748
Khan, Khaled M. \ Qatar University, Qatar .................................................................................. 1113
Kidd, Paul T. \ Cheshire Henbury, UK ............................................................................................. 314
Kifor, Tamás \ Computer and Automation Research Institute, Hungary .......................................... 972
Kimble, Chris \ Euromed Marseille École de Management, France ................................................. 35
King, Lisa \ University of Western Ontario, Canada ........................................................................ 292
Kolp, Manuel \ Université Catholique de Louvain Place des Doyens, Belguim ............................ 1141
Koopman, Gerwin \ Syntess Software, The Netherlands ............................................................... 1349
Kotzab, Herbert \ Copenhagen Business School, Denmark ............................................................ 638
Koufopoulos, Dimitrios N. \ Brunel University, UK ...................................................................... 1447
Koumpis, Adamantios \ ALTEC S.A., Greece ................................................................................ 1593
Krcmar, Helmut \ Technische Universität München, Germany ....................................................... 169
Krishnankutty, K. V. \ College of Engineering, Trivandrum, India .............................................. 1960
Krogstie, John \ Norwegian University of Science and Technology (NTNU), Norway, &
SINTEF ICT, Norway .................................................................................................................. 731
Kumar, Kuldeep \ Florida International University, USA ............................................................. 1060
Kumta, Gita A. \ SVKM’s NMIMS University, School of Business Management, Mumbai,
India ............. 112
Kusters, Rob \ Eindhoven University of Technology and Eindhoven University of Technology,
The Netherlands ........................................................................................................................ 1847
Lal, Shikha \ Banaras Hindu University (BHU), India .................................................................. 1553
Lämmer, Anne \ sd&m AG, and University of Potsdam, Germany ........................................ 946, 1265
Lane, Jo Ann \ University of Southern California, USA .................................................................. 986
Lastres-Segret, José A. \ University of La Laguna, Spain ............................................................. 1341
Lee, Tzong-Ru \ National Chung Hsing University, Taiwan, ROC ................................................ 1537
Lei, Chang \ Harbin Institute of Technology, China ....................................................................... 1021
León de Mora, Carlos \ University of Seville, Spain ....................................................................... 472
Li, Shing-Han \ Tatung University, Taiwan ...................................................................................... 452
Li, Zhang \ Harbin Institute of Technology, China ......................................................................... 1021
Liang, Xiaoya \ Fudan University, China ......................................................................................... 617
Lin, Chad \ Curtin University of Technology, Australia ................................................................. 1030
Lin, Koong \ Tainan National University of the Arts, Taiwan ........................................................ 1030
Loonam, John \ Dublin City University, Ireland ............................................................................ 1631
Lorenzo, Oswaldo \ Instituto de Empresa Business School, Spain .................................................. 648
Lu, June \ University of Houston-Victoria, USA ............................................................................... 533
Lübke, Daniel \ Leibniz Universität Hannover, Germany ................................................................ 508
Lukácsy, Gergely \ Budapest University of Technology and Economics, Hungary ......................... 972
Mabert, Vincent A. \ Indiana University, USA ............................................................................... 1924
Makuch, Paul \ Institute for Information Systems at German Research Centre for Artificial
Intelligence, Germany ................................................................................................................. 817
Martínez-Ruiz, María Pilar \ University of Castilla-La Mancha, Spain ...................................... 1730
Mathrani, Sanjay \ Massey University, New Zealand .............................................................. 53, 1233
Mazumdar, Chandan \ Jadavpur University, India ......................................................................... 154
McDonagh, Joe \ University of Dublin, Trinity College, Ireland ................................................... 1631
McGaughey, Ronald E. \ University of Central Arkansas, USA ........................................................ 21
McHaney, Roger \ Kansas State University, USA .......................................................................... 1121
McLaughlin, Stephen \ University of Glasgow, UK ...................................................................... 1513
Meza, Justin \ HP Labs, USA ........................................................................................................... 267
Middleton, Michael \ Queensland University of Technology, Australia ........................................ 1470
Millán, Rocío \ University of Seville, Spain ...................................................................................... 472
Millham, Richard C. \ Catholic University of Ghana, Ghana ......................................................... 181
Mishra, Alok \ Atilim University, Turkey .............................................................................. 1279, 1318
Modrák, Vladimír \ Technical University of Košice, Slovakia ........................................................ 625
Mohan, Ashutosh \ Banaras Hindu University (BHU), India ........................................................ 1553
Mohandas, K. \ Amrita School of Engineering, India .................................................................... 1537
Møller, Charles \ Aalborg University, Denmark ............................................................................. 1789
Motwani, Jaideep \ Grand Valley State University, USA ............................................................... 1447
Moynihan, Gary P. \ The University of Alabama, USA ................................................................... 235
Nair, Prashant R. \ Amrita University, Coimbatore, India ............................................................... 596
Nandhakumar, Joe \ University of Warwick, UK ........................................................................... 1209
Nicolescu, Valentin \ Technische Universität München, Germany ................................................... 169
Núñez-Gorrín, José M. \ University of La Laguna, Tenerife, Spain .............................................. 1341
Pahl, Claus \ Dublin City University, Ireland ................................................................................... 997
Paige, Richard \ University of York, UK ............................................................................................. 35
Papajorgji, Petraq \ Center for Applied Optimization, University of Florida, USA ........................ 687
Pardalos, Panos M. \ Center for Applied Optimization, University of Florida, USA ...................... 687
Parthasarathy, S. \ Thiagarajar College of Engineering, India .................................................... 1172
Pei, Z. J. \ Kansas State University, USA ........................................................................................ 1121
Perko, Igor \ University of Maribor, Slovenia ................................................................................ 1871
Petkov, Don \ Eastern Connecticut State University, USA ............................................................... 706
Piazza, Franca \ Saarland University, Germany ............................................................................ 1013
Prabhu, Vittal \ Penn State Unviersity, USA ...................................................................................... 86
Protogeros, Nikos \ University of Macedonia, Greece ................................................................... 1593
Put, Dariusz \ Cracow University of Economics, Poland ............................................................... 1039
Raghu, T.S. \ Arizona State University, USA .................................................................................... 584
Rahayu, Wenny \ La Trobe University, Australia ............................................................................ 879
Rahimifard, Aysin \ Loughborough University, UK ......................................................................... 370
Ramadoss, B. \ National Institute of Technology, Tiruchirappalli, India ......................................... 279
Ramamohanarao, Kotagiri \ University of Melbourne, Australia ................................................ 1081
Rashid, Mohammad A. \ Massey University, New Zealand .................................................... 53, 1233
Reimers, Kai \ RWTH Aachen University, Germany .......................................................................... 68
Ruël, Huub \ University of Twente, The Netherlands & American University of Beirut,
Lebanon ............................................................................................................................. 752, 1715
Rusu, Laura Irina \ La Trobe University, Australia ......................................................................... 879
Salaka, Vamsi \ Penn State Unversity, USA ........................................................................................ 86
Salim, Ricardo \ Universidad Autónoma de Barcelona, Spain, & Cautus Networks Corp.,
Venezuela ... 1816
Sammon, David \ University College Cork, Ireland ...................................................................... 1738
Schoenherr, Tobias \ Michigan State University, USA ................................................................... 1924
Sedera, Darshana \ Queensland University of Technology, Australia ............................................. 958
Sengupta, Anirban \ Jadavpur University, India ............................................................................. 154
Shakir, Maha \ Zayed University, United Arab Emirates ............................................................... 1797
Sharpanskykh, Alexei \ Vrije Universiteit Amsterdam, The Netherlands ...................................... 1196
Sherringham, Keith \ IMS Corp, Australia ............................................................................. 795, 805
Shrivastava, Monica \ Shri Shankaracharya Institute of Management & Technology,
India ........... 1183
Sindre, Guttorm \ Norwegian University of Science and Technology (NTNU), Norway ................ 731
Singla, Ashim Raj \ Indian Institute of Foreign Trade, New Delhi. India ..................................... 1617
Skytøen, Øyvind \ Norwegian University of Science and Technology (NTNU), Norway ................ 731
Soja, Piotr \ Cracow University of Economics, Poland .................................................................. 1039
Soni, Ashok K. \ Indiana University, USA ...................................................................................... 1924
Subramoniam, Suresh \ Prince Sultan University, Saudi Arabia .................................................. 1960
Sun, Chia-Ming \ National Yunlin University of Science & Technology, Taiwan .......................... 1605
Tabatabaie, Malihe \ University of York, UK ..................................................................................... 35
Taniar, David \ Monash University, Australia .................................................................................. 879
Tansel, Abdullah Uz \ Baruch College – CUNY, USA ................................................................... 1461
Targowski, Andrew \ Western Michigan University, USA ............................................................... 397
Tektonidis, Dimitrios \ ALTEC S.A., Greece .................................................................................... 550
ter Horst, Vincent \ Saxion Knowledge Center Innovation and Entrepreneurship,The
Netherlands ............................................................................................................................... 1379
Tounsi, Mohamed \ Prince Sultan University, Saudi Arabia ......................................................... 1960
Trigo, Antonio \ Escola Superior de Tecnologia e Gestão de Oliveira do Hospital,
Portugal ..... 1769
Triviño, Félix Biscarri \ University of Seville, Spain ....................................................................... 472
Triviño, Jesús Biscarri \ University of Seville, Spain ...................................................................... 472
Trojer, Thomas \ University of Innsbruck, Austria ......................................................................... 1902
Tufekci, Ozgur \ Cybersoft Information Technologies, Turkey ...................................................... 1422
Unhelkar, Bhuvan \ MethodScience.com & University of Western Sydney,
Australia ...................................................................................................... 356, 522, 668, 795, 805
Unruh, Amy \ University of Melbourne Australia .......................................................................... 1081
Valerio, Gabriel \ Tecnologico de Monterrey, Mexico .................................................................... 1836
Varajão, João \ Centro Algoritmi and University of Trás-os-Montes e Alto Douro,
Portugal ..... 1769
Varga, László Z. \ Computer and Automation Research Institute, Hungary .................................... 972
Varga, Mladen \ University of Zagreb, Croatia .............................................................................. 1695
Venkataramanan, M.A. \ Indiana University, USA ....................................................................... 1924
Veres, Csaba \ Norwegian University of Science and Technology (NTNU), Norway ....................... 731
Viehland, Dennis \ Massey University, New Zealand ............................................................... 53, 1233
Vining, Aidan R. \ Simon Fraser University, Canada .................................................................... 1399
Vinze, Ajay \ Arizona State University, USA .................................................................................... 584
Vrecko, Igor \ University of Maribor, Slovenia .............................................................................. 1871
Wagner, Thomas \ RWTH Aachen University, Germany ................................................................... 68
Wang, John \ Montclair State University, USA ................................................................................ 533
Wang, Mingzhong \ University of Melbourne, Australia ............................................................... 1081
Wang, Minhong \ The University of Hong Kong, Hong Kong ....................................................... 1060
Wautelet, Yves \ Université Catholique de Louvain Place des Doyens, Belguim .......................... 1141
Werth, Dirk \ Institute for Information Systems at German Research Centre for Artificial
Intelligence, Germany ................................................................................................................. 817
Weston, Richard \ Loughborough University, UK ........................................................................... 370
Wheeler, Zachary B. \ SDDM Technology, USA .............................................................................. 217
Wiggisser, Karl \ University of Klagenfurt, Austria ......................................................................... 566
Wittges, Holger \ Technische Universität München, Germany ........................................................ 169
Wu, Jiming \ California State University-East Bay, USA .............................................................. 1099
Wu, Ming-Chien (Mindy) \ University of Western Sydney, Australia ............................................. 356
Yao, James \ Montclair State University, USA .................................................................................. 533
Yen, David C. \ Miami University, USA ............................................................................................ 452
Zhu, Qin \ HP Labs, USA .................................................................................................................. 267
Zhu, Yaoling \ Dublin City University, Ireland ................................................................................. 997
Zhuang, Yi \ Zhejiang Gongshang University, China .................................................................... 1902
Contents
Volume ,
Section I. Fundamental Concepts and Theories
This section serves as the foundation for this exhaustive reference tool by addressing crucial theories
essential to the understanding of enterprise information systems. Chapters found within these pages
provide an excellent framework in which to position enterprise information systems within the field of
information science and technology. Individual contributions provide overviews of the history of enter-
prise information systems, the impact of information systems on organizations, and overviews on vari-
ous enterprise information system processes such as enterprise resource planning and decision support
systems. Within this introductory section, the reader can learn and choose from a compendium of expert
research on the elemental theories underscoring enterprise information systems.
Chapter 1.1. Principles and Experiences: Designing and Building Enterprise Information
Systems ................... 1
Mehmet S. Aktas, TUBITAK (Turkish National Science Foundation), Turkey
Chapter 1.2. Evolution of Enterprise Resource Planning ..................................................................... 21

Ronald E. McGaughey, University of Central Arkansas, USA
Angappa Gunasekaran, University of Massachusetts—Dartmouth, USA
Chapter 1.3. Exploring Enterprise Information Systems ...................................................................... 35

Malihe Tabatabaie, University of York, UK
Richard Paige, University of York, UK
Chris Kimble, Euromed Marseille École de Management, France
Chapter 1.4. Enterprise Systems in Small and Medium-Sized Enterprises .......................................... 53

Sanjay Mathrani, Massey University, New Zealand
Mohammad A. Rashid, Massey University, New Zealand
Dennis Viehland, Massey University, New Zealand
Chapter 1.5. A Conceptual Framework for Developing and Evaluating ERP Implementation
Strategies in Multinational Organizations............................................................................................. 68
Kai Kelzenberg, RWTH Aachen University, Germany
Thomas Wagner, RWTH Aachen University, Germany
Kai Reimers, RWTH Aachen University, Germany
Chapter 1.6. Integrated Research and Training in Enterprise Information Systems ............................. 86
Chen-Yang Cheng, Penn State University, USA
Vamsi Salaka, Penn State Unversity, USA
Vittal Prabhu, Penn State Unviersity, USA
Chapter 1.7. Free and Open Source Enterprise Resources Planning .................................................... 99
Rogerio Atem de Carvalho, Federal Center for Technological Education of Campos,
Brazil
Chapter 1.8. E-Government and ERP: Challenges and Strategies...................................................... 112

Gita A. Kumta, SVKM’s NMIMS University, School of Business Management, Mumbai,
India
Chapter 1.9. Enterprise Application Integration (EAI) ....................................................................... 128

Christoph Bussler, Merced Systems Inc., USA
Chapter 1.10. Enterprise Tomography: An Efficient Approach for Semi-Automatic Localization

of Integration Concepts in VBLAs ..................................................................................................... 134
Jan Aalmink, Carl von Ossietzky University Oldenburg, Germany
Jorge Marx Gómez, Carl von Ossietzky University Oldenburg, Germany
Chapter 1.11. Enterprise Information System Security: A Life-Cycle Approach ............................... 154
Chandan Mazumdar, Jadavpur University, India
Mridul Sankar Barik, Jadapur University, India
Anirban Sengupta, Jadavpur University, India
Chapter 1.12. From ERP to Enterprise Service-Oriented Architecture .............................................. 169

Valentin Nicolescu, Technische Universität München, Germany
Holger Wittges, Technische Universität München, Germany
Helmut Krcmar, Technische Universität München, Germany
Chapter 1.13. Data Reengineering of Legacy Systems....................................................................... 181

Richard C. Millham, Catholic University of Ghana, Ghana
Chapter 1.14. Semantically Modeled Databases in Integrated Enterprise Information Systems ....... 189
Cheryl L. Dunn, Grand Valley State University, USA
Gregory J. Gerard, Florida State University, USA
Severin V. Grabski, Michigan State University, USA
Chapter 1.15. An Overview of Ontology-Driven Data Integration .................................................... 207
Agustina Buccella, Universidad Nacional del Comahue, Argentina
Alejandra Cechich, Universidad Nacional del Comahue, Argentina
Chapter 1.16. A Fundamental SOA Approach to Rebuilding Enterprise Architecture for a Local
Government after a Disaster ............................................................................................................... 217
Zachary B. Wheeler, SDDM Technology, USA
Chapter 1.17. An Overview of Executive Information Systems ......................................................... 235

Gary P. Moynihan, The University of Alabama, USA
Chapter 1.18. Making Sense of e-HRM: Transformation, Technology and Power Relations ............ 250
Steve Foster, University of Hertfordshire and NorthgateArinso, UK
Chapter 1.19. Mix, Match, Rediscovery: A Mashup Experiment of Knowledge Organization in

an Enterprise Environment ................................................................................................................. 267
Justin Meza, HP Labs, USA
Qin Zhu, HP Labs, USA
Chapter 1.20. Testing Guidelines for Developing Quality EAI Projects ............................................ 279
S.R. Balasundaram, National Institute of Technology, Tiruchirappalli, India
B. Ramadoss, National Institute of Technology, Tiruchirappalli, India
Chapter 1.21. Technology and Human Resources Management in Health Care ................................ 292
Stefane M. Kabene, University of Western Ontario, Canada
Lisa King, University of Western Ontario, Canada
Candace J. Gibson, University of Western Ontario, Canada
Section II. Development and Design Methodologies
This section provides in-depth coverage of conceptual architectures, frameworks and methodologies
related to the design and implementation of enterprise information systems. Throughout these contribu-
tions, research fundamentals in the discipline are presented and discussed. From broad examinations
to specific discussions on particular frameworks and infrastructures, the research found within this
section spans the discipline while also offering detailed, specific discussions. Basic designs, as well as
abstract developments, are explained within these chapters, and frameworks for designing successful
decision support systems, integrating new technologies, and developing and implementing efficient
processes are included.
Chapter 2.1. Enterprise Information Systems: Aligning and Integrating Strategy, Technology,
Organization and People ..................................................................................................................... 314
Paul T. Kidd, Cheshire Henbury, UK
Chapter 2.2. An Adaptive E-Commerce Architecture for Enterprise Information Exchange ............. 329
Youcef Aklouf, University of Science and Technology, Algeria
Habiba Drias, University of Science and Technology, Algeria
Chapter 2.3. A Structured Approach to Developing a Business Case for New Enterprise
Information Systems ........................................................................................................................... 346
Francisco Chia Cua, Otago Polytechnic, New Zealand
Tony C. Garrett, Korea University, Republic of Korea
Chapter 2.4. Extending Enterprise Architecture with Mobility .......................................................... 356

Ming-Chien (Mindy) Wu, University of Western Sydney, Australia
Bhuvan Unhelkar, MethodScience.com & University of Western Sydney, Australia
Chapter 2.5. Enterprise Modelling in Support of Organisation Design and Change .......................... 370
Joseph Ajaefobi, Loughborough University, UK
Aysin Rahimifard, Loughborough University, UK
Richard Weston, Loughborough University, UK
Chapter 2.6. The Enterprise Systems Approach ................................................................................. 397

Andrew Targowski, Western Michigan University, USA
Chapter 2.7. Designing Data Marts from XML and Relational Data Sources.................................... 427
Yasser Hachaichi, Mir@cl Laboratory, Faculté des Sciences Economiques et de Gestion,
Tunisia
Jamel Feki, Mir@cl Laboratory, Faculté des Sciences Economiques et de Gestion,
Tunisia
Hanene Ben-Abdallah, Mir@cl Laboratory, Faculté des Sciences Economiques et de
Gestion, Tunisia
Chapter 2.8. Migrating Legacy Systems to Web Services Architecture ............................................. 452
Shing-Han Li, Tatung University, Taiwan
Shi-Ming Huang, National Chung Cheng University, Taiwan
David C. Yen, Miami University, USA
Cheng-Chun Chang, National Chung Cheng University, Taiwan
Chapter 2.9. EIS for Consumers Classification and Support Decision Making in a Power
Utility Database .. 472
Juan Ignacio Guerrero Alonso, University of Seville, Spain
Carlos León de Mora, University of Seville, Spain
Félix Biscarri Triviño, University of Seville, Spain
Iñigo Monedero Goicoechea, University of Seville, Spain
Jesús Biscarri Triviño, University of Seville, Spain
Rocío Millán, University of Seville, Spain
Chapter 2.10. An ERP Adoption Model for Midsize Businesses........................................................ 487

Fahd Alizai, Victoria University, Australia
Stephen Burgess, Victoria University, Australia
Chapter 2.11. Developing and Customizing Federated ERP Systems ................................................ 508
Daniel Lübke, Leibniz Universität Hannover, Germany
Jorge Marx Gómez, University Oldenburg, Germany
Chapter 2.12. Creation of a Process Framework for Transitioning to a Mobile Enterprise ............... 522
Chapter 2.13. Development and Design Methodologies in DWM ..................................................... 533

James Yao, Montclair State University, USA
John Wang, Montclair State University, USA
Qiyang Chen, Montclair State University, USA
June Lu, University of Houston-Victoria, USA
Chapter 2.14. Facilitating Design of Efficient Components by Bridging Gaps Between Data
Model and Business Process via Analysis of Service Traits of Data .................................................. 544
Ning Chen, Xi’an Polytechnic University, China
Chapter 2.15. The Utilization of Semantic Web for Integrating Enterprise Systems ......................... 550
Dimitrios Tektonidis, ALTEC S.A., Greece
Albert Bokma, University of Sunderland, UK
Section III. Tools and Technologies
This section presents extensive coverage of the technology that informs and impacts enterprise infor-
mation systems. These chapters provide an in-depth analysis of the use and development of innumer-
able devices and tools, while also providing insight into new and upcoming technologies, theories, and
instruments that will soon be commonplace. Within these rigorously researched chapters, readers are
presented with examples of the tools that facilitate and support the emergence and advancement of en-
terprise information systems. In addition, the successful implementation and resulting impact of these
various tools and technologies are discussed within this collection of chapters.
Chapter 3.1. Data Warehouse Maintenance, Evolution and Versioning ............................................. 566
Johann Eder, University of Klagenfurt, Austria
Karl Wiggisser, University of Klagenfurt, Austria
Chapter 3.2. Information Supply Chains: Restructuring Relationships, Chains, and Networks ........ 584
Hina Arora, Arizona State University, USA
T.S. Raghu, Arizona State University, USA
Ajay Vinze, Arizona State University, USA
Chapter 3.3. Benefits of Information Technology Implementations for Supply Chain Management:
An Explorative Study of Progressive Indian Companies ................................................................... 596
Prashant R. Nair, Amrita University, Coimbatore, India
Chapter 3.4. Transforming Compensation Management Practices through Web-Based Enterprise

Technologies ....... 617
Xiaoya Liang, Fudan University, China
Chapter 3.5. Business Process Management as a Critical Success Factor in EIS

Implementation ... 625
Vladimír Modrák, Technical University of Košice, Slovakia
Chapter 3.6. E-Markets as Meta-Enterprise Information Systems ..................................................... 638
Martin Grieger, Accenture, Germany
Evi Hartmann, SMI Supply Management Institute, Germany
Herbert Kotzab, Copenhagen Business School, Denmark
Chapter 3.7. Enterprise Systems as an Enabler of Fast-Paced Change: The Case of Global B2B
Procurement in Ericsson ..................................................................................................................... 648
Oswaldo Lorenzo, Instituto de Empresa Business School, Spain
Angel Díaz, Instituto de Empresa Business School, Spain
Volume ,,
Chapter 3.8. Extending Enterprise Application Integration (EAI) with Mobile and Web Services
Technologies ....... 668
Abbass Ghanbary, MethodScience.com & University of Western Sydney, Australia
Chapter 3.9. Towards a Model-Centric Approach for Developing Enterprise Information

Systems ............... 687
Petraq Papajorgji, Center for Applied Optimization, University of Florida, USA
Panos M. Pardalos, Center for Applied Optimization, University of Florida, USA
Chapter 3.10. Impact of Portal Technologies on Executive Information Systems ............................. 706
Udo Averweg, Information Services, eThekwini Municipality & University of
KwaZulu-Natal, South Africa
Geoff Erwin, Cape Peninsula University of Technology, South Africa
Don Petkov, Eastern Connecticut State University, USA
Chapter 3.11. A Voice-Enabled Pervasive Web System with Self-Optimization Capability for
Supporting Enterprise Applications .................................................................................................... 714
Shuchih Ernest Chang, National Chung Hsing University, Taiwan
Chapter 3.12. Achieving System and Business Interoperability by Semantic Web Services ............. 731
John Krogstie, Norwegian University of Science and Technology (NTNU), Norway,
& SINTEF ICT, Norway
Csaba Veres, Norwegian University of Science and Technology (NTNU), Norway
Guttorm Sindre, Norwegian University of Science and Technology (NTNU), Norway
Øyvind Skytøen, Norwegian University of Science and Technology (NTNU), Norway
Chapter 3.13. In-House vs. Off-the-Shelf e-HRM Applications......................................................... 752

Nawaf Al-Ibraheem, KNET, Kuwait
Huub Ruël, University of Twente, The Netherlands, & American University of Beirut,
Lebanon
Chapter 3.14. Enterprise Resource Planning Under Open Source Software ...................................... 776
Ashley Davis, University of Georgia, USA
Chapter 3.15. Real Time Decision Making and Mobile Technologies ............................................... 795
Keith Sherringham, IMS Corp, Australia
Chapter 3.16. Business Driven Enterprise Architecture and Applications to Support Mobile
Business .............. 805
Keith Sherringham, IMS Corp, Australia
Chapter 3.17. Mobile Technologies Extending ERP Systems ............................................................ 817

Dirk Werth, Institute for Information Systems at German Research Centre for Artificial
Intelligence, Germany
Paul Makuch, Institute for Information Systems at German Research Centre for Artificial
Intelligence, Germany
Chapter 3.18. Convergence in Mobile Internet with Service Oriented Architecture and Its Value
to Business .......... 823
Marco Garito, Digital Business, Italy
Chapter 3.19. Enterprise Specific BPM Languages and Tools ........................................................... 835
Steen Brahe, Danske Bank, Denmark
Chapter 3.20. Semantic Business Process Mining of SAP Transactions ............................................ 866
Jon Espen Ingvaldsen, The Norwegian University of Science and Technology, Norway
Jon Atle Gulla, The Norwegian University of Science and Technology, Norway
Chapter 3.21. Mining Association Rules from XML Documents....................................................... 879

Laura Irina Rusu, La Trobe University, Australia
Wenny Rahayu, La Trobe University, Australia
David Taniar, Monash University, Australia
Section IV. Utilization and Application
This section introduces and discusses the utilization and application of enterprise information systems
around the world. These particular selections highlight, among other topics, enterprise information
systems in multiple countries, data mining applications, and critical success factors of enterprise in-
formation systems implementation. Contributions included in this section provide excellent coverage of
the impact of enterprise information systems on the fabric of our present-day global village.
Chapter 4.1. QoS-Oriented Grid-Enabled Data Warehouses .............................................................. 901

Rogério Luís de Carvalho Costa, University of Coimbra, Portugal
Pedro Furtado, University of Coimbra, Portugal
Chapter 4.2. EIS Systems and Quality Management .......................................................................... 921
Bart H. M. Gerritsen, TNO Netherlands Organization for Applied Scientific Research,
The Netherlands
Chapter 4.3. A Procedure Model for a SOA-Based Integration of Enterprise Systems...................... 946
Anne Lämmer, sd&m AG, Germany
Sandy Eggert, University of Potsdam, Germany
Norbert Gronau, University of Potsdam, Germany
Chapter 4.4. Size Matters! Enterprise System Success in Medium and Large Organizations ........... 958
Darshana Sedera, Queensland University of Technology, Australia
Chapter 4.5. Web Services as XML Data Sources in Enterprise Information Integration.................. 972
Ákos Hajnal, Computer and Automation Research Institute, Hungary
Tamás Kifor, Computer and Automation Research Institute, Hungary
Gergely Lukácsy, Budapest University of Technology and Economics, Hungary
László Z. Varga, Computer and Automation Research Institute, Hungary
Chapter 4.6. System-of-Systems Cost Estimation: Analysis of Lead System Integrator Engineering
Activities ............. 986
Jo Ann Lane, University of Southern California, USA
Barry Boehm, University of Southern California, USA
Chapter 4.7. Consistency and Modularity in Mediated Service-Based Data Integration

Solutions ............. 997
Yaoling Zhu, Dublin City University, Ireland
Claus Pahl, Dublin City University, Ireland
Chapter 4.8. Data Warehouse and Business Intelligence Systems in the Context of E-HRM.......... 1013
Martin Burgard, Saarland University, Germany
Franca Piazza, Saarland University, Germany
Chapter 4.9. Implementation of ERP in Human Resource Management ......................................... 1021

Zhang Li, Harbin Institute of Technology, China
Wang Dan, Harbin Institute of Technology, China
Chang Lei, Harbin Institute of Technology, China
Chapter 4.10. A Study of Information Requirement Determination Process of an Executive

Information System........................................................................................................................... 1030
Chad Lin, Curtin University of Technology, Australia
Koong Lin, Tainan National University of the Arts, Taiwan
Chapter 4.11. Towards Identifying the Most Important Attributes of ERP Implementations........... 1039
Piotr Soja, Cracow University of Economics, Poland
Dariusz Put, Cracow University of Economics, Poland
Chapter 4.12. Challenges and Solutions for Complex Business Process Management.................... 1060
Minhong Wang, The University of Hong Kong, Hong Kong
Kuldeep Kumar, Florida International University, USA
Chapter 4.13. Multiple-Step Backtracking of Exception Handling in Autonomous Business

Process Management ........................................................................................................................ 1081
Mingzhong Wang, University of Melbourne, Australia
Jinjun Chen, Swinburne University of Technology, Australia
Kotagiri Ramamohanarao, University of Melbourne, Australia
Amy Unruh, University of Melbourne, Australia
Chapter 4.14. A Resource-Based Perspective on Information Technology, Knowledge

Management, and Firm Performance................................................................................................ 1099
Clyde W. Holsapple, University of Kentucky, USA
Jiming Wu, California State University-East Bay, USA
Chapter 4.15. A Decision Support System for Selecting Secure Web Services................................ 1113
Khaled M. Khan, Qatar University, Qatar
Chapter 4.16. ERP Systems Supporting Lean Manufacturing in SMEs ........................................... 1121
Pritish Halgeri, Kansas State University, USA
Roger McHaney, Kansas State University, USA
Z. J. Pei, Kansas State University, USA
Chapter 4.17. Specifying Software Models with Organizational Styles........................................... 1141

Manuel Kolp, Université Catholique de Louvain Place des Doyens, Belguim
Yves Wautelet, Université Catholique de Louvain Place des Doyens, Belguim
Stéphane Faulkner, University of Namur Rempart de la Vierge, Belgium
Chapter 4.18. Mobile Strategy for E-Business Solution ................................................................... 1162

Anthony S. Atkins, Staffordshire University, UK
A. K. Hairul Nizam Pengiran Haji Ali, Staffordshire University, UK
Chapter 4.19. Application of Software Metrics in EPR Projects ...................................................... 1172

S. Parthasarathy, Thiagarajar College of Engineering, India
Section V. Organizational and Social Implications
This section includes a wide range of research pertaining to the social and organizational impact of
enterprise information systems. Chapters included in this section analyze the impact of power relation-
ships in system implementation, discusses how enterprise systems can be used to support internal mar-
keting efforts, and demonstrate that perceived shared benefits, system characteristic, and the degree
of knowledge of the system are significant influences on an individual’s willingness to use enterprise
resource planning systems. The inquiries and methods presented in this section offer insight into the
implications of enterprise information systems at both a personal and organizational level, while also
emphasizing potential areas of study within the discipline.
Chapter 5.1. Optimization of Enterprise Information Systems through a ‘User Involvement
Framework in Learning Organizations’ ............................................................................................ 1183
Sumita Dave, Shri Shankaracharya Institute of Management & Technology, India
Monica Shrivastava, Shri Shankaracharya Institute of Management & Technology, India
Chapter 5.2. Authority and Its Implementation in Enterprise Information Systems ........................ 1196
Alexei Sharpanskykh, Vrije Universiteit Amsterdam, The Netherlands
Chapter 5.3. Enterprise Systems, Control and Drift ......................................................................... 1209

Ioannis Ignatiadis, University of Bath, UK
Joe Nandhakumar, University of Warwick, UK
Chapter 5.4. The Impact of Enterprise Systems on Business Value ................................................. 1233
Sanjay Mathrani, Massey University, New Zealand
Mohammad A. Rashid, Massey University, New Zealand
Dennis Viehland, Massey University, New Zealand
Chapter 5.5. People-Oriented Enterprise Information Systems ........................................................ 1247

Giorgio Bruno, Politecnico di Torino, Italy
Chapter 5.6. A SOA-Based Approach to Integrate Enterprise Systems............................................ 1265

Anne Lämmer, University of Potsdam, Germany
Sandy Eggert, University of Potsdam, Germany
Norbert Gronau, University of Potsdam, Germany
Chapter 5.7. Achieving Business Benefits from ERP Systems ......................................................... 1279
Alok Mishra, Atilim University, Turkey
Chapter 5.8. Experiences of Cultures in Global ERP Implementation ............................................. 1295

Esther Brainin, Ruppin Academic Center, Israel
Chapter 5.9. Enterprise Resource Planning Systems: Effects and Strategic Perspectives in
Organizations .... 1318
Alok Mishra, Atilim University, Turkey
Volume ,,,
Chapter 5.10. The Management of CRM Information Systems in Small B2B Service
Organisations: A Comparison between French and British Firms .................................................... 1327
Calin Gurau, GSCM – Montpellier Business School, France
Chapter 5.11. Information Technologies as a Vital Channel for an Internal E-Communication

Strategy ............. 1341
José A. Lastres-Segret, University of La Laguna, Spain
José M. Núñez-Gorrín, University of La Laguna, Tenerife, Spain
Chapter 5.12. Early User Involvement and Participation in Employee Self-Service Application
Deployment: Theory and Evidence from Four Dutch Governmental Cases .................................... 1349
Gerwin Koopman, Syntess Software, The Netherlands
Ronald Batenburg, Utrecht University, The Netherlands
Chapter 5.13. Assessing Information Technology Capability vs. Human Resource Information
System Utilization............................................................................................................................. 1370
Ralf Burbach, Institute of Technology Carlow, Ireland
Tony Dundon, National University of Ireland-Galway, Ireland
Chapter 5.14. Exploring Perceptions about the Use of e-HRM Tools in Medium Sized
Organizations .... 1379
Tanya Bondarouk, University of Twente, The Netherlands
Vincent ter Horst, Saxion Knowledge Center Innovation and Entrepreneurship,
The Netherlands
Sander Engbers, COGAS BV. Business Unit Infra & Networkmanagement,
The Netherlands
Chapter 5.15. Adoption, Improvement, and Disruption: Predicting the Impact of Open Source
Applications in Enterprise Software ................................................................................................. 1399
Michael Brydon, Simon Fraser University, Canada
Aidan R. Vining, Simon Fraser University, Canada
Section VI. Managerial Impact
This section presents contemporary coverage of the managerial implications of enterprise informa-
tion systems. Particular contributions explore relationships among information technology, knowledge
management, and firm performance, while others discuss the evaluation, adoption, and technical in-
frastructure of enterprise information systems. The managerial research provided in this section allows
administrators, practitioners, and researchers to gain a better sense of how enterprise information
systems can inform their practices and behavior.
Chapter 6.1. A Domain Specific Strategy for Complex Dynamic Processes .................................... 1422
Semih Cetin, Cybersoft Information Technologies, Turkey
N. Ilker Altintas, Cybersoft Information Technologies, Turkey
Ozgur Tufekci, Cybersoft Information Technologies, Turkey
Chapter 6.2. Measuring the Impact of an ERP Project at SMEs: A Framework and Empirical
Investigation...... 1447
Maria Argyropoulou, Brunel University, UK
George Ioannou, Athens University of Economics and Business, Greece
Dimitrios N. Koufopoulos, Brunel University, UK
Jaideep Motwani, Grand Valley State University, USA
Chapter 6.3. Managing Temporal Data ............................................................................................. 1461

Abdullah Uz Tansel, Baruch College, CUNY, USA
Chapter 6.4. Integrative Information Systems Architecture: Document & Content
Management ...... 1470
Len Asprey, Practical Information Management Solutions Pty Ltd, Australia
Rolf Green, OneView Pty Ltd, Australia
Michael Middleton, Queensland University of Technology, Australia
Chapter 6.5. Identifying and Managing Stakeholders in Enterprise Information System

Projects .............. 1480
Albert Boonstra, University of Groningen, The Netherlands
Chapter 6.6. Understanding Information Technology Implementation Failure: An Interpretive

Case Study of Information Technology Adoption in a Loosely Coupled Organization ................... 1496
Marie-Claude Boudreau, University of Georgia, USA
Jonny Holmström, Umeå University, Sweden
Chapter 6.7. Improving Supply Chain Performance through the Implementation of Process
Related Knowledge Transfer Mechanisms ....................................................................................... 1513
Stephen McLaughlin, University of Glasgow, UK
Chapter 6.8. Meta-Heuristic Approach to Solve Mixed Vehicle Routing Problem with Backhauls
in Enterprise Information System of Service Industry ..................................................................... 1537
S. P. Anbuudayasankar, Amrita School of Engineering, India
K. Ganesh, Global Business Services – Global Delivery, IBM India Private Limited,
India
K. Mohandas, Amrita School of Engineering, India
Tzong-Ru Lee, National Chung Hsing University, Taiwan, ROC
Chapter 6.9. Achieving Supply Chain Management (SCM): Customer Relationship Management
(CRM) Synergy Through Information and Communication Technology (ICT) Infrastructure
in Knowledge Economy.................................................................................................................... 1553
Ashutosh Mohan, Banaras Hindu University (BHU), India
Shikha Lal, Banaras Hindu University (BHU), India
Section VII. Critical Issues
This section addresses conceptual and theoretical issues related to the field of enterprise information
systems, which include issues related to customer relationship management, critical success factors,
and business strategies. Within these chapters, the reader is presented with analysis of the most current
and relevant conceptual inquires within this growing field of study. Particular chapters address the
successes of enterprise resource planning through technology, and presents strategies for overcoming
challenges related to enterprise system adoption. Overall, contributions within this section ask unique,
often theoretical questions related to the study of enterprise information systems and, more often than
not, conclude that solutions are both numerous and contradictory.
Chapter 7.1. Preparedness of Small and Medium-Sized Enterprises to Use Information and
Communication Technology as a Strategic Tool .............................................................................. 1573
Klara Antlova, Technical University, Czech Republic
Chapter 7.2. Doing Business on the Globalised Networked Economy: Technology and Business
Challenges for Accounting Information Systems ............................................................................. 1593
Adamantios Koumpis, ALTEC S.A., Greece
Nikos Protogeros, University of Macedonia, Greece
Chapter 7.3. Factors Influencing Information System Flexibility: An Interpretive Flexibility

Perspective ........ 1605
Ruey-Shun Chen, China University of Technology, Taiwan
Chia-Ming Sun, National Yunlin University of Science & Technology, Taiwan
Marilyn M. Helms, Dalton State College, USA
Wen-Jang (Kenny) Jih, Middle Tennessee State University, USA
Chapter 7.4. Challenges in Enterprise Information Systems Implementation: An Empirical

Study ................. 1617
Ashim Raj Singla, Indian Institute of Foreign Trade, New Delhi, India
Chapter 7.5. A Grounded Theory Study of Enterprise Systems Implementation: Lessons Learned
from the Irish Health Services .......................................................................................................... 1631
John Loonam, Dublin City University, Ireland
Joe McDonagh, University of Dublin, Trinity College, Ireland
Chapter 7.6. An Object-Oriented Abstraction Mechanism for Generic Enterprise Modeling .......... 1646
Islam Choudhury, London Metropolitan University, UK
Sergio de Cesare, Brunel University, UK
Emily Di Florido, Brunel University, UK
Chapter 7.7. Integrating Enterprise Systems..................................................................................... 1657

Mark I. Hwang, Central Michigan University, USA
Chapter 7.8. Analyzing Diffusion and Value Creation Dimensions of a Business Case of Replacing
Enterprise Systems ............................................................................................................................ 1663
Francisco Chia Cua, University of Otago, New Zealand
Tony C. Garrett, Korea University, Republic of Korea
Chapter 7.9. Challenges of Data Management in Always-On Enterprise Information Systems ...... 1695
Mladen Varga, University of Zagreb, Croatia
Chapter 7.10. Studying Human Resource Information Systems Implementation using Adaptive
Structuration Theory: The Case of HRIS Implementation at Dow Chemical Company .................. 1715
Huub Ruël, University of Twente, The Netherlands, & American University of Beirut,
Lebanon
Chapter 7.11. Consequences and Strategic Implications of Networked Enterprise and Human
Resources .......... 1730
Ana Isabel Jiménez-Zarco, Open University of Catalonia, Spain
María Pilar Martínez-Ruiz, University of Castilla-La Mancha, Spain
Óscar González-Benito, University of Salamanca, Spain
Chapter 7.12. An Extended Model of Decision Making: A Devil’s Advocate Workshop ................ 1738
David Sammon, University College Cork, Ireland
Chapter 7.13. Feral Systems and Other Factors Influencing the Success of Global ERP
Implementations 1748
Don Kerr, University of the Sunshine Coast, Australia
Section VIII. Emerging Trends
This section highlights research potential within the field of enterprise information systems while ex-
ploring uncharted areas of study for the advancement of the discipline. Chapters within this section
highlight new trends in adaptive information integration, as well as the challenges faced in cross-
organizational enterprise resource planning projects. The contributions that conclude this exhaustive,
multi-volume set provide emerging trends and suggestions for future research within this rapidly ex-
panding discipline.
Chapter 8.1. Motivations and Trends for IT/IS Adoption: Insights from Portuguese
Companies......... 1769
João Varajão, Centro Algoritmi and University of Trás-os-Montes e Alto Douro, Portugal
Antonio Trigo, Escola Superior de Tecnologia e Gestão de Oliveira do Hospital, Portugal
João Barroso, Instituto Superior de Engenharia do Porto, Portugal
Chapter 8.2. Next-Generation Enterprise Systems ........................................................................... 1789

Charles Møller, Aalborg University, Denmark
Chapter 8.3. ERP Trends, Opportunities, and Challenges: A Focus on the Gulf Region in the
Middle East ....... 1797
Maha Shakir, Zayed University, UAE
Chapter 8.4. The Future of ERP and Enterprise Resource Management Systems ........................... 1816
Carlos Ferran, The Pennsylvania State University, USA
Ricardo Salim, Universidad Autónoma de Barcelona, Spain, & Cautus Networks Corp.,
Venezuela
Chapter 8.5. Next-Generation IT for Knowledge Distribution in Enterprises .................................. 1836

Ramón Brena, Tecnologico de Monterrey, Mexico
Gabriel Valerio, Tecnologico de Monterrey, Mexico
Jose-Luis Aguirre, Tecnologico de Monterrey, Mexico
Chapter 8.6. Sizing ERP Implementation Projects: An Activity-Based Approach ........................... 1847
Guy Janssens, Open University Netherland, The Netherlands
Rob Kusters, Open University Netherland, The Netherlands & Eindhoven University
of Technology, The Netherlands
Fred Heemstra, Open University Netherland, The Netherlands & KWD Result
Management, The Netherlands
Chapter 8.7. Conducting Multi-Project Business Operations in SMEs and IS Support ................... 1871
Igor Vrecko, University of Maribor, Slovenia
Anton Hauc, University of Maribor, Slovenia
Vesna Čančer, University of Maribor, Slovenia
Igor Perko, University of Maribor, Slovenia
Chapter 8.8. Flow-Based Adaptive Information Integration ............................................................ 1902

Dickson K.W. Chiu, Dickson Computer Systems, Hong Kong
Thomas Trojer, University of Innsbruck, Austria
Hua Hu, Zhejiang Gongshang University, China
Haiyang Hu, Zhejiang Gongshang University, China
Yi Zhuang, Zhejiang Gongshang University, China
Patrick C. K. Hung, University of Ontario Institute of Technology, Canada
Chapter 8.9. Enterprise System in the German Manufacturing Mittelstand ..................................... 1924
Tobias Schoenherr, Michigan State University, USA
Ditmar Hilpert, Reutlingen University, Germany
Ashok K. Soni, Indiana University, USA
M.A. Venkataramanan, Indiana University, USA
Vincent A. Mabert, Indiana University, USA
Chapter 8.10. Engineering the Coordination Requirements in Cross-Organizational ERP Projects:

A Package of Good Practices ............................................................................................................ 1941
Maya Daneva, University of Twente, The Netherlands
Chapter 8.11. ERP and Beyond......................................................................................................... 1960

Suresh Subramoniam, Prince Sultan University, Saudi Arabia
Mohamed Tounsi, Prince Sultan University, Saudi Arabia
Shehzad Khalid Ghani, Prince Sultan University, Saudi Arabia
K. V. Krishnankutty, College of Engineering, Trivandrum, India
xxvii
Preface
The data collected by organizations is growing in volume and complexity. As such, businesses are
abandoning traditional methods and relying more heavily on enterprise information systems to aid in
the analysis and utilization of time-sensitive data and organizational knowledge. Enterprise information
systems have gained in popularity and even SMEs, recognizing the competitive advantage afforded by
real-time decision support, have begun to adopt the technologies.
The growth in enterprise information system adoption makes it challenging for experts and practi-
tioners to stay informed of the field’s most up-to-date research. That is why Business Science Reference
is pleased to offer this three-volume reference collection that will empower students, researchers, and
academicians with a strong understanding of critical issues within enterprise information systems by
providing both broad and detailed perspectives on cutting-edge theories and developments. This reference
is designed to act as a single reference source on conceptual, methodological, technical, and managerial
issues, as well as provide insight into emerging trends and future opportunities within the discipline.
Enterprise Information Systems: Concepts, Methodologies, Tools and Applications is organized into
eight distinct sections that provide comprehensive coverage of important topics. The sections are: (1)
Fundamental Concepts and Theories, (2) Development and Design Methodologies, (3) Tools and Tech-
nologies, (4) Utilization and Application, (5) Organizational and Social Implications, (6) Managerial
Impact, (7) Critical Issues, and (8) Emerging Trends. The following paragraphs provide a summary of
what to expect from this invaluable reference tool.
Section 1, Fundamental Concepts and Theories, serves as a foundation for this extensive reference
tool by addressing crucial theories essential to the understanding of enterprise information systems.
Chapters such as Exploring Enterprise Information Systems by Malihe Tabatabaie, Richard Paige, and
Chris Kimble, and Enterprise Systems in Small and Medium-Sized Enterprises by Sanjay Mathrani,
Mohammad Rashid, and Dennis Viehland give an introduction and overview of enterprise information
systems in a contemporary business environment. Free and Open Source Enterprise Resources Plan-
ning by Rogerio de Carvalho adds an important dimension to the under-researched subject of free/open
source enterprise resource planning systems by comparing it to proprietary systems and highlighting its
innovative potential. Additional selections, including Location-Based Service (LBS) System Analysis
and Design by Yuni Xia, Jonathan Munson, and David Wood, and An Overview of Executive Informa-
tion Systems by Gary Moynihan focus on providing backgrounds and introductions to specific concepts
within enterprise information systems. These and several other foundational chapters provide a wealth
of expert research on the elemental concepts and ideas surrounding enterprise information systems.
Section 2, Development and Design Methodologies, presents in-depth coverage of the conceptual
design and architecture of enterprise information systems, focusing on aspects including enterprise
resource planning, service-oriented architecture, and decision support systems. Designing and imple-
menting effective processes and strategies are the focus of such chapters as Development and Design
Methodologies in DWM by James Yao, John Wang, Qiyang Chen, and June Lu, and Enterprise Model-
xxviii
ing in Support of Organisation Design and Change by Joseph Ajaefobi, Aysin Rahimifard, and Richard
Weston. An ERP Adoption Model for Midsize Businesses by Fahd Alizai and Stephen Burgess offers a
model that contains implementation processes, stages, factors, and issues associated with ERP adoption
in midsize businesses. Youcef Aklouf and Habiba Drias’s An Adaptive E-Commerce Architecture for
Enterprise Information Exchange presents architecture that allows partners to exchange information
with other organizations without modifying their own systems. With contributions from leading inter-
national researchers, this section offers copious developmental approaches and design methodologies
for enterprise information systems.
Section 3, Tools and Technologies, presents extensive coverage of the various tools and technologies
used in the development and implementation of enterprise information systems. This comprehensive
section includes such chapters as Real Time Decision Making and Mobile Technologies, by Keith Sher-
ringham and Bhuvan Unhelkar, and Mobile Technologies Extending ERP Systems by Dirk Werth and
Paul Makuch, which describe various techniques and models for using mobile technology to support
enterprise information systems. Extending Enterprise Application Integration (EAI) with Mobile and Web
Services Technologies by Abbass Ghanbary and Bhuvan Unhelkar demonstrates how the technologies of
web services open up the doors to collaborative enterprise architecture integration and service oriented
architecture resulting in business integration. Finally, chapters such as Data Warehouse Maintenance,
Evolution and Versioning by Johann Eder and Karl Wiggisser, and Hybrid Data Mining for Medical
Applications by Syed Hassan and Brijesh Verma present tools to adapt to the challenges of various data
warehousing mechanisms. In all, this section provides coverage of a variety of tools and technologies
that inform and enhance modern enterprise information systems.
Section 4, Utilization and Application, describes how enterprise information systems have been
utilized and offers insight on important lessons for their continued use and evolution. Including chap-
ters such as A Decision Support System for Selecting Secure Web Services by Khaled Khan, and EIS
Systems and Quality Management by Bart Gerritsen, this section investigates numerous methodologies
that have been proposed and enacted in enterprise information systems, as well as their results. As this
section continues, a number of case studies in the use of enterprise information systems are presented
such as Enterprise System in the German Manufacturing Mittelstand by Tobias Schoenherr, Ditmar
Hilpert, Ashok Soni, M.A. Venkataramanan, and Vincent Mabert, and Size Matters! Enterprise System
Success in Medium and Large Organizations by Darshana Sedera. Contributions found in this section
provide comprehensive coverage of the practicality and current use of enterprise information systems.
Section 5, Organizational and Social Implications, includes chapters discussing the organizational
and social impact of enterprise information systems. People-Oriented Enterprise Information Systems
by Giorgio Bruno proposes a notation system called People-Oriented Business Process Notation to solve
the problem of effectively integrating conversation and business processes. Experiences of Cultures in
Global ERP Implementation by Esther Brainin examines how cultural differences in global enterprises
effect the implementation of enterprise resource planning systems. The Impact of Enterprise Systems
on Business Value by Sanjay Mathrani, Mohammad Rashid, and Dennis Viehland explores two case
studies that illustrate how enterprise information systems implementation can impact organizational
functions. This section continues with Authority and Its Implementation in Enterprise Information
Systems by Alexei Sharpanskykh, which discusses how power dynamics effect enterprise information
systems integration and proposes a logic-based specification language for representing power relations.
Overall, these chapters present a detailed investigation of the complex relationship between individuals,
organizations and enterprise information systems.
Section 6, Managerial Impact, presents focused coverage of enterprise information systems as it
relates to improvements and considerations in the workplace. Managing Temporal Data by Abdullah
Tansel addresses modeling and design issues related to temporal databases. Other chapters, such as
xxix
Integrative Information Systems Architecture by Len Asprey, discuss management considerations, docu-
ment and web content management, and the technical infrastructure supporting these systems. In all, the
chapters in this section offer specific perspectives on how managerial perspectives and developments in
enterprise information systems inform each other to create more meaningful user experiences.
Section 7, Critical Issues, addresses vital issues related to enterprise information systems, which
include customer relationship management, critical success factors and the business strategies. Chapters
such as The Feral Systems and Other Factors Influencing the Success of Global ERP Implementations by
Don Kerr, and Challenges in Enterprise Information Systems Implementations by Ashim Singla discuss
the success of enterprise information systems implementation based on technology, people, and processes.
Additional selections, such as An Extended Model of Decision Making by David Sammon, Integrating
Enterprise Systems by Mark Hwang, and Preparedness of Small and Medium-Sized Enterprises to Use
Information and Communication Technology as a Strategic Tool by Klara Antlova address critical suc-
cess factors in the deployment of enterprise information systems.
Section 8, Emerging Trends, highlights areas for future research within the field of enterprise
information systems, while exploring new avenues for the advancement of the discipline. Beginning
this section is ERP Trends, Opportunities, and Challenges by Maha Shakir. This selection provides a
richer understanding of key ERP issues through by discussing emerging industry trends. The evolution
of enterprise resource planning is discussed in The Future of ERP and Enterprise Resource Manage-
ment Systems by Carlos Ferran and Ricardo Salim, and Engineering the Coordination Requirements in
Cross-organizational ERP Projects by Maya Daneva explores the complexity of cross-organizational
enterprise resource planning implementation. These and several other emerging trends and suggestions
for future research can be found within the final section of this exhaustive multi-volume set.
Although the primary organization of the contents in this multi-volume work is based on its eight
sections, offering a progression of coverage of the important concepts, methodologies, technologies,
applications, social issues, and emerging trends, the reader can also identify specific contents by utilizing
the extensive indexing system listed at the end of each volume. Furthermore to ensure that the scholar,
researcher and educator have access to the entire contents of this multi volume set as well as additional
coverage that could not be included in the print version of this publication, the publisher will provide
unlimited multi-user electronic access to the online aggregated database of this collection for the life
of the edition, free of charge when a library purchases a print copy. This aggregated database provides
far more contents than what can be included in the print version in addition to continual updates. This
unlimited access, coupled with the continuous updates to the database ensures that the most current
research is accessible to knowledge seekers.
As a comprehensive collection of research on the latest findings related to using technology to
providing various services, Enterprise Information Systems: Concepts, Methodologies, Tools and Ap-
plications, provides researchers, administrators and all audiences with a complete understanding of the
development of applications and concepts in enterprise information systems. Given the vast number of
issues concerning usage, failure, success, policies, strategies, and applications of enterprise information
systems in organizations, Enterprise Information Systems: Concepts, Methodologies, Tools and Appli-
cations addresses the demand for a resource that encompasses the most pertinent research in enterprise
information systems development, deployment, and impact.
Section I
Fundamental Concepts
and Theories
This section serves as the foundation for this exhaustive reference tool by addressing crucial theories essential
to the understanding of enterprise information systems. Chapters found within these pages provide an excellent
framework in which to position enterprise information systems within the field of information science and technology.
Individual contributions provide overviews of the history of enterprise information systems, the impact of information
systems on organizations, and overviews on various enterprise information system processes such as enterprise
resource planning and decision support systems. Within this introductory section, the reader can learn and choose
from a compendium of expert research on the elemental theories underscoring enterprise information systems.
1
Chapter 1.1
Principles and Experiences:
Designing and Building Enterprise
Information Systems
Mehmet S. Aktas
TUBITAK (Turkish National Science Foundation), Turkey
AbstrAct the local information services and assembles their

metadata instances under one hybrid architecture.
The data requirements of e-business applications It integrates different information services using
have been increased over the years. These applica- unification and federation concepts. In this chap-
tions present an environment for acquiring, process- ter, we summarize the principles and experiences
ing, and sharing data among interested parties. To gained in designing and building the semantics,
manage information in such data-intensive appli- architecture, and implementation for the hybrid
cation domain, independent enterprise e-business information service.
applications have developed their own solutions
to information services. However, these solutions
are not interoperable with each other, target vastly IntroductIon
different systems, and address diverse sets of re-
quirements. They require greater interoperability to The data requirements of e-business applications
enable communication between different systems, so have been increased over the years. These ap-
that they can share and utilize each other’s resources. plications present an environment for acquiring,
To address these challenges, we discuss principles processing and sharing data among interested par-
and experiences for designing and building of a ties. In order to manage data in such data-intensive
novel enterprise information system. We introduce a enterprise business application domain, Service
novel architecture for a hybrid information service, Oriented Architecture (SOA) principles have gained
which provides unification, federation, and interop- great importance. A Service Oriented Architecture is
erability of major Web-based information services. simply a collection of services that are put together
The hybrid information service is designed as an to achieve a common goal and that communicate
add-on information system, which interacts with with each other for either data passing or coordinat-
ing some activity. There is an emerging need for
DOI: 10.4018/978-1-60566-723-2.ch004 Web-based Enterprise Information Systems (EIS)
Copyright © 2011, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Principles and Experiences
that manage all the information that may be as- study are two-fold. First is to describe a generic
sociated with wide-scale SOA-based e-business Information Service architecture, which supports
applications. one-to-many information service implementations
Over the years, independent enterprise e- as local data sources and integrates different kinds
business applications have developed their own of Web Service metadata at a higher conceptual
customized implementations of Information level, while ignoring the implementation details
Service Specifications. These EIS solutions are of the local data-systems. Second is to describe
not interoperable with each other, target vastly the customized implementations two widely-used
different systems and address diverse sets of Web Service Specifications: the WS-I compatible
requirements (Zanikolas & Sakellariou, 2005). Web Service Context (WS-Context) (Bunting, et
They require greater interoperability to enable al., 2003) and Universal Description, Discovery
communication between different systems, so that and Integration (UDDI) (Bellwood, Clement, &
they can share and utilize each other’s resources. von Riegen, 2003) Specifications.
Furthermore, they do not provide uniform inter- The organization of this chapter is as follows.
faces for publishing and discovery of information. Section 2 reviews the relevant work. Section 3
In turn, this creates a limitation on the client-end gives an overview of the system. Section 4 presents
(e.g. fat client-end applications), as the users have the semantics of the Hybrid Service. Sections 5-6
to interact with more than one EIS implementation. present the architectural design details and the
For example, large-scale e-business applica- prototype implementation of the system. Finally,
tions require management of large amounts of Section 7 contains a summary of the chapter.
relatively slowly varying metadata. Another ex-
ample, dynamic Web service collections, gathered
together at any one time to perform an e-business relevAnt Work
operation, require greater support for dynamic
metadata. Previous solutions do not address Unifying heterogeneous data sources under a
management requirements of both large-scale, single architecture has been target of many in-
static and small-scale, highly-dynamic metadata vestigations (Ziegler & Dittrich, 2004). Informa-
associated to Web Services (Zanikolas & Sakel- tion integration is mainly studied by distributed
lariou, 2005). None of the existing solution enables database systems research and investigates how
communication between different e-business to share data at a higher conceptual level, while
applications, so that they can share, utilize each ignoring the implementation details of the local
other’s resources, have unified access interface data systems (Ozsu, 1999; Valduriez & Pacitti,
and address diverse sets of application require- 2004). Previous work on merger between the
ments (OGF GIN-CG). We therefore see this as heterogeneous information systems may broadly
an important area of investigation especially for be categorized as global-as-view and local-as-view
enterprise e-business applications domain. integration (Florescu, Levy, & Mendelzon, 1998).
This chapter introduces a Hybrid Service as a In former category, data from several sources are
EIS that addresses metadata management require- transformed into a global schema and may be
ments of both large-scale, static and small-scale, queried with a uniform query interface. Much
highly-dynamic metadata domains. The main nov- work has been done on automating information
elty of this chapter is to describe the semantics, ar- federation process using global schema approach.
chitecture, and implementation of a EIS integrating In the latter category, queries are transformed into
different Information Services by using unification specialized queries over the local databases and
and federation concepts. The implications of this integration is carried out by transforming queries.
2
Although the global schema approach captures ShaikhAli, Rana, Al-Ali, & Walker, 2003; Verma,
expressiveness capabilities of customized local Sivashanmugam, Sheth, Patil, Oundhakar, &
schemas, it does not scale up to high number Miller; GRIMOIRES). UDDI-M (Dialani, 2002)
of data sources. In the local-as-view approach, and UDDIe (ShaikhAli, Rana, Al-Ali, & Walker,
each local-system’s schema is need to be mapped 2003) projects introduced the idea of associating
against each other to transform the queries. In metadata and lifetime with UDDI Registry service
turn, this leads to large number of mappings that descriptions, where retrieval relies on the matches
need to be created and managed. of attribute name-value pairs between service
The proposed system differs from local-as- description and service requests. METEOR-S
view approaches, as its query transformation (Verma, Sivashanmugam, Sheth, Patil, Ound-
happens between a unified schema and local hakar, & Miller) leveraged UDDI Specification
schemas. It utilizes and leverages previous work by utilizing semantic web languages to describe
on global-as-view approach for integrating het- service entries. Grimories (GRIMOIRES) extends
erogeneous local data services. The previous work the functionalities of UDDI to provide a semantic
mainly focuses on solutions that automates the enabled registry designed and developed for the
information federation process at semantics level. MyGrid project (MyGrid). It supports third-party
Different from the previous work, the proposed attachment of metadata about services and repre-
approach presents a system architecture that en- sents all published metadata in the form of RDF
ables information integration at application level. triples either in a database, or in a file, or in a
To our best knowledge, the proposed system is memory. Although out-of-box UDDI Specification
a pioneer work, as it describes an Information is a widely used standard, it is limited to key-
Service architecture that enables unification and word-based query capabilities. Neither it allows
federation of information coming from different metadata-oriented queries, nor it takes into account
metadata systems. One limitation is that it does the volatile behavior of services. The previous
not scale to high number of local data-systems due work on UDDI-Extensions have centralized and
to low-level manual semantic schema integration. database-based solutions. Thus, they present low
To facilitate testing of our system, we did the low- fault-tolerance and low performance as opposed to
level information federation manually through a decentralized and in-memory based data-systems.
delicate analysis of the structure and semantic of The UDDI Specification is not designed to coor-
each target schema. Since the main focus of our dinate activities of Web Services participating in
research is to explore the information integration a work-flow style applications. Thus, it does not
at application level, we leave out the investigation support data-sharing and metadata management
of a low-level automated information federation requirements of rich-interacting systems.
capability for a future study. The proposed system addresses the limita-
Locating services of interest in Web service tions of previous work by introducing an add-on
intensive environments has recently become architecture, which runs one layer above the
important, since the service oriented architecture implementations of UDDI and its extensions. It
based systems increased in numbers and gain leverages previous work on UDDI and improves
popularity in recent years. The UDDI Specifica- the quality of UDDI-based metadata-systems in
tion is a widely used standard that enables services terms of fault-tolerance and high-performance.
advertise themselves and discover other services. To our best knowledge, this approach is unique,
A number of studies extends and improves the since it improves the qualities of existing imple-
out-of-box UDDI Specification (Open_GIS_Con- mentations of Information Services without
sortium_Inc.; Sycline; Galdos; Dialani, 2002; changing their code. Different from the previous
3
UDDI-work, the proposed system supports data- Context has some limitations as described below.
sharing and manages stateful interactions of Web Firstly, the context service has limited functional-
Services. The proposed research also introduces ities such as the two primary operations: GetCon-
semantics, communication protocols, and imple- text and SetContext. Secondly, the WS-Context
mentation of an extended UDDI version, which Specification is only focused on defining stateful
addresses metadata management requirements of interactions of Web Services. It does not define a
aforementioned application use domains. searchable repository for interaction-independent
Managing stateful interactions of Web Services information associated to the services involved in
is an important problem in rich-interacting sys- an activity. Thirdly, the WS-Context Specification
tems. There are varying specifications focusing does not define a data model to manage stateful
on point-to-point service communication, such Web Service information.
as Web Service Resource Framework (WSRF), The proposed system differs from previous
WS-Metadata Exchange (WS-ME). The WSRF work focusing on point-to-point service com-
specification, proposed by Globus alliance, IBM munication, since it adopts a centralized metadata
and HP, defines conventions for managing state, management strategy to regulate the interactions.
so that collaborating applications can discover, It adopts the WS-Context Specification and
inspect, and interact with stateful resources in a presents an extended version of WS-Context
standard way. The WS-ME provides a mechanism Specification and its implementation. The pro-
a) to share information about the capabilities of totype implementation manages dynamically
participating Web Services and b) to allow query- generated session-related metadata. To our best
ing a WS Endpoint to retrieve metadata about what knowledge, the proposed approach is unique,
to know to interact with them. Communication since none of the previous work on Information
among services is also achieved with a centralized Services implemented the WS-Context Specifica-
metadata management strategy, the Web Services tion to manage stateful interactions of services.
Context (WS-Context) Specification (Bunting, The proposed Hybrid Service leverages the ex-
et al., 2003). The W-Context defines a simple tended WS-Context Service implementation and
mechanism to share and keep track of common improves its quality in terms of fault-tolerance
information shared between multiple participants and high-performance.
in Web service interactions. It is a lightweight
storage mechanism, which allows the participant’s
of an activity to propagate and share context in- HybrId servIce
formation. Although point-to-point methodologies
successfully manage the stateful information, We designed and built a novel Hybrid Information
they provide service conversation with metadata Service Architecture called Hybrid Service, which
coming from the two services that exchange in- provides unification, federation and interoper-
formation. This is a limitation, since they become ability of major Information Services. The Hybrid
inefficient, when the number of communicating Service forms an add-on architecture that interacts
services increased. We find the WS-Context with the local information services and unifies
Specification as promising approach to tackle the them in a higher-level hybrid system. In other
problem of managing distributed session state, words, it provides a unifying architecture, where
since it models a session metadata repository as one can assemble metadata instances of different
an external entity, where more than two services information services. To facilitate testing of the
can easily access/store highly dynamic, shared system, we integrated the Hybrid Service with the
metadata. Even though it is promising, the WS- two local information service implementations:
4
WS-Context and Extended UDDI. We discuss b) extended business service entity (businessSer-
semantics of the Hybrid Service in the follow- vice). We describe the additional/modified data
ing section followed by a section discussing its model entities as follows.
architecture. Business service entity structure: The
UDDI’s business service entity structure contains
descriptive, yet limited information about Web
semAntIcs Services. A comprehensive description of the out-
of-box business service entity structure defined
The semantics of the system may be analyzed under by UDDI can be found in (Bellwood, Clement,
four categories: extended UDDI, WS-Context, & von Riegen, 2003). Here, we only discuss the
Unified Schema, and Hybrid Service Schema. additional XML structures introduced to expand
The extended UDDI Specification extends ex- on existing business service entity.
isting out-of-box UDDI Specification to address These additional XML elements are a) ser-
its aforementioned limitations. The WS-Context vice attribute and b) lease. The service attribute
Specification improves the existing out-of-box XML element corresponds to a static metadata
Web-Service Context Specification to meet the (e.g. WSDL of a given service). A lease structure
aforementioned application requirements. The describes a period of time during which a service
Unified Schema Specification integrates these two can be discoverable.
information service specifications. The Hybrid Service attribute entity structure: A service
Service Schema consists of two small schemas: attribute (serviceAttribute) data structure describes
Hybrid Schema and SpecMetadata Schema. information associated with service entities.
These define the necessary abstract data models Each service attribute corresponds to a piece of
to achieve a generic architecture for unification metadata, and it is simply expressed with (name,
and federation of different information service value) pairs. Apart from similar approaches (Di-
implementations. alani, 2002; ShaikhAli, Rana, Al-Ali, & Walker,
2003), in the proposed system, a service attribute
the extended uddI specification includes a) a list of abstractAtttributeData, b) a
categoryBag and c) a boundingBox XML struc-
We have designed extensions to the out-of-box tures: An abstractAttributeData element is used
UDDI Data Structure (Bellwood, Clement, & to represent metadata that is directly related with
von Riegen, 2003) to be able to associate both functionality of the service and store/maintain
prescriptive and descriptive metadata with service these domain specific auxiliary files as-is. This
entries. This way the proposed system addresses allows us to add third-party data models such as
limitations of UDDI (explained in Section 2) and “capabilities.xml” metadata file describing the
interoperates with existing UDDI clients without data coverage of domain-specific services such
requiring an excessive change in the implemen- as the geospatial services. An abstractAttribute-
tations. We name our version of UDDI as the Data can be in any representation format such
extended UDDI. as XML or RDF. This data structure allows us
The Extended UDDI Schema: The schema to pose domain-specific queries on the metadata
addresses the metadata requirements of Geo- catalog. Say, an abstractAttributeData of a geo-
graphical Information System/Sensor applications spatial service entry contains “capabilities.xml”
by extending the out-of-box UDDI data model. It metadata file. As it is in XML format, a client
includes following additional/modified entities: may conduct a find_service operation with an
a) service attribute entity (serviceAttribute) and XPATH query statement to be carried out on the
5
abstractAttributeData, i.e. “capabilities.xml”. In serviceAttribute elements. The find serviceAt-

this case, the results will be the list of geospatial tribute API call returns a list of serviceAttribute
service entries that satisfy the domain-specific structure matching the conditions specified in
XPATH query. A categoryBag element is used the arguments. get serviceAttributeDetail: Used
to provide a custom classification scheme to to retrieve semi-static metadata associated with a
categorize serviceAttribute elements. A simple unique identifier. The get serviceAttributeDetail
classification could be whether the service attribute API call returns the serviceAttribute structure
is prescriptive or descriptive. A boundingBox ele- corresponding to each of the attributeKey values
ment is used to describe both temporal and spatial specified in the arguments. get serviceDetail: Used
attributes of a given geographic feature. This way to retrieve service entity structure associated with
the system enables spatial query capabilities on a unique identifier.
the metadata catalog. Using Extended UDDI Schema XML API:
Extended UDDI Schema XMLAPI: We pres- Given the capabilities of the Extended-UDDI
ent extensions/modifications to existing UDDI Service, one can simply populate metadata in-
XML API set to standardize the additional capa- stances using the Extended-UDDI XML API as
bilities of our implementation. These additional in the following scenario. Say, a user publishes a
capabilities can be grouped under two XML API new metadata to be attached to an already exist-
categories: Publish and Inquiry. ing service in the system. In this case, the user
The Publish XML API is used to publish meta- constructs a serviceAttribute element. Based on
data instances belonging to different entities of aforementioned extended UDDI data model, each
the extended UDDI Schema. It extends existing service entry is associated with one or more ser-
UDDI Publish XML API Set. It consists of the viceAttribute XML elements. A serviceAttribute
following functions: save service: Used to extend corresponds to a piece of interaction-independent
the out-of-box UDDI save service functionality. metadata and it is simply expressed with (name,
The save service API call adds/updates one or value) pair. We can illustrate a serviceAttribute as
more Web Services into the service. Each service in the following example: ((throughput, 0.9)). A
entity may contain one-to-many serviceAttribute serviceAttribute can be associated with a lifetime
and may have a lifetime (lease). save serviceAt- and categorized based on custom classification
tribute: Used to register or update one or more schemes. A simple classification could be whether
semi-static metadata associated with a Web Ser- the serviceAttribute is prescriptive or descriptive.
vice. delete service: Used to delete one or more In the aforementioned example, the throughput
service entity structures. delete serviceAttribute: service attribute can be classified as descriptive.
Used to delete existing serviceAttribute elements In some cases, a serviceAttribute may correspond
from the service. The Inquiry XML API is used to to a domain-specific metadata where service meta-
pose inquiries and to retrieve metadata from the data could be directly related with functionality
Extended UDDI Information Service. It extends of the service. For instance, Open Geographical
existing UDDI Inquiry XML API set. It consists Concorcium compatible Geographical Informa-
of the following functions: find service: Used to tion System services provide a “capabilities.
extend the out-of-box UDDI find service func- xml” metadata file describing the data coverage
tionality. The find service API call locates specific of geospatial services. We use an abstractAttrib-
services within the service. It takes additional uteData element to represent such metadata and
input parameters such as serviceAttributeBag and store/maintain these domain specific auxiliary
Lease to facilitate the additional capabilities. find files as-is. As the serviceAttribute is constructed,
serviceAttribute: Used to find the aforementioned it can then be published to the Hybrid Service
6
by using “save_serviceAttribute” operation of structure contains dynamic metadata associated

the extended UDDI XML API. On receiving a to a Web Service or a session instance or both.
metadata publish request, the system extracts the Each sessionEntity is associated with its partici-
instance of the serviceAttribute entity from the pant sessionServices. The sessionService entity
incoming requests, assigns a unique identifier structure is used as an information container for
to it and stores in in-memory storage. Once the holding limited metadata about a Web Service par-
publish operation is completed, a response is sent ticipating to a session. A lease structure describes
to the publishing client. a period of time during which a sessionEntity or
serviceService or a context entity instances can
the Ws-context specification be discoverable.
Session service entity structure: The ses-
We have designed extensions and a data model sionService entity contains descriptive, yet limited
for the WS-Context Specifications to tackle the information about Web Services participating to a
problem of managing distributed session state. session. A service key identifies a sessionService
Unlike the point-to-point approaches, WS-Context entity. A sessionService may participate one or
models a third-party metadata repository as an more sessions. There is no limit on the number
external entity where more than two services of sessions in which a service can participate.
can easily access/store highly dynamic, shared These sessions are identified by session keys.
metadata. Each sessionService has a name and description
The extended WS-Context Schema: The associated with it. This entity has an endpoint ad-
schema is comprised of following entities: ses- dress field, which describes the endpoint address
sionEntity, sessionService and context. of the sessionService. Each sessionService may
Session entity structure: A sessionEntity have one or more context entities associated to it.
describes a period of time devoted to a specific The lease structure identifies the lifetime of the
activity, associated contexts, and serviceService sessionService under consideration.
involved in the activity. A sessionEntity can Context entity structure: A context entity
be considered as an information holder for the describes dynamically generated metadata. An
dynamically generated information. An instance instance of a context entity is uniquely identified
of a sessionEntity is uniquely identified with a with a context key, which is generated by the sys-
session key. A session key is generated by the tem when an instance of the entity is published.
system when an instance of the entity is published. If the context key is specified in a publication
If the session key is specified in a publication operation, the system updates the corresponding
operation, the system updates the corresponding entry with the new information. When retrieving
entry with the new information. When retrieving an instance of a context, a context key must be
an instance of a session, a session key must be presented.
presented. A sessionEntity may have name and A context is associated with a sessionEntity.
description associated with it. A name is a user- The session key element uniquely identifies the
defined identifier and its uniqueness is up to the sessionEntity that is an information container
session publisher. for the context under consideration. A context
A user-defined identifier is useful for the in- has also a service key, since it may also be as-
formation providers to manage their own data. A sociated with a sessionService participating a
description is optional textual information about session. A context has a name associated with
a session. Each sessionEntity contains one-to- it. A name is a user-defined identifier and its
many context entity structures. The context entity uniqueness is up to the context publisher. The
7
information providers manage their own data in find sessionService API call returns a service list
the interaction-dependent context space by using matching the criteria specified in the arguments.
this user-defined identifier. The context value can get sessionDetail: Used to retrieve sessionEntity
be in any representation format such as binary, data structure corresponding to each of the session
XML or RDF. Each context has a lifetime. Thus, key values specified in the arguments. get con-
each context entity contains the aforementioned textDetail: Used to retrieve the context structure
lease structure describing the period of time dur- corresponding to the context key values specified.
ing which it can be discoverable. get sessionServiceDetail: Used to retrieve ses-
WS-Context Schema XML API: We present sionService entity data structure corresponding
an XML API for the WS-Context Service. The to each of the sessionService key values specified
XML API sets of the WS-Context XML Meta- in the arguments. The Proprietary XML API is
data Service can be grouped as Publish, Inquiry, implemented to provide find/add/modify/delete
Proprietary, and Security. operations on the publisher list, i.e., authorized
The Publish XML API: The API is used to users of the system. We adapt semantics for the
publish metadata instances belonging to different proprietary XML API from existing UDDI Speci-
entities of the WS-Context Schema. It extends the fications. This XML API is as in the following:
WS-Context Specification Publication XML API find publisher: Used to find publishers registered
set. It consists of the following functions: save with the system matching the conditions speci-
session: Used to add/update one or more session fied in the arguments. get publisherDetail: Used
entities into the hybrid service. Each session may to retrieve detailed information regarding one or
contain one-to-many context entity, have a lifetime more publishers with given publisherID(s). save
(lease), and be associated with service entries. save publisher: Used to add or update information
context: Used to add/update one or more context about a publisher. delete_publisher: Used to delete
(dynamic metadata) entities into the service. save information about a publisher with a given pub-
sessionService: Used to add/update one or more lisherID from the metadata service. The Security
session service entities into the hybrid service. XML API is used to enable authenticated access to
Each session service may contain one-to-many the service. We adopt the semantics from existing
context entity and have a lifetime (lease). delete UDDI Specifications. The Security API includes
session: Used to delete one or more sessionEntity the following function calls. get_authToken: Used
structures. delete context: Used to delete one or to request an authentication token as an ‘authInfo’
more contextEntity structures. delete sessionSer- (authentication information) element from the
vice: Used to delete one or more session service service. The authInfo element allows the system
structures. The Inquiry XML API is used to pose implement access control. To this end, both the
inquiries and to retrieve metadata from service. publication and inquiry API set include authentica-
It extends the existing WS-Context XML API. tion information in their input arguments. discard_
The extensions to the WS-Context Inquiry API authToken: Used to inform the hybrid service that
set are outlined as follows: find session: Used to an authentication token is no longer required and
find sessionEntity elements. The find session API should be considered invalid.
call returns a session list matching the conditions Using WS-Context Schema XMLAPI: Given
specified in the arguments. find context: Used to the capabilities of the WS-Context Service, one can
find contextEntity elements. The find context API simply populate metadata instances using the WS-
call returns a context list matching the criteria Context XML API as in the following scenario.
specified in the arguments. find sessionService: Say, a user publishes a metadata under an already
Used to find session service entity elements. The created session. In this case, the user first constructs
8
a context entity element. Here, a context entity is businessEntity is used to aggregate one-to-many
used to represent interaction-dependent, dynamic services and sites managed by the same people. The
metadata associated with a session or a service or sessionEntity is used to aggregate session services
both. Each context entity has both system-defined participating to a session. Therefore, businessEn-
and user-defined identifiers. The uniqueness of the tity (from ExtendedUDDI) can be considered as
system-defined identifier is ensured by the system matching concepts with the sessionEntity (from
itself, whereas, the user-defined identifier is sim- WS-Context schema) as their intentional domains
ply used to enable users to manage their memory are similar. The cardinality between these entities
space in the context service. As an example, we differs, as the businessEntity may contain one to
can illustrate a context as in ((system-defined-uuid, may sessionEntities. The second mapping is be-
user-defined-uuid, “Job completed”)). A context tween: ExtendedUDDI.service and WS-Context.
entity can be also associated with service entity sessionService: These entities are equivalent as
and it has a lifetime. Contexts may be arranged the intentional domains that they represent are
in parent-child relationships. One can create a the same. The cardinality between these entities
hierarchical session tree where each branch can is also the same. In the integrated schema, we
be used as an information holder for contexts with unify these entities as service entity. The third
similar characteristics. This enables the system mapping is between ExtendedUDDI.metadata and
to be queried for contexts associated to a session WS-Context.context: These entities are equivalent
under consideration. This enables the system as the intentional domains that they represent are
to track the associations between sessions. As the same. The cardinality between these entities is
the context elements are constructed, they can also the same. We continue unification by merg-
be published with save_context function of the ing the two schemas based on the mappings that
WS-Context XML API. On receiving publishing we identified and create a unified schema. The
metadata request, the system processes the request, Unified Schema unifies matching and disjoint
extracts the context entity instance, assigns a entities of the two schemas.
unique identifier, stores in the in-memory storage The Unified Schema is comprised of the fol-
and returns a respond back to the client. lowing entities: businessEntity, sessionEntity,
service, bindingTemplate, metadata, tModel,
the unified schema specification publisherAssertions. A businessEntity describes
a party who publishes information about a ses-
This research investigates a system architecture sion (in other words service activity), site or
that would support information federation and service. The publisherAssertions entity defines
unification at application-level. To facilitate test- the relationship between the two businessEntities.
ing of such system architecture, a unified schema The sessionEntity describes information about a
is needed. We achieved semantic-level unification service activity that takes place. A sessionEntity
manually through a delicate analysis of the struc- may contain one-to-many service and metadata
ture and semantics of the two schemas: extended entities. The service entity provides descriptive
UDDI and WS-Context. We introduced an abstract information about a Web Service family. It may
data model and query/publish XML API and contain one-to-many bindingTemplate entities that
named it as the Unified Schema Specification. define the technical information about a service
We begin unification by finding the mappings end-point. A bindingTemplate entity contains
between the similar entities of the two schemas. references to tModel that defines descriptions of
First mapping is between ExtendedUDDI.busi- specifications for service end-points. The service
nessEntity and WS-Context.sessionEntity: The entity may also have one-to-many metadata at-
9
tached to it. A metadata contains information get businessDetail: Used to retrieve businessEntity
about both interaction-dependent, interaction- data structure of the Unified Schema corresponding
independent metadata and service data associated to each of the business key values specified in the
to Web Services. A metadata entity describes the arguments. get sessionDetail: Used to retrieve ses-
information pieces associated to services or sites sionEntity data structure corresponding to each of
or sessions as (name, value) pairs. the session key values specified in the arguments.
The Unified Schema XML API: To facilitate get serviceDetail: Used to retrieve service entity
testing of the federation capability, we introduce a data structure corresponding to each of the service
limited Query/Publish XML API that can be car- key values specified in the arguments. get meta-
ried out on the instances of the parts of the Unified dataDetail: Used to retrieve the metadata structure
Schema. We can group the Unified Schema XML corresponding to the metadata key values specified.
API under two categories: Publish and Inquiry. Using the Unified Schema XML API: Given
The Publish XML API: This API is used to these capabilities, one can simply populate the
publish metadata instances belonging to different Hybrid Service with Unified Schema metadata
entities of the Unified Schema. It consists of the instances using its XML API as in the following
following functions: save business: Used to add/ scenario. Say, a user wants to publish both session-
update one or more business entities into the hybrid related and interaction-independent metadata as-
service. save session: Used to add/update one or sociated to an existing service. In this case, the user
more session entities into the hybrid service. Each constructs metadata entity instance. Each metadata
session may contain one-to-many metadata, one- entity has both system-defined and user-defined
to-many service entities and have a lifetime (lease). identifiers. The uniqueness of the system-defined
save service: Used to add/update one or more ser- identifier is ensured by the system itself, whereas,
vice entries into the hybrid service. Each service the user-defined identifier is simply used to enable
entity may contain one-to-many metadata element users to manage their memory space in the context
and may have a lifetime (lease). save metadata: service. As an example, we can illustrate a context
Used to register or update one or more metadata as in the following examples: a) ((throughput,
associated with a service. delete business: Used to 0.9)) and b) ((system-defined-uuid, user-defined-
delete one or more business entity structures. delete uuid, “Job completed”)). A metadata entity can
session: Used to delete one or more sessionEntity be also associated with site, or sessionEntity of
structures. delete service: Used to delete one or the Unified Schema and it has a lifetime. As the
more service entity structures. delete metadata: metadata entity instances are constructed, they
Used to delete existing metadata elements from can be published with “save_metadata” function
the hybrid service. The Inquiry XML API is used of the Unified Schema XML API. On receiving
to pose inquiries and to retrieve metadata from publishing metadata request, the system processes
service. It consists of following functions: find the request, extracts the metadata entity instance,
business: This API call locates specific businesses assigns a unique identifier, stores in the in-memory
within the hybrid services. find session: Used to storage and returns a respond back to the client.
find sessionEntity elements. The find session API
call returns a session list matching the conditions the Hybrid service semantics
specified in the arguments. find service: Used to
locate specific services within the hybrid service. The Hybrid Service introduces an abstraction layer
find metadata: Used to find service entity elements. for uniform access interface to be able to support
The find service API call returns a service list one-to-many information service specification
matching the criteria specified in the arguments. (such as WS-Context, Extended UDDI, or Unified
10
Schema). To achieve the uniform access capability, mapping file of the schema under consideration.
the system presents two XML Schemas: a) Hybrid Then, the system stores the newly-inserted schema
Schema and b) Specification Metadata (Spec- entity instances into the in-memory storage. de-
Metadata) Schema. The Hybrid Schema defines lete_schemaEntity: The delete_schemaEntity is
the generic access interface to the Hybrid Service. used to delete an instance of any schema entities
The SpecMetadata Schema defines the necessary of a given specification. The delete_schemaEntity
information required by the Hybrid Service to be API call deletes existing service entities associ-
able to process instances of supported information ated with the specified key(s) from the system.
service schemas. We discuss the semantics of the On receiving a schema entity deletion request
Hybrid Schema and the SpecMetadata Schema in message, the system processes the incoming mes-
the following sections. sage based on information given in the mapping
file of the schema under consideration. Then the
The Hybrid Schema system, deletes the correct entity associated with
the key. find_schemaEntity: This API call locates
The Hybrid Schema is designed to achieve a schemaEntities whose entity types are identified
unifying access interface to the Hybrid Service. It in the arguments. This function allows the user to
is independent from any of the local information locate a schema entity among the heterogeneous
service schemas supported by the Hybrid Service. metadata space. On receiving a find_schemaEntity
It defines a set of XML API to enable clients/pro- request message, the system processes the incom-
viders to send specification-based publish/query ing message based on information given in the
requests (such as WS-Context’s “save_context” schema mapping file of the schema under con-
request) in a generic way to the system. sideration. Then the system, locates the correct
The Hybrid Service XML API allows the entities matching the query under consideration.
system support one-to-many information service get_schemaEntity: The get_schemaEntityDetail is
communication protocols. It consists of the follow- used to retrieve an instance of any schema enti-
ing functions: hybrid_service: This XML API call ties of a given specification. It returns the entity
is used to pose inquiry/publish requests based on structure corresponding to key(s) specified in the
any specification. With this function, the user can query. On receiving a get_schemaEntityDetail
specify the type of the schema and the function. retrieval request message, the system processes
This function allows users to access an information the incoming message based on information given
service back-end directly. The user also specifies in the mapping file of the schema under consider-
the specification-based publish/query request in ation. Then the system retrieves the correct entity
XML format based on the specification under associated with the key. Finally, the system sends
consideration. On receiving the hybrid_function the result to the user.
request call, the system handles the request based Given these capabilities, one can simply
on the schema and function specified in the query. populate the Hybrid Service using the “save_sche-
save_schemaEntity: This API call is used to save maEntity” element and publish metadata instances
an instance of any schema entities of a given of the customized implementations of information
specification. The save_schemaEntity API call service specifications. The “save_schemaEntity”
is used to update/add one or more schema entity element includes an “authInfo” element, which
elements into the Hybrid Information Service. describes the authentication information; “lease”
On receiving a save_schemaEntity publication element, which is used to identify the lifetime of
request message, the system processes the incom- the metadata instance; “schemaName” element,
ing message based on information given in the which is used to identify a specification schema
11
(such as Extended UDDI Schema); “schemaFunc- information is needed to process the incoming
tionName”, which is used to identify the function request and execute the requested operation on
of the schema (such as “save_ serviceAttribute”); the appropriate information service backend. This
“schema_SAVERequestXML”, which is an ab- information defines the function name, its argu-
stract element used for passing the actual XML ments, return values and the class, which needs to
document of the specific publish function of a be executed in the information service back-end.
given specification. The Hybrid Service requires The MappingRules XML element describes
a specification metadata document that describes all required information regarding the mapping
all necessary information to be able to process rules that provide mapping between the Unified
XML API of the schema under consideration. Schema and the local information service schemas
We discuss the specification metadata semantics such as extended UDDI and WS-Context. The
in the following section. MappingRules element consists of one-to-many
MappingRule sub-elements. Each MappingRule
The SpecMetadata Schema describes information about how to map a unified
schema XML API to a local information service
The SpecMetadata XML Schema is used to de- schema XML API. The MappingRule element
fine all necessary information required for the contains the necessary information to identify
Hybrid Service to support an implementation of functions that will be mapped to each other.
information service specification. The Hybrid Given these capabilities, one can simply popu-
System requires an XML metadata document, late the Hybrid Service as in the following scenario.
which is generated based on the SpecMetadata Say, a user wants to publish a metadata into the
Schema, for each information service specifica- Hybrid Service using WS-Context’s “save_con-
tion supported by the system. The SpecMetadata text” operation through the generic access inter-
XML file helps the Hybrid System to know how face. In this case, firstly, the user constructs an
to process instances of a given specification XML instance of the “save_context” XML document
API. The SpecMetadata includes Specname, (based on the WS-Context Specification) as if s/
Description, and Version XML elements. These he wants to publish a metadata instance into the
elements define descriptive information to help WS-Context Service. Once the specification-
the Hybrid Service to identify the local informa- based publish function is constructed, it can be
tion service schema under consideration. The published into the Hybrid Service by utilizing
FunctionProperties XML element describes the “save_schemaEntity” operation of the Hybrid
all required information regarding the functions Service Access API.
that will be supported by the Hybrid Service. As for the arguments of the “save_sche-
The FunctionProperties element consists of one- maEntity” function, the user needs to pass the
to-many FunctionProperty sub-elements. The following arguments: a) authentication informa-
FunctionProperty element consists of function tion, b) lifetime information, c) schemaName
name, memory-mapping and information-service- as “WS-Context”, d) schemaFunctionName as
backend mapping information. Here the memory- “save_context” and e) the actual save_context
mapping information element defines all necessary document which was constructed based on the
information to process an incoming request for WS-Context Specification. Recall that, for each
in-memory storage access. The memory-mapping specification, the Hybrid Service requires a
information element defines the name, user- SpecMetadata XML document (an instance of
defined identifier and system-defined identifier the Specification Metadata Schema). On receipt
of an entity. The information-service-backend of the “save_schemaEntity” publish operation, the
12
Hybrid Service obtains the name of the schema control capability is responsible for enforcing
(such as WS-Context) and the name of the publish controlled access to the Hybrid Information Ser-
operation (such as save_context) from the pass- vice. The investigation and implementation of
ing arguments. In this case, the Hybrid Service access control mechanism for the decentralized
consults with the WS-Context SpecMetadata information service is left out for future study.
document and obtains necessary information about (3) TupleSpaces Access API allows access to
how to process incoming “save_context” opera- in-memory storage. This API supports all query/
tion. Based on the memory mapping information publish operations that can take place on the Tuple
obtained from user-provided SpecMetadata file, Pool. (4) The Tuple Pool implements a lightweight
the system processes the request, extracts the implementation of JavaSpaces Specification
context metadata entity instance, assigns a unique (Sun_Microsystems, 1999) and is a generalized in-
identifier, stores in the in-memory storage and memory storage mechanism. It enables mutually
returns a response back to the client. exclusive access and associative lookup to shared
data. (5) The Tuple Processor layer is designed to
process metadata stored in the Tuple Pool. Once
ArcHItecture the metadata instances are stored in the Tuple Pool
as tuple objects, the system starts processing the
We designed and built a Hybrid Information tuples and provides the following capabilities. The
Service (Hybrid Service) to support handling first capability is the LifeTime Management. Each
and discovery of metadata associated with Web metadata instance may have a lifetime defined
Services. The Hybrid Service is an add-on archi- by the user. If the metadata lifetime is exceeded,
tecture that interacts with the local information then it is evicted from the TupleSpace. The second
systems and unifies them in a higher-level hybrid capability is the Persistency Management. The
system. It provides a unifying architecture where system checks with the tuple space every so often
one can assemble metadata instances of different for newly added /updated tuples and stores them
information services. into the database for persistency of information.
The third capability is the Fault Tolerance Man-
Abstraction layers agement. The system checks with the tuple space
every so often for newly-added/updated tuples and
Figure 1 illustrates the detailed architectural replicates them in other Hybrid Service instances
design and abstraction layers of the system. (1) using the publish-subscribe messaging system.
The Uniform Access layer imports the XML API This capability also provides consistency among
of the supported Information Services. It is de- the replicated datasets. The fourth capability is
signed as generic as possible, so that it can support the Dynamic Caching Management. With this
one-to-many XML API, as the new information capability, the system keeps track of the requests
services are integrated with the system. (2) The coming from the pub-sub system and replicates/
Request-processing layer is responsible for ex- migrates tuples to other information services where
tracting incoming requests and process operations the high demand is originated. (6) The Filtering
on the Hybrid Service. It is designed to support layer supports the federation capability. This layer
two capabilities: notification and access control. provides filtering between instances of the Unified
The notification capability enables the interested Schema and local information service schemas
clients to be notified of the state changes happen- such as WS-Context Schema based on the user
ing in a metadata. It is implemented by utilizing defined mapping rules to provide transformations.
publish-subscribe based paradigm. The access (7) The Information Resource Manager layer is
13
Figure 1.
Client
Client
Extended WS-Context
ClientHybrid
….
UDDIExtended
API API
WS-Context API Hybrid Mapping
UDDI WS API API
…. Files
WS API
Request
UNIFORMprocessor
ACCESS INTERFACE (XML)
Request processor
Access Control
Access Control
Request processor
TUPLE SPACE ACCESS API Mapping
TUPLE SPACE API
Access Control
Rule
Files (XSLT)
TUPLE SPACE
POOL API SERVICE
A HYBRIGTUPLE POOL
TUPLE
GRID INFORMATION
MEMORY-IN STORAGE
Persistency Dynamic
Persistency Dynamic Caching Fault
Caching Fault Tolerance
Tolerance
Management
Management TUPLE
Management
Management POOL ( JAVA SPACES)
Management
Management Management
Management
TUPLE
TUPLE processor
processor
Filter
Filter PUB PUB-SUB Network
– SUB Network Manager
Manager
ManagerManager PublisherMANAGER
INFORMATION RESOURCE Subscriber
Resource Resource ……
Handler Handler
Extended UDDI WS-Context HYBRID ….
GIS NETWORK
Service - I Service - II CONNECTED….WITH PUB-
DB1 DB2 …
SUB SYSTEM 10 of 34
Extended UDDI WS-Context …..
responsible for managing low-level information

service implementations. It provides decoupling execution logic Flow
between the Hybrid Service and sub-systems. (8)
The Pub-Sub Network layer is responsible for The execution logic for the Hybrid Service hap-
communication between Hybrid Service instances. pens as follows. Firstly, on receiving the client
request, the request processor extracts the in-
distribution coming request. The request processor processes
the incoming request by checking it with the
Figure 2 illustrates the distribution in Hybrid specification-mapping metadata (SpecMetadata)
Service and shows N-node decentralized services files. For each supported schema, there is a Spec-
from the perspective of a single service interact- Metadata file, which defines all the functions that
ing with two clients. To achieve communication can be executed on the instances of the schema
among the network nodes, the Hybrid Service under consideration. Each function defines the
utilizes a topic-based publish-subscribe software required information related with the schema
multicasting mechanism. This is a multi-publisher, entities to be represented in the Tuple Pool. (For
multicast communication mechanism, which example; entity name, entity identifier key, etc…).
provides message-based communication. In the Based on this information, the request processor
prototype implementation, we use an open source extracts the inquiry/publish request from the in-
implementation of publish-subscribe paradigm coming message and executes these requests on
(NaradaBrokering (Pallickara & Fox, 2003)) for the Tuple Pool. We apply the following strategy
message exchanges between peers. to process the incoming requests. First off all,
14
Figure 2. Distributed hybrid services

client client
Wsdl Wsdl
HTTP(S)
Wsdl Wsdl
HybrId service
Subscriber
ext uddI Ws-context
...
Publisher
database database
topic based Publish-subscribe
Replica Server-1
messaging system
Wsdl Wsdl
HybrId service HybrId service
Ext UDDI WS-Context Ext UDDI WS-Context

... ...
Database Database Database Database
Replica Server-2 Replica Server-N
the system keeps all locally available metadata second capability is the notification management.
keys in a table in the memory. On receipt of a Here, the system informs the interested parties of
request, the system first checks if the metadata the state changes happening in the metadata. This
is available in the memory by checking with the way the requested entities can keep track of infor-
metadata-key table. If the requested metadata is mation regarding a particular metadata instance.
not available in the local system, the request is Thirdly, if the request is to be handled in the
forwarded to the Pub-Sub Manager layer to probe memory, the Tuple Space Access API is used to
other Hybrid Services for the requested metadata. enable the access to the in-memory storage. This
If the metadata is in the in-memory storage, then API allows us to perform operations on the Tuple
the request processor utilizes the Tuple Space Pool. The Tuple Pool is an in-memory storage. The
Access API and executes the query in the Tuple Tuple Pool provides a storage capability where
Pool. In some cases, requests may require to be the metadata instances of different information
executed in the local information service back- service schemas can be represented.
end. For an example, if the client’s query requires Fourthly, once the metadata instances are
SQL query capabilities, it will be forwarded to stored in the Tuple Pool as tuple objects, the tuple
the Information Resource Manager, which is processor layer is being used to process tuples
responsible of managing the local information and provide a variety of capabilities. The first
service implementations. capability is the LifeTime Management. Each
Secondly, once the request is extracted and metadata instance may have a lifetime defined
processed, the system presents abstraction layers by the user. If the metadata lifetime is exceeded,
for some capabilities such as access control and then it is evicted from the Tuple Pool. The second
notification. First capability is the access control capability is the Persistency Management. The
management. This capability layer is intended to system checks with the tuple space every so often
provide access controlling for metadata accesses. for newly-added / updated tuples and stores them
As the focus of our investigation is distributed into the local information service back-end. The
metadata management aspects of information third capability is the Dynamic Caching Man-
services, we leave out the research and imple- agement. The system keeps track of the requests
mentation of this capability as future study. The coming from the other Hybrid Service instances
15
and replicates/migrates metadata to where the implementation has a Resource Handler that en-
high demand is originated. The fourth capability ables interaction with the Hybrid Service.
is the Fault Tolerance Management. The system Sixthly, if the metadata is to be replicated/
again checks with the tuple space every so often stored into other Hybrid Service instances, the
for newly-added / updated tuples and replicates Pub-Sub Management Layer is used for manag-
them in other information services using the ing interactions with the Pub-Sub network. On
pub-sub system. This service is also responsible receiving the requests from the Tuple Processor,
for providing consistency among the replicated the Pub-Sub Manager publishes the request to the
datasets. As the main focus of this paper is to corresponding topics. The Pub-Sub Manager may
discuss information federation in Information also receive key-based access/storage requests
Services, the detailed discussion on replication, from the pub-sub network. In this case, these
distribution, consistency enforcement aspects of requests will be carried out on the Tuple Pool by
the system is left out as the focus of another paper utilizing TupleSpace Access API. The Pub-Sub
(Aktas, Fox, & Pierce, 2008). Manager utilizes publisher and subscriber sub-
The Hybrid Service supports a federation components in order to provide communication
capability to address the problem of providing among the instances of the Hybrid Services.
integrated access to heterogeneous metadata. To
facilitate the testing of this capability, a Unified
Schema is introduced by integrating different PrototyPe ImPlementAtIon
information service schemas. If the metadata is
an instance of the Unified Schema, such metadata The Hybrid Information Service prototype
needs to be mapped into the appropriate local implementation consists of various modules
information service back-end. To achieve this, such as Query and Publishing, Expeditor, Filter
the Hybrid Service utilizes the filtering layer. and Resource Manager, Sequencer, Access and
This layer does filtering based on the user-defined Storage. This software is open source project and
mapping rules to provide transformations between available at (Aktas). The Query and Publishing
the Unified Schema instances and local schema module is responsible for processing the incom-
instances. If the metadata is an instance of a lo- ing requests issued by end-users. The Expeditor
cal schema, then the system does not apply any module forms a generalized in-memory storage
filtering, and backs-up this metadata to the cor- mechanism and provides a number of capabilities
responding local information service back-end. such as persistency of information. The Filter and
Fifthly, if the metadata is to be stored to the Resource Manager modules provide decoupling
information service backend (for persistency of between the Hybrid Information Service and the
information), the Information Resource Manage- sub-systems. The Sequencer module is respon-
ment layer is used to provide connection with the sible for labeling each incoming context with
back-end resource. The Information Resource a synchronized timestamp. Finally, the Access
Manager handles with the management of local and Storage modules are responsible for actual
information service implementations. It provides communication between the distributed Hybrid
decoupling between the Hybrid Service and sub- Service nodes to support the functionalities of a
systems. With the implementation of Information replica hosting system.
Resource Manager, we have provided a uniform, The Query and Publishing module is respon-
single interface to sub-information systems. The sible for implementing a uniform access interface
Resource Handler implements the sub-information for the Hybrid Information Service. This module
system functionalities. Each information service implements the Request Processing abstraction
16
layer with access control and notification capa- an in-memory storage based on the TupleSpaces
bilities. On completing the request processing paradigm (Carriero & Gelernter, 1989). The
task, the Query and Publishing module utilizes Tuple-processing layer introduces a number of
the Tuple Space API to execute the request on capabilities: LifeTime Management, Persistency
the Tuple Pool. On completion of operation, the Management, Dynamic Caching Management and
Query and Publication module sends the result to Fault Tolerance Management. Here, the LifeTime
the client. As discussed earlier, context informa- Manager is responsible for evicting those tuples
tion may not be open to anyone, so there is a need with expired leases. The Persistency Manager is
for an information security mechanism. We leave responsible for backing-up newly-stored / updated
out the investigation and implementation of this metadata into the information service back-ends.
mechanism as a future study. We must note that to The Fault Tolerance Manager is responsible for
facilitate testing of the centralized Hybrid Service creating replicas of the newly added metadata.
in various application use domains, we imple- The Dynamic Caching Manager is responsible
mented a simple information security mechanism. for replicating/migrating metadata under high
Based on this implementation, the centralized demand onto replica servers where the demand
Hybrid Service requires an authentication token originated.
to restrict who can perform inquiry/publish opera- The Filtering module implements the filtering
tion. The authorization token is obtained from the layer, which provides a mapping capability based
Hybrid Service at the beginning of client-server on the user defined mapping rules. The Filtering
interaction. In this scenario, a client can only ac- module obtains the mapping rule information from
cess the system if he/she is an authorized user by the user-provided mapping rule files. As the map-
the system and his/her credentials match. If the ping rule file, we use the XSL (stylesheet language
client is authorized, he/she is granted with an au- for XML) Transformation (XSLT) file. The XSLT
thentication token which needs to be passed in the provides a general purpose XML transformation
argument lists of publish/inquiry operations. The based on pre-defined mapping rules. Here, the
Query and Publishing module also implements a mapping happens between the XML APIs of the
notification scheme. This is achieved by utilizing Unified Schema and the local information service
a publish-subscribe based messaging scheme. This schemas (such as WS-Context or extended UDDI
enables users of Hybrid Service to utilize a push- schemas).
based information retrieval capability where the The Information Resource Manager module
interested parties are notified of the state changes. handles with management of local information
This push-based approach reduces the server load service implementations such as the extended
caused by continuous information polling. We UDDI. The Resource Manager module separates
use the NaradaBrokering software (Pallickara & the Hybrid System from the sub-system classes. It
Fox, 2003) as the messaging infrastructure and its knows which sub-system classes are responsible
libraries to implement subscriber and publisher for a request and what method needs to be executed
components. by processing the specification-mapping metadata
The Expeditor module implements the file that belongs the local information service
Tuple Spaces Access API, Tuple Pool and Tuple- under consideration. On receipt of a request, the
processing layer. The Tuple Spaces Access API Information Resource Manager checks with the
provides an access interface on the Tuple Pool. corresponding mapping file and obtain informa-
The Tuple Pool is a generalized in-memory stor- tion about the specification-implementation. Such
age mechanism. Here, to meet the performance information could be about a class (which needs
requirement of the proposed architecture, we built to be executed), it’s function (which needs to be
17
invoked), and function’s input and output types, In order to impose an order on updates, each
so that the Information Resource Manager can context has to be time-stamped before it is stored
delegate the handling of incoming request to ap- or updated in the system. The responsibility of the
propriate sub-system. By using this approach, the Sequencer module is to assign a timestamp to each
Hybrid Service can support one-to-many infor- metadata, which will be stored into the Hybrid
mation services as long as the sub-system imple- Service. To do this, the Sequencer module inter-
mentation classes and the specification-mapping acts with Network Time Protocol (NTP)-based
metadata (SpecMetadata) files are provided. The time service (Bulut, Pallickara, & Fox, 2004)
Resource Handler is an external component to implemented by NaradaBrokering (Pallickara
the Hybrid Service. It is used to interact with & Fox, 2003) software. This service achieves
sub-information systems. Each specification has synchronized timestamps by synchronizing the
a Resource Handler, which allows interaction machine clocks with atomic timeservers available
with the database. The Hybrid System classes across the globe.
communicate with the sub-information systems
by sending requests to the Information Resource
Manager, which forwards the requests to the ap- conclusIon
propriate sub-system implementation. Although
the sub-system object (from the corresponding This chapter introduced the principles and expe-
Resource Handler) performs the actual work, the riences of designing and building a web-based
Information Resource Manager seems as if it is Enterprise Information Service. Within this em-
doing the work from the perspective of the Hybrid phasis, it also introduced a novel architecture for
Service inner-classes. This approach separates the an Enterprise Information Service, called Hybrid
Hybrid Service implementation from the local Service, supporting handling and discovery of
schema-specific implementations. not only quasi-static, stateless metadata, but also
The Resource Manager module is also used session related metadata.
for recovery purposes. We have provided a recov- The Hybrid Service is an add-on architecture
ery process to support persistent in-memory stor- that runs one layer above existing information
age capability. This type of failure may occur if the service implementations. Although, it mainly man-
physical memory is wiped out when power fails or ages metadata that maybe associated to Web Ser-
machine crashes. This recovery process converts vices, it can also be used to manage any metadata
the database data to in-memory storage data (from about Web resources on the Internet. It provides
the last backup). It runs at the bootstrap of the Hy- unification, federation and interoperability of En-
brid Service. This process utilizes user-provided terprise Information Services. To achieve unifica-
“find_schemaEntity” XML documents to retrieve tion, the Hybrid Service is designed as a generic
instances of schema entities from the information system with front and back-end abstraction layers
service backend. Each “find_schemaEntity” XML supporting one-to-many local information systems
document is a wrapper for schema specific “find” and their communication protocols. To achieve
operations. At the bootstrap of the system, firstly, federation, the Hybrid Service is designed to sup-
the recovery process applies the schema-specific port information integration technique in which
find functions on the information service backend metadata from several heterogeneous sources
and retrieves metadata instances of schema enti- are transferred into a global schema and queried
ties. Secondly, the recovery process stores these with a uniform query interface. To manage both
metadata instances into the in-memory storage to quasi-static and dynamic metadata and provide
achive persistent in-memory storage. interoperability with wide-range of Web Service
18
applications, the Hybrid Service is integrated Dialani, V. (2002). UDDI-M version 1.0 API
with two local information services: WS-Context specification. Southampton, UK: University of
XML Metadata Service and Extended UDDI XML Southampton.
Metadata Service. The WS-Context Service is
Florescu, D., Levy, A., & Mendelzon, A. (1998).
implemented based on WS-Context Specification
Database techniques for the World Wide Web:
to manage dynamic, session related metadata. The
A survey. SIGMOD Record, 27(3), 59–74.
Extended UDDI Service is implemented based on
doi:10.1145/290593.290605
an extended version of the UDDI Specification to
manage semi-static, stateless metadata. Galdos. Galdos Inc. Retrieved from http://www.
galdosinc.com
GRIMOIRES. (n.d.).UDDI compliant Web ser-
reFerences
vice registry with metadata annotation extension.
Aktas, M. S. (n.d.). Fault tolerant high performance Retrieved from http://sourceforge.net/projects/
information service-FTHPIS-hybrid WS-context grimoires
service Web site. Retrieved from http://www. MyGrid. (n.d.). UK e-science project. Retrieved
opengrids.org/wscontext from http://www.mygrid.org.uk
Aktas, M. S., Fox, G. C., Pierce, M. E. (2008). OGF Grid Interoperation Now Community Group
Distributed high performance grid information (GIN-CG). (n.d.). Retrieved from https://forge.
service. Submitted to Journal of Systems and gridforum.org/projects/gin
Software.
Open_GIS_Consortium_Inc. (2003). OWS1.2
Bellwood, T., Clement, L., & von Riegen, C. UDDI experiment. OpenGIS interoperability
(2003). UDDI version 3.0.1: UDDI spec technical program report OGC 03-028. Retrieved from
committee specification. Retrieved from http:// http://www.opengeospatial.org/docs/03-028.pdf
uddi.org/pubs/uddi-v3.0.1-20031014.htm
Ozsu, T. P. V. (1999). Principles of distributed
Bulut, H., Pallickara, S., & Fox, G. (2004, June database systems, 2nd edition. Prentice Hall.
16-18). Implementing a NTP-based time service
within a distributed brokering system. In ACM Pallickara, S., & Fox, G. (2003). NaradaBrokering:
International Conference on the Principles and A distributed middleware framework and archi-
Practice of Programming in Java, Las Vegas, tecture for enabling durable peer-to-peer grids. In
NV. Proceedings of ACM/IFIP/USENIX International
Middleware Conference Middleware-2003, Rio
Bunting, B., Chapman, M., Hurley, O., Little, M., Janeiro, Brazil.
Mischinkinky, J., Newcomer, E., et al. (2003).
Web services context (WS-context) version 1.0. Pallickara, S., & Fox, G. (2003). NaradaBroker-
Retrieved from http://www.arjuna.com/library/ ing: A middleware framework and architecture
specs/ws_caf_1-0/WS-CTX.pdf for enabling durable peer-to-peer grids. (LNCS).
Springer-Verlag.
Carriero, N., & Gelernter, D. (1989). Linda in
context. Communications of the ACM, 32(4),
444–458. doi:10.1145/63334.63337
19
ShaikhAli. A., Rana, O., Al-Ali, R., & Walker, Verma, K., Sivashanmugam, K., Sheth, A., Patil,
D. (2003). UDDIe: An extended registry for Web A., Oundhakar, S., & Miller, J. (n.d.). METEOR–S
services. In Proceedings of the Service Oriented WSDI: A scalable P2P infrastructure of registries
Computing: Models, Architectures, and Applica- for semantic publication and discovery of Web
tions, Orlando, FL. SAINT-2003 IEEE Computer services. Journal of Information Technology and
Society Press. Management.
Sun_Microsystems. (1999). JavaSpaces specifica- Zanikolas, S., & Sakellariou, R. (2005). A
tion revision 1.0. Retrieved from http://www.sun. taxonomy of grid monitoring systems. Future
com/jini/specs/js.ps Generation Computer Systems, 21(1), 163–188.
doi:10.1016/j.future.2004.07.002
Sycline. Sycline Inc. Retrieved from http://www.
synclineinc.com Ziegler, P., & Dittrich, K. (2004). Three decades
of data integration-all problems solved? In WCC,
Valduriez, P., & Pacitti, E. (2004). Data manage-
3-12.
ment in large-scale P2P systems. Int. Conf. on
High Performance Computing for Computational
Science (VecPar2004). ( [). Springer.]. LNCS,
3402, 109–122.
This work was previously published in Always-On Enterprise Information Systems for Business Continuance: Technologies
for Reliable and Scalable Operations, edited by Nijaz Bajgoric , pp. 58-77 , copyright 2010 by Information Science Reference
(an imprint of IGI Global).
20
21
Chapter 1.2
Evolution of Enterprise
Resource Planning
Ronald E. McGaughey
University of Central Arkansas, USA
Angappa Gunasekaran
University of Massachusetts—Dartmouth, USA
AbstrAct systems to meet evolving business needs, then their

research should focus at least in part on the chang-
Business needs have driven the design, developing business environment, its impact on business
ment, and use of the enterprise-wide information needs, and the requirements for enterprise systems
systems we call Enterprise Resource Planning that meet those needs.
(ERP) systems. Intra enterprise integration was a
driving force in the design, development, and use of
early ERP systems. Changing business needs have IntroductIon
brought about the current business environment,
wherein supply chain integration is desirable, if not Twenty years ago supplier relationship manage-
essential, thus current and evolving ERP systems ment was unique to the Japanese (those firms who
demonstrate an expanded scope of integration that embraced the JIT philosophy), China was still a
encompasses limited inter-enterprise integration. slumbering economic giant, the Internet was largely
This chapter explores the evolution, the current for academics and scientists and certainly not a
status, and future of ERP, with the objective of consideration in business strategy, the very idea of a
promoting relevant future research in this important network of businesses working together as a virtual
area. If researchers hope to play a significant role in enterprise was almost like science fiction, and hardly
the design, development, and use of suitable ERP anyone had a cell phone. The world has changed.
The cold war is over and economic war is on. We
have moved rapidly toward an intensely competi-
DOI: 10.4018/978-1-60566-146-9.ch002
Evolution of Enterprise Resource Planning
tive, global economic environment. Countries like efficiency and effectiveness through integrat-
China and India are fast positioning themselves as ing, streamlining and improving fundamental
key players and threatening the economic order back-office business processes. Traditional ERP
that has existed for decades. Information Technol- systems were called back-office systems because
ogy (IT) is more sophisticated than ever, yet we they involved activities and processes in which
still struggle with how to best use it in business, the customer and general public were not typically
and on a personal level as well. E-commerce (B2B, involved, at least not directly. Functions supported
B2C, C2C, G2C, B2G) has become commonplace by ERP typically included accounting, manufac-
and M-commerce is not far behind, especially in turing, human resource management, purchasing,
Europe and Japan. In 2007, for the first time, there inventory management, inbound and outbound
are more cell phones than tethered phones in the logistics, marketing, finance and to some extent
US, and increasingly sophisticated cell phones engineering. The objective of traditional ERP
have capabilities that exceed the capabilities of systems in general was greater efficiency, and to
older PCs. This is the backdrop against which we a lesser extent effectiveness. Contemporary ERP
will discuss the evolving enterprise information systems have been designed to streamline and in-
system. At this point we will call it ERP, but is tegrate operation processes and information flows
should become evident in the course of reading within a company to promote synergy (Nikolo-
this manuscript that ERP is a label that may no poulos, Metaxiotis, Lekatis and Assimakopoulos,
longer be appropriate for evolving enterprise and 2003) and greater organizational effectiveness.
inter-enterprise systems. These newer ERP systems have moved beyond
In this chapter we define ERP and discuss the the back-office to support front-office processes
evolution of ERP, the current state of ERP and and activities like those fundamental to customer
the future of ERP. We will emphasize how the relationship management. The goal of most firms
evolution of ERP has been influenced by changing implementing ERP is to replace diverse functional
business needs and by evolving technology. We systems with a single integrated system that does
present a simple framework to explain that evolu- it all faster, better, and cheaper. Unfortunately, the
tion. Some general directions for future research “business and technology integration technology
are indicated by our look at the past, present and in a box” has not entirely met expectations (Koch,
particularly the future of ERP. 2005). While there are some success stories, many
companies devote significant resources to their
ERP effort only to find the payoff disappointing
erP deFIned (Dalal, Kamath, Kolarik and Sivaraman, 2003;
Koch, 2005). Let us examine how we have come
The ERP system is an information system that to this point in the ERP lifecycle.
integrates business processes with the aim of
creating value and reducing costs by making the
right information available to the right people at tHe evolutIon oF erP
the right time to help them make good decisions in
managing resources productively and proactively. The origin of ERP can be traced back to Materi-
An ERP is comprised of multi-module applica- als Requirement Planning (MRP). While the
tion software packages that serve and support concept of MRP was understood conceptually
multiple business functions (Sane, 2005). These and discussed in the 1960s, it was not practical
large automated cross functional systems were for commercial use. It was the availability of
designed to bring about improved operational computing power (processing capability and stor-
22
age capacity) that made commercial use of MRP production schedule. Manpower planning and
possible and practical. While many early MRP support for human resources management were
systems were built in-house, often at great expense, incorporated into MRP. Distribution management
MRP became one of the first popular off-the-shelf capabilities were added. The enhanced MRP and
business applications (Orlicky, 1975). In essence, its many modules provided data useful in the fi-
MRP involves taking inventory records, the master nancial planning of manufacturing operations, thus
production schedule, and bills of materials and financial planning capabilities were added. Busi-
calculating time phased material, component ness needs, primarily for operational efficiency,
and sub-assembly requirements (gross and net), and to a lesser extent for greater effectiveness, and
planned orders, delivery dates, and more. Note the advancements in computer processing and storage
term calculating was used rather than forecasting. technology brought about MRP and influenced its
With a realistic MPS, accurate inventory records, evolution. What started as an efficiency oriented
lead times that are known and predictable, and tool for production and inventory management was
current and correct bills of materials, it is possible becoming increasingly a cross functional system.
to calculate material, component, and assembly A very important capability to evolve in MRP
requirements rather than forecast them. The shear systems was the ability to close the loop (control
volume of calculations necessary for MRP with loop). This was largely because of the develop-
multiple orders for even a few items made the use ment of real time (closed loop) MRP systems to
of computers essential. Initially, batch processing replace regenerative MRP systems in response to
systems were used and regenerative MRP systems changing business needs and improved computer
were the norm, where the plan would be updated technology—time-sharing was replacing batch
periodically, often weekly. MRP employed a type processing as the dominant computer processing
of backward scheduling wherein lead times were mode. With time-sharing mainframe systems
used to work backwards from a due date to an the MRP system could run 24/7 and update
order release date. While the primary objective continuously. Use of the corporate mainframe
of MRP was to compute material requirements, that performed other important computing task
the MRP system proved also to be a useful sched- for the organization was not practical for some
uling tool. Order placement and order delivery companies, because MRP consumed too many
were planned by the MRP system. Not only were system resources; subsequently, some companies
orders for materials and components generated opted to use mainframes (now growing smaller
by an MRP system, but also production orders and cheaper, but increasing in processing speed
for manufacturing operations that used those and storage capability) or mini-computers (could
materials and components to make higher level do more, faster than old mainframes) that could
items like sub assemblies and finished products. be dedicated to MRP. MRP could now respond
As MRP systems became popular and more and (update relevant records) to timely data fed into
more companies started using them, practitioners, the system and produced by the system. This
vendors and researchers started to realize that the closed the control loop with timely feedback
data and information produced by the MRP system for decision making by incorporating current
in the course of material requirements planning and data from the factory floor, warehouse, vendors,
production scheduling could be augmented with transportation companies, and other internal and
additional data and used for other purposes. One external sources, thus giving the MRP system the
of the earliest add-ons was the Capacity Require- capability to provide current (almost real-time)
ments Planning module which could be used in information for better planning and control. These
developing capacity plans to produce the master closed loop systems better reflected the realities
23
of the production floor, logistics, inventory, and sophisticated back-office systems were more ap-
more. It was this transformation of MRP into a propriately labeled Enterprise Resource Planning
planning and control tool for manufacturing by systems (Nikolopoulos, Metaxiotis, Lekatis, and
closing the loop, along with all the additional Assimakopoulos, 2003).
modules that did more than plan materials—they MRP II was mostly for automating the busi-
planned and controlled various manufacturer re- ness processes within an organization, but ERP,
sources—that led to MRPII. Here too, improved while primarily for support of internal processes,
computer technology and the evolving business started to support processes that spanned enterprise
needs for more accurate and timely information boundaries (the extended enterprise). While ERP
to support decision making and greater organiza- systems originated to serve the information needs
tional effectiveness contributed to the evolution of manufacturing companies, their domain was not
from MRP to MRPII. just manufacturing anymore. Early ERP systems
The MRP in MRPII stands for Manufacturing typically ran on mainframes like their predeces-
Resource Planning rather than materials require- sors, MRP and MRPII, but many migrated to cli-
ments planning. The MRP system had evolved ent/server systems where networks were central
from a material requirements planning system and distributed databases more common. The
into a planning and control system for resources growth of ERP and the migration to client/server
in manufacturing operations—an enterprise systems really got a boost from the Y2K scare.
information system for manufacturing. As time Many companies were convinced of the need to
passed, MRPII systems became more widespread, replace older main-frame based systems, some
and more sophisticated, particularly when used in ERP and some not, with the newer client/server
manufacturing to support and complement com- architecture. After all, since they were going to
puter integrated manufacturing (CIM). Databases have to make so many changes in the old systems
started replacing traditional file systems allowing to make them Y2K compliant and avoid serious
for better systems integration and greater query problems—this was what vendors and consultants
capabilities to support decision makers, and the often told them to create FUD (fear, uncertainty
telecommunications network became an integral and doubt)--they might as well bite the bullet and
part of these systems in order to support commu- upgrade. Vendors and consultants benefited from
nications between and coordination among system the Y2K boost to ERP sales, as did some of their
components that were sometimes geographically customers. Since Y2K, ERP systems have evolved
distributed, but still within the company. In that rapidly, bringing us to the ERP systems of today.
context the label CIM II was used for a short time Present day ERP systems offer more and more
to describe early systems with capabilities now capabilities and are becoming more affordable,
associated with ERP (Lope, 1992). The need for even for SMEs (Dahlen and Elfsson, 1999).
greater efficiency and effectiveness in back-office
operations was not unique to manufacturing, but
was also common to non-manufacturing opera- erP todAy
tions. Companies in non-manufacturing sectors
such as healthcare, financial services, air trans- As ERP systems continue to evolve, vendors like
portation, and the consumer goods sector (Chung PeopleSoft (Conway, 2001) and Oracle (Green,
and Snyder, 2000) started to use MRPII-like 2003) are moving to an Internet-based architec-
systems to manage critical resources, thus the ture, in large part because of the ever-increasing
M for manufacturing seemed not always to be importance of E-commerce and the globalization
appropriate. In the early 90s, these increasingly of business (Abboud and Vara, 2007). Beyond
24
Table 1. Enterprise system modules

that, perhaps the most salient trend in the continu-
ing evolution of ERP is the focus on front-office Modules
applications and inter-organizational business Enterprise Resource Planning (ERP)
processes, particularly in support of supply chain Customer Relationship Management (CRM)
management (Scheer and Habermann, 2000; Sales Management
Al-Mashari, 2002). ERP is creeping out of the Field Service Management
Retail Management
back-office into the front and beyond the enter-
Asset Management
prise to customers, suppliers and more, in order Financial Management
to meet changing business needs (Burns, 2007). Yield Management
Front-office applications involve interaction with Business Collaboration
external constituents like customers, suppliers, Supplier Relationship Management (SRM)

partners and more—hence the name front-office Inventory Management
Order Processing
because they are visible to “outsiders.” Key play-
Business Intelligence
ers like Baal, Oracle, PeopleSoft, and SAP have Data Warehouse
incorporated Advanced Planning and Scheduling Knowledge Management
Analytics and Reporting
(APS), Sales Force Automation (SFA), Customer
Online Business Services
Relationship Management (CRM), Supply Chain User Services
Management (SCM), Business Intelligence, and E-Commerce
E-commerce modules/capabilities into their sys- M-Commerce
tems, or repositioned their ERP systems as part Facilities Management

Maintenance Management
of broader Enterprise Suites incorporating these
Warehouse Management
and other modules/capabilities. ERP products Logistics Management
reflect the evolving business needs of clients and Distribution Management
the capabilities of IT, perhaps most notably those Project Management
related to the Web. Human Resource Management
While some companies are expanding their
ERP system capabilities (adding modules) and still
calling them ERP systems, others have started to state that today’s ERP systems, or ERP-like sys-
use catchy labels like enterprise suite, E-commerce tems, typically include modules/capabilities as-
suite, and enterprise solutions to describe their sociated with front-office processes and activities.
enterprise solution clusters that include ERP ERP capabilities are packaged with other modules
among other modules/capabilities. Table 1 lists that support front-office and back-office pro-
the various modules/capabilities taken from the cesses and activities, and nearly anything else that
product descriptions of vendors like PeopleSoft, goes on within organizations. ERP proper (the
Oracle, J.D. Edwards, and SAP, who are major back office system) has not become unimportant
players in the ERP/enterprise systems market. because back-office efficiency and effectiveness
Perhaps, most notable about ERP today is that was, is, and will always be important. Today’s
it is much more than manufacturing resource focus, however, seems more to be external as
planning. ERP and ERP-like systems have become organizations look for ways to support and improve
popular with non-manufacturing operations like relationships and interactions with customers,
universities, hospitals, airlines and more, where suppliers, partners and other stakeholders (Knap-
back-office efficiency is important and so too is en, 2007). While integration of internal functions
front-office efficiency and effectiveness (Chung is still important, and in many organizations still
and Snyder, 2000). In general, it is accurate to has not been achieved to a great extent, external
25
integration is now receiving much attention. services (also require global positioning systems)
Progressive companies desire to do things--all a reality, as well as a host of other capabilities
things--faster, better and cheaper (to be agile), we categorize as M-Business. One can expect to
and they want systems and tools that will improve see ERP geared more to the support of both E-
competitiveness, increase profits, and help them Commerce and M-Commerce. Internet, mobile,
not just to survive, but to prosper in a dynamic and wireless technology should figure prominently
global economy. Today, that means working with in new and improved system modules and capabili-
suppliers, customers and partners like never be- ties (Bhattacharjee, Greenbaum, Johnson, Martin,
fore. Vendors are using the latest technology to Reddy, Ryan, White and McKie, 2002; O’Brien,
respond to these evolving business needs, as 2002; Sane, 2005). Vendors and their customers
evidenced in the products and services they now will find it necessary to make fairly broad, sweep-
offer or plan to offer their customers. Will ERP ing infrastructure changes to meet the demands of
be the all-encompassing system comprised of the E-Commerce and M-Commerce (Bhattacharjee,
many modules and capabilities mentioned, or will Greenbaum, Johnson, Martin, Reddy, Ryan, White
it be relegated to the status of modules/function- and McKie, 2002; Higgins, 2005). Movement
ality in all-encompassing systems for the future? away from client-server systems to Internet based
architectures is likely. In fact, it has already started
(Scheer and Habermann, 2000; Conway, 2001;
erP And tHe Future Abboud and Vara, 2007). New systems will have
to incorporate existing and evolving standards and
New, multi-enterprise business models like Value older systems will have to be adapted to existing
Collaboration Networks, customer-centric net- and evolving standards, and that may make the
works that coordinate all players in the supply transition a little uncomfortable and expensive for
chain, are evolving as we enter the 21st century vendors and their customers. Perhaps the biggest
(Nattkemper, 2000). These new business models business challenge with E-Commerce, and even
reflect an increased business focus on external more so with M-Commerce, is understanding how
integration. While no one can really predict the to use these new and evolving capabilities to bet-
future of ERP very far into the future, current ter serve the customer, work with suppliers and
management concerns and emphasis, vendor other business partners, and to be generally more
plans, and the changing business and technological efficient and effective. Businesses are just begin-
environments, provide some clues about the future ning to understand E-Commerce and how it can
of ERP. We turn our attention now to evolving be used to meet changing business needs as well
business needs and technological changes that as how it changes them, and now M-Commerce
should shape the future of ERP. poses a whole new challenge. E-Commerce and
E-Commerce is arguably one of the most im- M-Commerce pose challenges for vendors and
portant developments in business in the last 50 for their clients. Back-office and front-office
years (Jack Welch reportedly called it the “Viagra processes and activities are being affected by E-
of business”), and M-Commerce is poised to takes Commerce/E-Business and will most certainly be
it place along side or within the rapidly growing affected by M-Commerce/M-Business.
area of E-Commerce. Internet technology has The current business emphasis on intra and
made E-Commerce in its many forms (B2B, B2C, inter organizational process integration and ex-
B2G, G2C, C2C, etc.) possible. Mobile and wire- ternal collaboration should remain a driving force
less technology are expected to make “always on” in the evolution of ERP in the foreseeable future.
Internet and “anytime/anywhere” location based Some businesses are attempting to transform
26
themselves from traditional, vertically integrated partners or a value added service provider that
organizations into multi-enterprise, “recombinant coordinates the entire supply chain, and powered
entities” reliant on core-competency-based strate- by a single system serving all participants, could
gies (Genovese, Bond, Zrimsek and Frey, 2001). be the model for the future.
Integrated SCM and business networks will re- Regardless of the means used to achieve greater
ceive great emphasis, reinforcing the importance external integration to complement internal inte-
of IT support for cross enterprise collaboration and gration, the new focus on supporting/facilitating
inter-enterprise processes (Bhattacharjee, Green- inter-organizational processes will be important
baum, Johnson, Martin, Reddy, Ryan, White and to the future of ERP like systems, whatever they
McKie, 2002; Al-Mashari, 2002). Collaborative are called! Solution providers and consultants will
commerce (C-commerce) has become not only a strive to enable companies to communicate and
popular buzz word, but also a capability businesses collaborate with other entities that comprise the
desire/need. C-Commerce is label used to describe extended enterprise (Bhattacharjee, Greenbaum,
Internet-based (at least at present) electronic col- Johnson, Martin, Reddy, Ryan, White and McKie,
laboration among businesses, typically supply 2002; Knapen, 2007). Internet based technologies
chain partners, in support of inter-organizational will unquestionably be key in supporting cross-
processes that involve not necessarily transactions, enterprise ERP in the foreseeable future, and the
but rather information sharing and coordination ASP (also called SAAS—software as a service)
(Sane, 2005). ERP systems, or their successors model may just be the future of ERP in a busi-
will have to support the required interactions and ness world focused on supply chain management
processes among and within business entities, (Burns, 2007). Note that ASP solutions are gener-
and work with other systems/modules that do the ally less costly, and consequently, they reduce one
same. The back-office processes and activities particular component of financial risk—the cost
of business network partners will not exist in a of product obsolescence. It is noteworthy also that
vacuum--many will overlap. There will be great ASP (SAAS) solutions are moving ERP within
need for business processes to span organizational reach of SMEs, as it costs much less to “rent” than
boundaries (some do at present), possibly requir- to “buy.” The lower cost of ASPs makes them
ing a single shared inter-enterprise system that more attractive to SMEs, increasingly targeted
will do it (we might call it a distributed ERP), by vendors seeking new customers (Dahlen and
or at least ERP systems that can communicate Elfsson, 1999; Muscatello, Small and Chen, 2003,
with and co-process (share/divide processing Burns, 2007).
tasks) with other ERP systems—probably the Some expect Web services to play a prominent
most practical solution, at least in the near fu- role in the future of ERP (O’Brien, 2002; ACW
ture. Middleware, ASPs, and enterprise portal Team, 2004; Abboud and Vara, 2007). Web Ser-
technologies may play an important role in the vices range from simple to complex, and they can
integration of such modules and systems (Bhat- incorporate other Web services. The capability of
tacharjee, Greenbaum, Johnson, Martin, Reddy, Web Services to allow businesses to share data,
Ryan, White and McKie, 2002). Widespread applications, and processes across the Internet
adoption of a single ASP solution among supply (O’Brien, 2002) may result in ERP systems of
chain partners may facilitate interoperability as the future relying heavily on the Service Oriented
all supply chain partners essentially use the same Architecture (SOA), within which Web Services
system. Could one ASP database serve all supply are created and stored, providing the building
chain partners? Alternatively, a supply chain portal blocks for programs and systems. Web Service
(vertical portal), jointly owned by supply chain technology could put the focus where it belongs:
27
on putting together the very best functional solu- So will it be single source, or best in breed?
tion to automate a business process (Bhattacharjee, The best in breed approach will be good if greater
Greenbaum, Johnson, Martin, Reddy, Ryan, White interoperability/integration among vendor prod-
and McKie, 2002). The use of “best in breed” Web ucts is achieved (Bhattacharjee, Greenbaum,
Service-based solutions might be more palatable to Johnson, Martin, Reddy, Ryan, White and McKie,
businesses, since it might be easier and less risky 2002). There is a need for greater “out of the box”
to plug-in a new Web Service based solution than interoperability, thus a need for standards. Ideally,
replace or add-on a new product module. A greater products will reach a level of standardization where
role for Web Services (the SOA) is expected, and software modules exhibit behavior similar to the
that too would heighten the importance of an plug and play hardware--you just plug in a new
Internet Based Architecture to the future of ERP module, the system recognizes it, configures itself
(Abboud and Vara, 2007; Burns, 2007). It should to accommodate the new module, and eureka, it
be noted that a Service Oriented Architecture is works! While this is much to hope for, increased
not incompatible with ASPs, and in fact it should standardization brought about by developments
make building, maintaining and providing ASP like the Service Oriented Architecture, and
products/services more efficient and effective. XML-based XBRL, might make interoperability
All from one, or best in breed? Reliance on a a reality, though probably not anytime soon. The
single vendor would seem best from a vendor’s fact that some are embracing standards for XML
perspective, but it may not be best from the client’s (Garbellotto, 2007) and more, does provide some
standpoint. While it may be advantageous to have reason for hope, but whether the future of ERP
only one proprietary product to install and operate, software trends toward the single source or best
and a single contact point for problems, there are in breed approach remains to be seen. Regardless
risks inherent in this approach. Switching cost of the direction, integration technologies will be
can be substantial, and if a single vendor does not important in the new breed of modular, but linked
offer a module/solution needed by the client, then enterprise applications.
the client must develop it internally, do without it, Data warehouses, data mining, and various
or purchase it from another vendor. The market analytic capabilities are needed in support of
is in fact changing in some ways because of this front office and back office processes and activi-
situation. The market was about what a product ties involved in CRM, SRM, SCM, Field Service
would do, but it is evolving such that now the Management, business collaboration and more.
focus is on what the product will do in the future Likewise they are important in business intelli-
(Maguire, 2006). Potential customers are “thinking gence (the process and BI systems) and strategic
ahead” and seem better to understand the future management. An important trend at present is
consequences of today’s system choice. Still, it is merging ERP and CRM with BI (Burns, 2007).
not uncommon for a client to be faced with trying Data warehouses will play an important role in
to get diverse products to work together, and the the future of ERP, because they will require data
problems of doing so are well documented. The from ERP and provide data to support decision
single source approach means an organization making in the context of ERP—ERP the process.
must place great faith in the vendor, and with the Ideally, the Data warehouse would be integrated
consolidation and changes taking place among with all front-office, back-office and strategic
enterprise solution providers, that can be risky. systems to the extent that it helps close loops by
Never the less, the “one source” alternative seems providing accurate and timely data to support deci-
most popular at present (Burns, 2007). sion making in any context, in the form of on-line
28
analytical processing. Knowledge management Internet (O’Brien, 2002). Back-office systems will
systems (KMS) endowed with neural networks not be unimportant, but they are more mature as
and expert system capabilities should play a key a consequence of past emphasis and many work
role in decision making as they will be capable quite well. Emphasis will be on more thorough
of capturing, modeling and automating decision integration of the modules that comprise back-
making processes. Data warehouses and KMS office systems, integration of back-office systems
should enable future ERP systems to support with front-office and strategic systems, and inte-
more automated business decision making (Bhat- gration of front-office, back-office and strategic
tacharjee, Greenbaum, Johnson, Martin, Reddy, systems with the systems of other organizations,
Ryan, White and McKie, 2002) and they should especially within the context of the supply chain
be helpful in the complex decision making needed (Knapen, 2007). At present greater organizational
in the context of fully integrated supply chain effectiveness in managing the entire supply chain
management. More automated decision making in all the way to the end customer is a priority in
both front-office and back-office systems, should business. The greater emphasis on front-office
eliminate/minimize human variability and error, functions and cross enterprise communications
greatly increase decision speed, and hopefully and collaboration via the Internet simply reflects
improve decision quality. Business Intelligence changing business needs and priorities. A 2004
(BI) tools, which are experiencing a significant ITtoolbox survey of ERP users in Europe, North
growth in popularity, take internal and external America, Asia, India and elsewhere showed
data and transform it into information used in great interest in improved functionality and ease
building knowledge that helps decision makers of integration and implementation (top motives
to make more “informed” decisions—no pun for adding new modules or purchasing new ERP
intended. Current Business Intelligence (BI) tools systems). Furthermore the same survey showed
are largely designed to support strategic planning greatest interest in modules for CRM, Data Ware-
and control but will likely trickle down to lower housing and SCM (top 3 on list). The demand for
level decision makers, where their capabilities will specific modules/capabilities in particular shows
be put to use in tactical and perhaps operational that businesses are looking beyond the enterprise.
decision contexts. BI tools use data, typically This external focus is encouraging vendors to
from a data warehouse, along with data mining, seize the moment by responding with the mod-
analytic, statistical, query, reporting, forecast- ules/systems that meet evolving business needs.
ing and decision support capabilities to support The need to focus not just on new front office
managerial planning and control. The combined tools but also on strategy will encourage greater
capabilities of the data warehouse, KMS and BI vendor emphasis on tools like Data Warehouses
should contribute to faster, better, and less costly and capabilities like Business Intelligence that
(in terms of time and effort involved) decisions at support strategy development, implementation
all organizational levels. They should be helpful and control. A CAMagazine enterprise software
in making decisions in the inter-organizational survey showed that still in 2007, vendors remain
context of supply chain management, where com- focused on providing support for BI and strategic
plexity is increased by the need make decisions planning and control (Burns, 2007).
involving multiple supply chain partners. The evolving environment of business suggests
At least in the near future, it appears that a direction for these comprehensive enterprise
greater emphasis will be placed on front-office systems that would seem to make ERP an inap-
systems, as opposed to back office systems, and propriate label. The Garner group coined the term
sharing data, applications and processes across the ERPII to describe their vision of the enterprise
29
system of the future with increased focus on the necessary for competitive reasons, systems have
front office, strategy, and the internet. ERPII was evolved to meet those needs. One can hardly ignore
described as a business strategy and a set of collab- the technological changes that have taken place,
orative operational and financial processes inter- because the current state of technology is a limiting
nally and beyond the enterprise (Zrimsek, 2002). factor in the design of systems to meet evolving
The Gartner Group projected that by 2005 ERPII business needs. From our examination of the evolu-
would replace ERP as the key enabler of internal tion of ERP we would conclude that the next stage
and inter-enterprise efficiency (Zrimsek, 2002). of the evolution will come about and be shaped by
While not there yet, the systems they described the same forces that have shaped each stage, that
are evolving, but the ERPII label was lost! The being evolving business needs and advances in
name that seems to have stuck, at least for now, is technology. We expect ERP (the traditional back
Enterprise Information Systems, or just Enterprise office system) to take its place along with MRP and
Systems. Wikipedia (2007) describes Enterprise MRPII. The functions of ERP will remain impor-
Information Systems as technology platforms that tant and necessary, as have the functions of MRP
“enable organizations to integrate and coordinate and MRPII, but ERP will likely be absorbed by
their business processes. They provide a single something bigger, and ERP itself will take its place
system that is central to the organization and as an integral part of the Internet based enterprise
ensure that information can be shared across all or inter-enterprise system of the future. Whether
functional levels and management hierarchies. that all encompassing system is called Enterprise
Enterprise systems are invaluable in eliminating resource planning, Enterprise suite, Enterprise
the problem of information fragmentation caused Information System, or by a label that currently
by multiple information systems in an organiza- resides in the back of some vendor employee or
tion, by creating a standard data structure.” researchers mind, remains to be seen. One thing
While the ERP label may linger for a while, seems certain, the next stage in the evolution will
ERP will likely be relegated to module(s)/capabil- hinge on the same forces shaping systems of the
ity status, as a name more fitting for the evolving past—business need and technological change.
integrated, inter-enterprise, front office- back of-
fice, strategic systems will replace ERP, in much
the same way that ERP replaced MRPII. conclusIon
ERP has evolved over a long period of time.

tHe erP evolutIon FrAmeWork MRP gave way to MRPII, then MRPII to ERP,
ERP to ERPII, and in rather short order ERPII to
This framework simply summarizes the evolution of Enterprise System/Enterprise Information Sys-
ERP relating the stages in its evolution to business tem. It seems that Enterprise system is holding
needs driving the evolution, as well as changes in steady for now, but it too may give way to a new
technology. Table 2 presents the framework. As label that reflects the “inter-enterprise flavor” of
MRP evolved into MRPII, then ERP, and finally future systems. MRP capabilities still exists as
to ERPII/Enterprise Information System (present will ERP capabilities, but future systems must
state of ERP), the scope of the system expanded as provide an increasingly broad set of capabilities
organizational needs changed, largely in response to and modules that support the back-office, front-
the changing dynamics of the competitive environ- office, strategic planning and control, as well
ment. As business has become increasingly global as integrating processes and activities across
in nature and cooperation among enterprises more diverse enterprises comprising supply chains and
30
Table 2. The evolution of ERP
System Primary Business Need(s) Scope Enabling Technology

MRP Efficiency Inventory Management and Mainframe computers, batch process-
Production planning and ing, traditional file systems.
control.
MRPII Efficiency, Effectiveness and inte- Extending to the entire manu- Mainframes and Mini computers,
gration of manufacturing systems facturing firm (becoming cross realtime (time sharing) process-
functional). ing, database management systems
(relational)
ERP Efficiency (primarily back office), Entire organization (increas- Mainframes, Mini and micro Com-
Effectiveness and integration of all ingly cross functional), both puters, Client server networks with
organizational systems. manufacturing and non-manu- distributed processing and distributed
facturing operations databases, Data warehousing, and
mining, knowledge management.
ERPII Efficiency, effectiveness and integra- Entire organization extending Mainframes, Client Server systems,
tion within and among enterprises. to other organizations (cross distributed computing, knowledge
functional and cross enterprise- management, internet technology
-partners, suppliers, customers, (includes intranets extranets, portals).
etc.)
Inter-Enterprise Efficiency, effectiveness, coordina- Entire organization and its con- Internet, Service Oriented Architec-
Resource Planning, tion, and Integration within and stituents (increasingly global ture, Application Service Providers,
Enterprise Systems, among all relevant supply chain and cross cultural) compris- wireless networking, mobile wireless,
Supply Chain Man- members as well as other partners, ing global supply chain from knowledge management, grid comput-
agement, or whatever or stakeholders on a global scale. beginning to end as well as ing, artificial intelligence.
label gains common other industry and government
acceptance constituents
business networks. Whatever the name, current to remain a priority. An environment for busi-
trends suggest certain characteristics which we ness applications much like the “plug and play”
can reasonably expect. This future system will environment for hardware would make it easier
have to support E-commerce and M-commerce, for organizations to integrate their own systems
thus wireless technology, including mobile but not and have their systems integrated with other or-
limited to mobile, and the internet will play a role ganizations’ systems. Such an environment awaits
in the evolving architecture. An Internet-based greater standardization. This ideal “plug and play”
architecture now seems most likely, at least in environment would make it easier for firms to opt
the near future, and it may be a Service Oriented for a “best in breed” strategy for application/mod-
Architecture wherein Web Services are key. The ule acquisition as opposed to reliance on a single
increased emphasis on front-office systems and vendor for a complete package of front-office,
strategic planning and control will likely influ- back-office and strategic systems. At present it
ence new capabilities introduced by vendors for seems that selecting a single vendor is preferred
the next few years. Increased automation of deci- (Burns, 2007). Future developments in computer
sion making is to be expected with contributions (hardware and software) and telecommunications
from Knowledge Management Systems, Data technology will move us closer to effective inter-
Warehouses/Data Marts, and Business Intelligence organizational system integration and make fully
systems fueled by advancements in the field of integrated supply chain management a reality.
artificial intelligence. Greater interoperability of Perhaps we might call the evolving system an
diverse systems and more thorough integration Interprise Resource Planning System, Interprise
within and between enterprise systems is likely Management System, or Interprise Information
31
System to emphasize the inter-enterprise nature • Solution vendors embrace standards like
of these systems. Whatever they are called, it XML, the Service Oriented Architecture,
seems that what will be required of them goes and evolving wireless standards with
far beyond what the enterprise resource planning due consideration to global business
(ERP) label would aptly describe, even with the requirements;
“II” (ERPII) added!
From the discussion of ERP’s future one can For researchers and practitioners the advice
extrapolate certain desired capabilities of the in- is simple. The two primary drivers in the evolu-
terprise/enterprise system of the future. Following tion from MRP, to MRPII, to ERP, to ERPII,
is a list of desired/required capabilities: to Enterprise Systems were business need and
technological change. Technological change
• Supports interaction of supply chain part- made possible the development of systems to
ners and inter-organizational processes; meet changing business needs. The needs may
• A single corporate database to facilitate exist for a while before the technology can help
true functional system/module integration; meet them, and the technology can exists for a
• At some point in time, possibly an inter- while before someone recognizes its usefulness
organizational database to integrate supply in meeting a current or evolving business need.
chain partners--maybe supplied by an ASP In either case, the focus should be on monitoring
or some other entity that supports the entire business needs and monitoring technological
supply chain; change. Research that does both and is geared
• Any necessary data transfer among/integra- towards bringing the two together could make
tion of modules is smooth and consistent; significant contributions to business. The ERP
• Possesses flexibility to continuously sup- system of the future, whatever it may be called,
port agile companies responding to dy- will be found at the convergence of business need
namic business environment; and technological change.
• Employs a fluid yet robust architecture re-
flective of evolving enterprise models and
evolving technology like mobile wireless; reFerences
• Utilizes database and data warehouse mod-
els/solutions to support transaction inten- Abboud, L., & Vara, V. (2007, January 23). SAP
sive applications (front office and back of- Trails Nimble Start-Ups As Software Market
fice), query intensive applications, OLAP, Matures. Wall Street Journal, C1.
and any other necessary internal and/or ex- Al-Mashari, M. (2002). Enterprise resource plan-
ternal interaction with the database or data ning (ERP) systems: A research agenda. Industrial
warehouse; Management & Data Systems, 102(3), 165–170.
• Enterprise systems take into account doi:10.1108/02635570210421354
partnering enterprise characteristics like
culture, language, technology level, stan- Arinze, B., & Anandarajan, M. (2003). A frame-
dards, information flows, and provide flex- work for using OO mapping methods to rapidly
ibility to adapt as partnering relationship configure ERP systems. Association for Comput-
changes; ing Machinery. Communications of the ACM,
• Solution vendors form global alliances 46(2), 61. doi:10.1145/606272.606274
with other vendors to better meet needs of
clients in any country;
32
Bhattacharjee, D., Greenbaum, J., Johnson, R., Green, J. (2003). Responding to the challenge. Ca-
Martin, M., Reddy, R., & Ryan, H. L. (2002)... nadian Transportation Logistics, 106(8), 20–21.
Intelligent Enterprise, 5(6), 28–33.
Higgins, K. (2005, May 23). ERP Goes On The
Burns, M. (2007, September). Work in process: Road. Information Week, 1040, 52–53.
Enterprise software survey 2007. CAMagazine,
ITtoolbox ERP Implementation Survey.
18-20.
(2004). Retrieved July 7, 2005, from http://
Chung, S., & Snyder, C. (2000). ERP adoption: A supplychain.ittoolbox.com/research/survey.
technological evolution approach. International asp?survey=corioerp_survey&p=2
Journal of Agile Management Systems, 2(1),
Knapen, J. (2007, May 14). SAP Sees Growth
24–32. doi:10.1108/14654650010312570
Ahead. The Wall Street Journal Online, http://on-
Conway, C. (2001). Top 20 Visionaries. [VARbusi- line.wsj.article_print/SB1179166926818022214.
ness: Manhassett.]. Comments of Craig Conway, html accessed 11 November 2007.
1724, 35.
Koch, C. (2004). Koch’s IT Strategy: The ERP
Dahlen, C., & Elfsson, J. (1999). An analysis Pickle. Retrieved June 16, 2005, from http://www.
of the current and future ERP market. Master’s cio.com/blog_view.html?CID=935
Thesis, Industrial Economics and Management.
Kremers, M., & Dissel, H. V. (2000). ERP system
The Royal Institute of Technology, Stockholm,
migrations. Association for Computing Machin-
Sweden.
ery. Communications of the ACM, 43(4), 52–56.
Dalal, N. P., Kamath, M., Kolarik, W. J., & doi:10.1145/332051.332072
Sivaraman, E. (2004). Toward an Integrated
Kumar, K., & Hillegersberg, J. V. (2000). ERP
Framework for Modeling Enterprise Resources.
experiences and evolution. Association for Com-
Communications of the ACM, 47(3), 83–87.
puting Machinery. Communications of the ACM,
doi:10.1145/971617.971620
43(4), 22–26. doi:10.1145/332051.332063
Davison, R. (2002). Cultural complications
Lee, J., Siau, K., & Hong, S. (2003). Enterprise
of ERP. Association for Computing Machin-
integration with ERP and EAI. Association for
ery. Communications of the ACM, 45(7), 109.
Computing Machinery. Communications of the
doi:10.1145/514236.514267
ACM, 46(2), 54. doi:10.1145/606272.606273
Enterprise Information Systems. A definition
Letzing, J. (2007, April 25). Big Rivals Move
from Wikipedia.com http://en.wikipedia.org/
In on Salesforce.com’s Turf. The Wall Street
wiki/Enterprise_information_systems, accessed
Journal, B3G.
20 November 2007.
Lope, P. F. (1992). CIMII: the integrated manufac-
Garbellotto, G. (2007, October). The Data Ware-
turing enterprise. Industrial Engineering (Ameri-
housing Disconnect. Strategic Finance, 59-61.
can Institute of Industrial Engineers), 24, 43–45.
Genovese, Y., Bond, B. A., Zrimsek, B., & Frey,
Markus, M. L., Tanis, C., & van Fenema,
N. (2001). The Transition to ERP II: Meet-
P. C. (2000). Multisite ERP implementa-
ing the Challenges. http://www.gartner.com/
tions. Association for Computing Machinery.
DisplayDocument?doc_dc=101237, accessed
on 7 July 2005.
doi:10.1145/332051.332068
33
Muscatello, J., Small, M., & Chen, I. (2003). Scheer, A.-W., & Habermann, F. (2000). Mak-
Implementing enterprise resource planning ing ERP a Success. Association for Computing
(ERP) systems in small and midsize manufactur- Machinery. Communications of the ACM, 43(4),
ing firms. International Journal of Operations 57–61. doi:10.1145/332051.332073
& Production Management, 23(8), 850–871.
Soh, C., Kien, S. S., & Yap, J. T. (2000). Cul-
doi:10.1108/01443570310486329
tural fits and misfits: Is ERP a universal solu-
Nattkemper, J. (2000). An ERP evolution. HP tion. Association for Computing Machinery.
Professional, 14(8), 12–15. Communications of the ACM, 43(4), 47–51.
doi:10.1145/332051.332070
Nikolopoulos, K., Metaxiotis, K., Lekatis, N., &
Assimakopoulos, V. (2003). Integrating industrial ACW Team (2004, August 23). SSA Global
maintenance strategy into ERP. Industrial Man- releases converged ERP with manufacturing
agement & Data Systems, 103(3/4), 184–192. capabilities. Asia Computer Weekly, 1.
doi:10.1108/02635570310465661
Willcocks, L. P., & Stykes, R. (2000). The role of
O’Brien, J. M. (2002). J.D. Edwards follows 5 with the CIO and IT function in ERP. Association for
ERP upgrade. Computer Dealer News, 18(12), 11. Computing Machinery. Communications of the
ACM, 43(4), 32–38. doi:10.1145/332051.332065
Sane, V. (2005). Enterprise Resource Planning
Overview. Ezine articles. Retrieved July 2, Zrimsek, B. (2002). ERPII: The Boxed Set. Re-
2005, from http://ezinearticles.com/?Enterprise- trieved July 7, 2005, from http://www.gartner.
Resource-Planning-Overview&id=37656 com/pages/story.php.id.2376.s.8.jsp
This work was previously published in Global Implications of Modern Enterprise Information Systems: Technologies and Applica-
tions, edited by Angappa Gunasekaran, pp. 17-31, copyright 2009 by Information Science Reference (an imprint of IGI Global).
34
35
Chapter 1.3
Exploring Enterprise
Information Systems
Malihe Tabatabaie
University of York, UK
Richard Paige
University of York, UK
Chris Kimble
Euromed Marseille École de Management, France
AbstrAct concept of EIS, i.e., identifying what is and what is

not an EIS. Based on this domain analysis, a proposal
The concept of an Enterprise Information System for using goal-oriented modelling techniques for
(EIS) has arisen from the need to deal with the building EIS is constructed; the proposal is made
increasingly volatile requirements of modern large- more concrete through illustration via an example.
scale organisations. An EIS is a platform capable of
supporting and integrating a wide range of activities IntroductIon
across an organisation. In principle, the concept is
useful and applicable to any large and SMEs, inter- This chapter focuses on a grand challenge for
national or national business organisation. However, organisations: dealing with their evolving require-
the range of applications for EIS is growing and ments and goals, and the impact of these changes
they are now being used to support e-government, on their Information Technology (IT). In particular,
health care, and non-profit / non-governmental we are interested in large-scale organisations such
organisations. This chapter reviews research and as multi-national companies, or public-sector or-
development efforts related to EIS, and as a result ganisations, which are sometimes called enterprises
attempts to precisely define the boundaries for the in the literature.
Organisations use IT in many different ways: to
DOI: 10.4018/978-1-60566-856-7.ch021 facilitate communication, to support commercial
Exploring Enterprise Information Systems
transactions, to advertise, etc. In order to under- • The increasing rate of organisational and
stand the effect of organisational and enterprise social change.
changes on use of IT, we start by defining the
nature of an organisation. The current literature By investigating the validity of this assumption,
defines that an organisation is thus about a group and the importance of these factors, this chapter
of elements (human, automated system, structure, aims to contribute a better understanding of Enter-
policy etc) that are arranged in a specific manner prise Information Systems (EIS), their dimensions,
to accomplish a particular purpose (Buck, 2000; their boundaries, and the challenges that arise in
Laudon & Laudon, 2007; Terry, 1975). This defi- their construction and development.
nition applies to small, medium, and large-scale As part of this investigation, and as a result of
organisations. the analysis of the literature that commences in
As we said earlier, a large-scale organisation can the next section, we propose one key challenge
sometimes be designated by the word enterprise. for understanding and building EIS:
However, we find it helpful to be more precise in
defining enterprise; in our view, an enterprise is a • Understanding diverse and volatile stake-
large-scale organisation that is involved in, and must holder requirements.
orchestrate, more than one independent business
processes. We come to this definition by observing To aid in understanding these constructs,
that many organisations, such as small IT houses, we propose the use of goal-oriented modelling
engage in a single business process. Identically techniques; this is discussed in the last section
some large organisations, such as online retailers, of this chapter.
have a single business process. Organisations that The rest of the chapter is organised as follow:
have many different business processes, that must be The background section outlines the challenges
coordinated in some way, such as Mitsubishi, have in large-scale organisations as a motivation for
different requirements and different characteristics. discussing the systems that can address these
Such organisations are often very large scale (e.g., challenges. A specific instance of large-scale
public health organisations) and multi-national. In organisations is an enterprise; hence, section 2
our view, the need to coordinate different business also discusses the requirements of IT systems for
processes is a key characteristic in distinguishing enterprises. One of the main difficulties in this
an enterprise from another organisation. area is the imprecise definition for EIS, and how
This paper investigates the validity of an as- an EIS differs from a general purpose IT system.
sumption regarding the root of complexity of IT Hence, we provide a working definition for EIS
systems in complex organisations, where the IT in this section.
systems support business processes directly. The The Enterprise Information System section
assumption is that complexity is due to the fol- describes EIS in more detail by discussing state-
lowing factors: of-the-art definitions and effective elements, such
as business and organisation, based on a literature
• Increasing size of IT systems and the or- review. The future trend section describes goal-
ganisation itself; oriented modelling techniques as a promising
• The interactions between different IT approach for attacking one of the main challenges
systems; of building an EIS by making the system more
• The involvement of many different or- clear for its stakeholders. Section 4 also provides
ganisations in the constructions and use of an example to clarify this idea.
these IT systems; and,
36
bAckGround 1. Lack of clear connections between the project

and the organisation’s key priorities, includ-
A brief review of the history of enterprises and ing agreed measures of success
software systems helped us to construct a work- 2. Lack of clear senior management and
ing definition for EIS. This working definition is Ministerial ownership
our basis for presenting an argument about what 3. Lack of effective engagement with
is and what is not an EIS, and for refining our stakeholders
understanding of the objectives for this type of 4. Lack of skills and proven approach to project
systems. This section therefore discusses some management
examples of EIS to shape the argument. 5. Lack of understanding of and contact with
the supply industry at senior levels in the
challenges of large scale organisation
software system 6. Evaluation of proposals driven by initial
price rather than long term value for money
Since the 1950s organisations have been devel- (especially securing delivery of business
oping computer-based information systems to benefits)
support their business processes. Through im- 7. Too little attention to breaking development
provements to IT, computer based systems have and implementation into manageable steps
become more complex and yet more reliable; 8. Inadequate resources and skills to deliver
therefore increasing functional requirements have the total portfolio
been placed upon these systems (Edwards, Ward,
& Bytheway, 1993). However, building this kind The first item in this list refers to the concep-
of system has many challenges, including fun- tual gap between project priorities and those of
damental challenges regarding the construction organisations; later in this chapter, more discus-
of such systems, and the challenges of evolving sions address this challenge. In addition to these
systems to accommodate new requirements. causes, hidden challenges threaten the IT projects;
Understanding the challenges of building such IT in particular the large-scale ones. For example,
systems is essential for planning, designing, and stakeholders should understand the conditions
development in order to provide as early as pos- and limitations of the system. Having unreliable
sible risk understanding, as well as understanding expectations from the system can move the domain
of the potential means for mitigation. of the project out of its limits and cause failure.
The challenges of understanding and building Another important and hidden challenge is
large-scale software systems can be observed in the lack of visualisation in the software systems.
both the public and private sectors. In the public Software is not visible and tangible for the stake-
sector, understanding the challenges, and reflecting holders; therefore, stakeholders cannot picture the
based on these challenges during the development functionality of the software before it actually
process, is important because failure (whether built, which can cause unrealistic expectations
financial or otherwise) can result in significant and other undefined problems. For example, in
damage to the reputation of the government. the case of constructing a building, stakeholders
The National Audit Office/Office of Govern- can visualise the building by looking at its mock-
ment Commerce lists the common causes of the up; in the case of software there is no such a clear
project failure as follow (Projects, 2004): and easy to understand mock-up.
Flexibility and supporting changes are other
challenges that software systems should deal with.
37
It is important to note that software systems can What Is an enterprise?

improve the speed of the processes in organisa-
tions and deal with the complex and well-defined The literature is not rich on the history of enterpris-
processes. However, they are not intelligent es; however, Fruin (1992) is one of the researchers
enough to improve the business model; hence, that explained the history of enterprises briefly and
software systems are not the solution for the ill- with an eye on the Japanese revolution in industry
defined business model. This challenge can be and business. According to this book:
seen mainly in large-scale software systems that
deal with businesses in organisations, such as The enterprise system appeared around the turn
EIS. The term Enterprise Information System is of the twentieth century when the factory system
a common term in today’s industry, which suffers was effectively joined with a managerial hier-
from misinterpretation and an imprecise defini- archy in production and distribution. It is the
tion. The rest of this chapter discuss this type of emerging coordination of previously independent
systems in more detail. organizations for production, management, and
distribution-shop-floor, front office, and sales of-
large scale software system: fice- that generates the organizational innovation
enterprise Information system known as the Japanese enterprise system. (Fruin,
1992, p. 89).
A specific kind of large scale IT system is those
that support enterprises. We call these software According to Fruin (1992), the notion of
systems, EIS. The business aspect of organisations an enterprise system was established after the
motivates engineers to develop systems that satisfy First World War, when new industries came to
real requirements of organisations, particularly the market and many industries combined and
requirements associated with business processes. amalgamated. Three types of enterprises were
As a result, technologies such as Service Oriented identified: National, Urban and Rural; which all
Architecture (SOA) are currently popular in design have some common elements such as inter-firm
and implementation of systems for businesses. relations, Marketing, Mode of Competition,
However, the term business often implies a process Finance, Ownership, Management, Administra-
that focuses on delivering financial value; but in tive Coordination, Government Relations.
practice, large-scale processes, and their associ- Mitsubishi is an example of an enterprise
ated IT systems, i.e. EIS, can support delivery of dating back to 1926; it integrates distinct yet
different kinds of outcome, which are not always affiliated companies, particularly Mitsubishi
directly linked to financial value. In fact, today’s Heavy Industry, Mitsubishi Warehousing, Mit-
businesses include both financial organisations subishi trading, Mitsubishi Mining, Mitsubishi
as well as public organisations, which deliver Bank, Mitsubishi Electric, Mitsubishi Trust,
services to the public. The success or value of Mitsubishi property, Mitsubishi steel, Mit-
these types of services is not always evaluated subishi Oil, Nippon Industrial Chemicals, and
by the financial results they deliver. Mitsubishi Insurance (Fruin, 1992). There are
To commence our main discussion on EIS, many other examples of enterprises including
we first discuss enterprises; in our view, an EIS Boeing, General Electric, Kodak, IBM, Norwich
supports business processes of an enterprise. It is Union, Samsung, and Philips. From a consumer
important to have an understanding of an enterprise or client’s point of view, these enterprises are
to understand what an EIS is. often perceived as involving only one single
organisation (e.g., Mitsubishi’s car division).
38
Another example in this area is General Electric, enterPrIse InFormAtIon

which has independent divisions focusing on systems
healthcare, aviation, oil and gas, energy, elec-
trical distribution, security, and many others Introduction
(GeneralElectric, 2008).
History shows that enterprises have existed Based on the working definition developed in the
from the turn of the twentieth century; neverthe- last section, in this section we focus on refining the
less, the concept still suffers from an unclear definition to include additional detail, particularly
definition. in the organisational and business context. As a
result, this section proposes a concrete definition
conclusion for EIS. To help explain the definition further, and
partly to validate it, we relate it to well-known
Today’s large-scale IT systems increasingly examples of organisations.
provide support for the business processes of
organisations. The aim of using information challenges
systems is to increase the automation of the pro-
cesses within organisations. Enterprises integrate The notion of enterprise is a widely used term
organisations, departments, and even entire busi- for instance in the case of Mitsubishi. However,
nesses to achieve shared goals. Processes within a precise definition of what constitutes an enter-
enterprises can benefit from IT infrastructure; in prise – and hence, what precisely constitutes an
this section, we have argued for calling such IT EIS – is still missing. One of the main difficulties
infrastructure an EIS. in defining what is an EIS is in distinguishing it
from any other large-scale software system. For
Working definition example, perceived challenges in designing and
developing an EIS will arise in the form of hav-
From the discussion on the history of enterprises ing to meet fixed costs of development, in deal-
and challenges of large scale software systems, ing with volatile requirements, and in managing
we see that EIS are computer-based systems that the complexity of the EIS. However, these are
satisfy the requirements of enterprises. EIS are also challenges all kinds of large-scale software
designed and developed for enterprises rather than systems. Therefore, we do not aim to enumerate
a single business unit or department. They can deal all of the design and development challenges of
with the problems of a large organisation (which EIS; instead, this section will address one of the
includes different SMEs or different partners), essential challenges, which is unclear definition
and they can deal with the problems of a medium for EIS; hence, we aim to propose a definition
or small enterprise (which is an organisation that for this term.
includes different departments). To define EIS, this study reviewed the current
This working definition will be refined in definitions found in the literature; the next section
later sections. After providing a brief background will cover some of them.
for EIS, we will discuss the definition in more
detail. state of the Art definition
Organisations continue to find that they need

systems that span their entire organisation and
tie various functionalities together. As a result, an
39
understanding of enterprise systems is critical to tion should be visible. This security policy does
succeed in today’s competitive and ever changing not have any contrast with the idea of enterprise
world (Jessup & Valacich, 2006). processes, which their goal is to let the informa-
A good definition for EIS introduced it as a tion flow seamlessly.
software system with the specific ability to inte- Moreover, (Strong & Volkoff, 2004, p. 22)
grate other software systems of an organisation. defines an ES as a system which its task is to
support and “integrate a full range of business
Enterprise systems integrate the key business processes, uniting functional islands and mak-
processes of a firm into a single software system ing their data visible across the organization in
so that information can flow seamlessly through real time”. This definition adds to the previous
the organization, improve coordination, efficiency, definition, the fact that the data and information
and decision making. Enterprise software is based entailed by the system should be understandable
on a suite of integrated software modules and by all its business processes.
a common central database. The database col- Another definition for enterprise systems is
lects data from and feeds the data into numerous based on legacy systems; a legacy system is an
applications that can support nearly all of an existing computer system or application program,
organization’s internal business activities. When which continues to be used because the company
new information is entered by one process, the does not want to replace or redesign it (Robertson,
information is made available immediately to other 1997). Most established companies, who have been
business processes. Organization, which imple- using a system for long time, are in this group.
ments enterprise software, would have to adopt Legacy systems mainly suffer from deficiency of
the business processes embedded in the software documentation, slow hardware and difficulties in
and, if necessary, change their business processes improvement, maintenance and expansion. How-
to conform to those in the software. Enterprise ever, there is evidence that overtime EIS replaces
systems support organizational centralization by the stand alone applications and the functionality
enforcing uniform data standards and business of legacy systems (Strong & Volkoff, 2004). In
processes throughout the company and a single contrast to enterprise systems, legacy systems are
unified technology platform. (Laudon & Laudon, not designed to communicate with other applications
2007, p. 382) beyond departmental boundaries (Jessup & Valacich,
2006) even if middleware offers a potential solution
This definition seems very specific on what is to adapt the novel parts with the legacy system.
an EIS; however, there are points that are ignored Nevertheless, regarding the price of developing a
by this definition. For example, the argument that middleware, the following question comes to mind:
mentioned when new information is entered by can middleware alone solve the problem of integrat-
one process, the information is made available ing new subsystems with a legacy system?
immediately to all other business processes. How- In short, the common idea in the existing
ever, it can be argued that the information should definitions illustrates that an EIS is about various
be available to the other processes depending on businesses, business processes, organisations,
their access domain. By this, we mean the level information systems, and information that circu-
of access to the information should be different lates across the enterprise. In other words, EIS is
from process to process. It is not reasonable to about the businesses model in the organisation.
expose information to the processes, which do not Therefore, the two main elements of EIS are
require it. Therefore, based on the access level of organisation and business. The two following
processes, only the suitable and updated informa- sections cover these points.
40
organisation • Satisfy their owners
The EIS definitions that we extracted from the All the above goals focus on increasing the
literature linked the EIS to organisations (Laudon market demands for products or services.
& Laudon, 2007; Strong & Volkoff, 2004) or Examples of not for the profit organisations
large companies (Jessup & Valacich, 2006) and could be charities, mutual societies, etc, which
we assume that in both cases the definitions refer provide some services for the society. The cus-
to the same concept: organisation. Based on this tomers are also the member of the mutual society;
assumption, it is vital to review the different types therefore, they are the owner of the business. The
of organisations that can influence the different value for money rule exists in this group too. The
types of EIS. Therefore, this section discusses usual way to evaluate the success of this group of
categorise of organisations based on their goals. organisations is to measure how well they achieve
Elizabeth Buck categorises organisations in three their goals considering the available resources.
groups (Buck, 2000): Table 1 illustrates some of the characteristics of
organisations that were described; it also sum-
• Public Organisations marises the different type of organisations.
• Private Organisations By understanding the categories of organisa-
• Not for Profit Organisations tions, we can focus on understanding their goals.
By knowing the goals of organisations we can
The public organisations include central or design and develop an EIS that satisfy the defined
local government, where elected members (e.g., requirements and goals; but there are other ques-
minister) will decide on the goals of organisations, tions in this area: what are the EIS’ goals? Are
and may influence how goals are achieved. The aim the goals of EIS similar to the goals of organisa-
of this type of organisation is to supply services to tions? It seems that EIS’ goals could be a sub set
or for the public, considering a ‘value for money’ of the organisations’ goals. When the EIS’ goals
rule. Examples of this type of organisation can get closer to the goals of organisations it could
be health service, prison, police, social security, become a better EIS. The final and optimistic goal
environmental protection, the armed forces, etc. for an EIS is to improve the goals of the organisa-
Individuals or other private organisations tion it services. However, defining the goals of
own private sectors organisations. This group of an EIS is the path for analysing and developing
organisations can have the following goals: the organisation’s business model and thus the
next section will explore the role of business in
• Satisfy their customer the definition of EIS.
• Satisfy their staff
Table 1. Organisations’ categories
Type of Decision Makers Value for Owner Goal(s) Example

Organisation Money
Public Elected members Yes Public Supply Services to or UK central
for the Public Government
Private Share holders No Share holders Satisfy customers/ Satisfy Mitsubishi
staff/ Satisfy owners
Not for Profit Elected Manager Yes Members/ Provide some services for the NCH (Children
Customers society or members Charity)
41
business on one particular business, an enterprise focuses

on a collection of business processes which could
Another main factor that influences the architec- be relevant to each other or not but all of them
ture and functionality of an EIS is the business are under the arch of the main principals of the
model (Figure 1). Supporting the strong relation- enterprise. Indeed, making profit is not one of the
ship between business processes is the aim of essences of business model. There are non-profit
ES. In fact, the ability to define various business governmental or non-governmental organisations
processes in enterprise systems is the element such as healthcare organisations that can have
that distinguishes them from normal systems for their own business model which deals with the
a company or a department; for example BMW process of treating patients.
involves in a diversity of businesses to produce The presentation of Enterprise System in this
cars or engines for other car brands (e.g. Rolls- chapter is not about detailed implementation of
Royce), in addition to building bicycles and business functions; its focus is mainly about a
boats. A normal system in a company contains very top-level view on the whole business model
components and subsystems that belong to one of an organisation as defined by Clifton (2000):
specific business and satisfy its requirements.
A normal company may need to contact other business involves a complex mix of people, policy
companies to continue its business but involving and technology, and exist within the constraints of
partners or suppliers is not their main concern. In economics and society (Clifton, Ince, & Sutcliffe,
contrast to normal company where the focus is 2000, p. 1).
Figure 1. Business model [based on (Kaisler, Armoiur, & Valivullah, 2005)]
42
Figure 1 illustrates the general structure of a ies (Jessup & Valacich, 2006); hence the concept
business model where the business model includes of EIS has grown to fill this gap.
business processes and business functions. Busi- In short, the common idea in existing defini-
ness processes are “a set of logically related tasks tions illustrates that an EIS amalgamates concerns
performed to achieve a defined business outcome” from various businesses, business processes,
(Davenport & Short, 1990, p. 100). For example, organisations, information systems, and informa-
in the case of BMW, the business processes is put- tion that circulate across an enterprise. In other
ting new orders for part suppliers. When there is a words, it is about the business models of the
new demand for specific car (e.g. model Z5), this organisation. However, a definition for EIS that
new market request creates a business event that just emphasises the financial profit side of busi-
triggers a set of business processes such as increas- nesses for organisations is out of date. In the next
ing the amount of resources for producing the Z5 section, a definition that considers other aspect
(e.g. BP2 in Figure 1), and putting new orders for of organisations, the domain and objectives of
parts suppliers. Each of these business processes is EIS is proposed.
subdivided into different business functions (e.g.
BF2 and BF3 in Figure 1). Examples include the enterprise Information
functions required for inputting new orders such system definition
as checking the parts suppliers’ ability for new
demands, organising the time that is needed for This section proposes a definition for EIS, which
each part to arrive to assembly line, etc. is the result of our analysis of the state-of-the-art
According to (Kaisler, Armoiur, & Valivullah, definitions and of industrial case studies. The defi-
2005, p. 2) “business processes must be modelled nition that considers business and organisational
and aligned across the function, data and infor- aspects of EIS is as follow:
mation systems that implement the processes”. An Enterprise Information System is a software
Therefore, the term business function in our re- system that integrates the business processes of
search refers to the functionality that is required organisation(s) to improve their functioning.
for implementing a business process. Figure 1 is Integration of business processes plays an
a simple explanation for business process model. important role in this definition. Integration could
Each of these business functions can trigger a busi- be accomplished by providing standards for data
ness process too. Moreover, the business processes and business processes. These standards will be
can breakdown to other business processes, which applied to various part of the system such as a
is not shown in this diagram to keep it simple to database or clusters of databases. As the result
understand. The aim of this diagram is mainly of such integration, information may flow seam-
to explain business processes and functions in a lessly.
general business model. Another point in this definition is the software
Understanding business models is helpful for characteristics of EIS. At this stage, we consider
developing EIS because their role is to integrate EIS as a type of Information System; therefore,
a full range of business processes (Strong & this software system includes both humans and
Volkoff, 2004). Before defining the concept of hardware.
EIS, Legacy systems were the type of systems The next term, used in the definition is organisa-
that were developed to handle the requirements of tion. Different types of organisations are discussed
organisations (Robertson, 1997). However, legacy earlier in this chapter. Organisations may include
systems are not designed to communicate with an organisation with its partners, or a group of
other applications beyond departmental boundar- organisations. Table 2 refines the above definition
43
Table 2. EIS boundaries, objectives and challenges
Objective Integrity of the organisation and collaborators

Seamless Information flow
Suitable access to data and information for various stakeholders
Matching the software system structure with organisation structure
Goal Improving coordination, efficiency, and decision-making of business process in an organisation
Domain Covers the internal and external business activities of organisation
Challenge Security challenges that should be considered carefully for organisations’ processes. Otherwise, mixing the required informa-
tion of one business process with another one can cause problem for the organisation
Improve flexibility in organisation processes
and describes what we propose as the objectives, concern about where the data is and what are the
goals, domain, and challenges of EIS. various resources. As can be seen in this figure,
In addition, Figure 2 describes the definition the bigger rectangle describes the boundaries of
of EIS graphically. Note that BP in this figure are EIS, which is flexible.
business processes. As it can be seen in this figure, The two following sections aim at continu-
each organisation contains various business pro- ing the discussion about EIS by presenting some
cesses. Moreover, in Figure 2, the database could examples in this area. The results of reviewing
be a cluster of databases; however, it is highly these examples lead to a better clarification of
likely that there would be a single interface to what is an EIS and what is not.
exchange data with the database without having a
Figure 2. An enterprise information system
44
examples of enterprise What cannot be an enterprise

Information system Information system
The review of the industrial cases of what might As we were discussing the public, private, and
be considered as an EIS moves our discussion governmental examples for an EIS, the next
toward the example of Mitsubishi. As was men- step is introducing some examples of Informa-
tioned earlier, Mitsubishi with more than 400 tion Systems that are not EIS according to our
companies all around the word is an example of definition.
enterprises (Mitsubishi, 2007). Thirty top-level eBay is one of the well known international
managers manage all the individual Mitsubishi’s Information Systems that focuses in the auction
companies. This does not mean that each company industry. This online market which involves
does not have enough freedom to make their own around 147 million people (Gopalkrishnan &
decisions; it means that this group of thirty manag- Gupta, 2007) provides a platform for individuals
ers will make some of the top-level decisions and or companies to trade their products or services;
they provide the high-level standards that all these but it does not connect the business processes of
companies should consider. In this case, if there is organisations together. Therefore, according to
a computer based system that links various parts our definition, an EIS connects different busi-
of the Mitsubishi organisation (including high- ness processes of organisations or departments
level managers) together and makes information of organisation together to make the information
flow seamlessly between them, then we view this flow seamlessly and thus it seems that based on
system as an EIS. Developing such a system is this characteristic of EIS, eBay is not an EIS. The
a large and complex problem; hence, there is a information system is the element that processes
need for powerful, reusable solutions to develop data and put them online, there is no evidence of
this type of system in a manner that can benefit connection between business processes because
all of the enterprise. it is not a requirement in this Information System.
Another example in this area is the infrastruc- The same argument can be followed in the case of
ture being developed to support the National Amazon, therefore even though it is large-scale
Health Service (NHS) in the United Kingdom and international online shop but it is not an EIS.
where the information systems being developed
to support management of patients’ records and conclusion
prescriptions can be considered as an EIS, because
such IT infrastructure aims to connect independent In short, this section described EIS in more detail
departments within and outside of the NHS. While by providing the definition for EIS. Defining any
we are looking at the NHS, which is a public sec- kind of system is essential for defining its domain
tor organisation, we can raise e-Government as and objectives. Without this basic information,
another example of public sector organisation that the researches on the similar area will not be
may be supported by and hence benefit from, EIS consistent. However, there is no claim that the
infrastructure because it connects various govern- given definition is the only definition for EIS.
mental organisations or departments together to let This definition is based on our studies, observa-
information flow seamlessly between them. tions, interviews, and comparisons on the current
theoretical and practical definitions and case
studies. Part of this ongoing work is presented
in this chapter.
45
To make the results of our study on the defini- sweerde, 2004). Goals are also used in the safety
tion of EIS more clear, two examples are discussed and security research community – for example,
in this section. The earlier examples describe the to present safety cases and safety arguments (T.
case that can be an EIS and the case that cannot Kelly, 2004; T. Kelly & Weaver, 2004) - and in
be an EIS. This categorisation is based on this software assessment (Weiss, Bennett, Payseur,
chapter’s criteria, which is discussed in the given Tendick, & Zhang, 2002).
definition. Hence, each of these cases could be the Kelly (1998) defined a goal as ‘requirements
objective for more discussions on the possibility of statement’, Lamsweerde (2003) used goals as
being an EIS or not. Considering different point of criteria for deriving software architecture. Kim
views and the context of arguments, one Informa- et al (2006) defined goal model from (Axel van
tion System can be an EIS or not. Therefore, it is Lamsweerde, 2003) point of view as a criteria for
crucial to consider the writers’ point of view and designing the architecture for systems; therefore,
given definition, in the preceding examples. the aim of software architect is to implement a
After discussing what can be an EIS, the next system based on the architecture to accomplish
section will focus on an approach for developing goals (Kim, Park, & Sugumaran, 2006). Logically
this type of system. goals are the motivation for developing a system;
therefore, all the stakeholders should have a clear
understanding about the goals of the system. In
Future trends addition, the goals of the system should be realisti-
cally defined before continuing any other step of
Goal-based and goal-oriented thinking is used the development. There are attempts to show the
to plan for the future or to solve problems (Kim, goals in graphical notations such as GSN (Timothy
Park, & Sugumaran, 2006). The concept of us- Patrick Kelly, 1998), Kaos (Axel van Lamsweerde,
ing goal-oriented techniques has been proposed 2001), and (Kim, Park, & Sugumaran, 2006).
as one possible way to manage some of the dif- Moreover, Kaos defines the formal textual notation
ficulties associated with developing large-scale to describe the goals in addition to informal text.
complex systems (Kavakli, Loucopoulos, & This attempt is respectful because it considers the
Filippidou, 1996), particularly the challenge of larger group of audience to understand and benefit
clearly identifying and specifying requirements. from the goal model.
As we discussed in the previous section, an EIS Different stakehold ers require different forms
is an instance of large-scale complex system. This of presentation for their goals. For example, the
section promotes the idea of using goal-oriented high-level manager may not require seeing a for-
modelling techniques for developing EIS by mal explanation of the goals because they may not
briefly discussing them and their roles in defin- understand it; however, they can better understand
ing EIS system requirements. We will summarise an informal explanation in a simple diagram. On
our discussion on goal-oriented techniques by the other hand, there is a good possibility that a
presenting an example of goal graph. programmer’s team, requires the formal expla-
nation for the goals in detail to understand and
Goal oriented techniques implement the system in the correct and expected
way. It is important to bear in mind that goal dia-
Goal oriented techniques have been widely dis- grams are aimed at making the system more clear
cussed in the requirement engineering domain to different stakeholders, therefore goal-oriented
(T. P. Kelly, McDermid, Murdoch, & Wilson, ideas should prevent adding more confusions for
1998; Axel van Lamsweerde, 2001; A. V. Lam- different stakeholders. Any approach that makes
46
the goals of the system more clear for stakehold- unit for representing software behaviour, which
ers should be considered; it can be different goal can be observed at runtime and has a start and
models for different stakeholders. end point of execution”. This paper also argues
The next section explains an approach for de- that most methods in the class diagram can be
signing a goal model. This approach is very high action, but because the runtime of actions should
level without explaining the details. The aim is be observable, the size of the action should be
to introduce a possible approach for developing restricted in a manner that makes it possible to
a goal model to readers. This approach benefits be observed in the software model.
from the information in similar studies in this The issues in this case are the challenges and
area such as (Timothy Patrick Kelly, 1998; Kim, difficulties that occur when developers consider
Park, & Sugumaran, 2006; Axel van Lamsweerde, the implementation and the execution of a system.
2001). These challenges can be a technical difficulties,
or goal conflicts, etc.
designing a Goal model After producing the goals, actions, and prob-
lems lists, the relation between these elements
One of the main reasons for developing unsuc- should be created. The notation here is similar to
cessful software systems is unrealistic planning the notation in (Kim, Park, & Sugumaran, 2006),
and design. Hence, the aim of goal-oriented ap- which is as follows:
proaches, as discussed in the previous section,
is to provide an environment such that different (Gz, An) → Px
stakeholders can understand the goals at dif-
ferent levels of abstraction and decomposition. A represents an action, G represents a Goal,
One way to accomplish this is to use a graphical and z and n are the symbolized identification of
modelling language, such as GSN; another way random variables that present the ID of the goal,
is by documenting the requirements and design for instance, it can be G 1.1.2, which means goal
precisely and accurately using a textual format. with ID 1.1.2. An example of action could be
It is also possible to present the prototype of the 1.1.2/1, which presents the required action that
system and discuss it with various stakeholders. can be done to achieve this goal.
All these approaches and other similar ones could The next notation illustrates the relationship
be beneficial for different type of systems. The between a goal, action, and problem. Following
approach that is discussed in this section is a is an example of this notation:
simple approach for developing a goal model.
The aim is to develop a goal model that can Px → (Gz, An)
present the system’s high-level goals clearly.
Furthermore, it does not involve the details of The above notation means, the Action with ID
the goals or their descriptions; this can help to n which is required for satisfying Goal with ID z
provide an understandable top down model for can cause the problem with ID x. This notation
high-level goals of the system for non-technical describes the case where action that belongs to
decision makers. a goal causes a problem or problems. The next
The basics of this approach for designing a notation describes the case that a problem can be
goal model is to create a list of goals, a list of solved using a specific action:
actions, and a list of occurred problems. Goals
were defined earlier; actions according to (Kim, Px → (Gz, An)
Park, & Sugumaran, 2006, p. 543) “are the atomic
47
The above notation means to solve the problem is shown as a box surrounding it. The big arrow
P with ID x, the Goal G with ID z is required, in the left hand side of the box illustrates the fact
and to satisfy this goal Action with ID n should that this system is one of the information systems
be done. in the defined enterprise. The enterprise in this
In the case that the developer team does not figure is shown using a pyramid, which is mainly
know the required action yet, action n (An) can a symbol of organisation.
be replaced with ‘?’. Before starting to implement To make this figure simple and clear we did
the system, all the question marks (?) should be not include the option that this EIS can be shared
filled with actions as solutions to satisfy goals. and used with other enterprises around the globe.
Nevertheless, in the case that as the result of Note that by having a design for EIS, we try to
limitations in the technology, resources, etc. one have a big picture of enterprise that includes the
or more question mark (?) cannot be replaced by possible changes and extension in the future. The
solution, there could be a bottom-up check to see EIS does not have a local design that cannot be
if the system is still worth implementing; consider- used when changes occur. The current solution
ing the unsolved problem(s), the functionality of for extending a system or merging systems is
the system should not rely on the non available mainly developing middleware, which enterprise
solutions. architect should avoid relying just on middleware.
However as it was discussed before, a goal Considering that in some cases middleware can
model should have different levels of abstractions be so expensive that the organisation’s decision
for different users. Hence, designers should avoid makers may decide to use manual paper based
destroying the purpose of the goal model, which is system instead.
to make the systems goals clear for stakeholders After drawing an overall view of the requested
by mixing and presenting all the information to system, the goal of the system should be defined.
the ones who does not require it. The next sec- Each goal should have its own action, which
tion will provides an example of goal model for acts as a solution for the system and the possible
stroke care. problems. Figure 4 illustrates the goal diagram
for this system. This diagram is very high-level,
example which targeted non-technical decision makers.
This diagram is the starting point for creating a
The aim of the case study is to design the goal complete goal model for this system. As can be
model for a system that collects the data of treat- seen in this diagram, goals have their unit identity,
ment for a specific serious condition. The data which in this case is shown by numbers. These
can be collected from different sources such as numbers makes the traceability of goals possible
doctors, researchers, nurses, emergency staff, etc. within this model. In addition, it is possible to
Moreover, each of these stakeholders can have a implement it in tools for drawing diagrams. This
different way of communicating with the database, is a AND-OR graph and it means the parent goal
for instance, laptop, paper, phone, etc. The role with OR child can be satisfied when at least one of
of this system is to collect the data from various the child goal reach to the solution. It is similar to
sources, analyse them and provide some data as AND-OR in logic mathematic. Furthermore, goal
an output for different purposes. Based on the graph is a weight graph; hence, the goals in the
case study in (Bobrow.D.G, 2002) we can call same level can be prioritised over other goals.
this system a knowledge sharing system. Prioritizing goals is helpful in different context.
Figure 3 illustrates the described system. In For example, in allocating resources or in some
this figure the boundaries of the described system cases, when satisfying a lower priority goal is
48
Figure 3. Example of knowledge sharing system
Figure 4. Goal graph
49
depend on satisfying the higher priority goal. In cases is hardly the leading factor in developing an
general, Figure 4 shows the basic requirements EIS. The basic requirement for research on how to
for the goal graph. We emphasise that the aim of improve the development of EIS is to achieve more
this graph is to provide the high-level clear image knowledge on what an EIS is.
of the system’s goals and present it to the stake- Consequently, the main objective of this chap-
holders to be used for brainstorming for example. ter was to explore the boundaries of EIS; this was
achieved by developing a definition for EIS. This
conclusion definition captured what we believe are the impor-
tant characteristics that should be considered while
In conclusion, this section proposed the future we attempt to build an EIS; characteristics such as
work for the study on how to develop EIS. The organisations, their goals, business processes, and
fact that developing required components for the business model. None of these characteristics
EIS can be similar to other large-scale complex is based on the size of the organisation; therefore,
system makes this field of work valuable; because it can cover different sizes of enterprises, small,
finding better solutions for different challenges medium, or large. Accordingly, this chapter did
of Information Systems provides a platform for not use a specific term such as SME, Small and
developing various kinds of suitable systems. Medium Enterprises, to define EIS.
This effort and study on EIS provides an easier Any discussion of EIS encompasses a number
and safer life for individuals and organisations of facets, including general IT system develop-
that benefit from this type of IT products. It influ- ment, requirements, organisational theory, and
ences the government’s performance, it provides distributed systems technology. Our aim is to
better innovative platform for industries. All more precisely define what an EIS is, and what
these reasons bestow enough motivations for us it is not, to assist in providing better methodolo-
to continue improving this study. gies and techniques for building such increasingly
important software systems. We believe that it
is clear that the volatile requirements of modern
conclusIon organisations require special business processes,
and these business processes cannot be fully
By looking at various ways that the word ‘enterprise’ achieved without IT systems and in some cases
is used, it becomes clear that there is an ambiguity in without an EIS. A high-quality EIS can provide
this term. Yet this term and others such as ‘Enterprise a connection between the different, independent
Architecture’, ‘Enterprise Information System’, etc. business processes in an enterprise.
are increasingly used. This fact encourages us to look As discussed, we argue that goal-oriented
at these terms and clarify them for future use in our modelling techniques are important for helping
research and other relevant ones. The simple defini- to understand what is required for a business or
tion for enterprise is an entity engaged in economic organisation, and for understanding what an EIS
activities. This definition does not cover requirements should provide. Thus, we argue that a first step
for defining an EIS. The argument in this chapter for developing a system for an enterprise is to
illustrates that an EIS covering the requirements of find and justify the enterprise’s goals. When all
any entity engaged in economy activity is simply an the stakeholders have a clear idea about the goals
IS and they can hardly be categorized as a separate of the enterprise, their expectations will be real-
group with the name of EIS. The fact that the number istic in principle; the desired system’s boundary
of people employed by an organisation can increase should be more precisely defined, and in principle
the complexity of the software system in some building the system should be possible. We do not
50
claim that following this approach will provide Fruin, M. W. (1992). The Japanese enterprise
a full guarantee for developing a suitable EIS: system. New York: Oxford University Press.
such systems are always challenging to build, and
GeneralElectric. (2008). Product and services.
goal-oriented techniques only tackle an important
Retrieved September 1, 2008 from http://www.
part of a large problem. Additional research and
ge.com/products_services/index.html
experiments are needed to identify what further
techniques are needed to supplement goal-oriented Gopalkrishnan, J., & Gupta, V. K. (2007). eBay:
modelling for designing, implementing, deploy- “The world’s largest online marketplace” - A
ing, and maintaining Enterprise Information Case Study. Conference on Global Competition
Systems. and Competitiveness of Indian Corporate (pp.
543-549).
Jessup, L., & Valacich, J. (2006). Information
AcknoWledGment
systems today, why is matters (2nd ed.). NJ: Pearson
Education, Inc.
We would like to thank Dr. Fiona Polack for her
valuable suggestions. Kaisler, S., Armoiur, F., & Valivullah, M. (2005).
Enterprise architecting: Critical problems. Paper
presented at the 38th Annual Hawaii Interna-
reFerences tional Conference on System Sciences, Island of
Hawaii, HI.
Bobrow.D.G, W. J. (2002). Community knowledge
sharing in practice: The Eureka story. Society Kavakli, E. V., Loucopoulos, P., & Filippidou,
for organizational learning and Massachusetts D. (1996). Using scenarios to systematically
Institute of Technology, 4, 47-59. support goal-directed elaboration for informa-
tion system requirements. Paper presented at the
Buck, E. (2000). Different Types of organisation. IEEE Symposium and Workshop on Engineer-
NEBS Management/QMD Ltd. Retrieved Sep- ing of Computer Based Systems(ECBS ‘96),
tember 1, 2008 from http://www.teamsthatwork. Friedrichshafen, Germany.
co.uk/Organise%20&%20improve%20team%20
work%201.pdf Kelly, T. (2004). A Systematic approach to safety
case management. Paper presented at the SAE
Clifton, H., Ince, D. C., & Sutcliffe, A. G. (2000). 2004 World Congress, Detroit, MI.
Business information systems (6th ed.). Essex,
England: Pearson Education Limited. Kelly, T., & Weaver, R. A. (2004). The goal struc-
turing notation - A safety argument notation. Paper
Davenport, T. H., & Short, J. E. (1990). The New presented at the 2004 International Conference on
industrial engineering, information technology Dependable Systems and Networks (DSN 2004),
and business redesign. In M. Lewis & N. Slack Florence, Italy.
(Eds.), Operations management: Critical perspec-
tives on business and management (pp. 97-123). Kelly, T. P. (1998). Arguing Safety- A systematic
London and New York: Routledge. approach to managing safety cases. University
of York, York.
Edwards, C., Ward, J., & Bytheway, A. (1993).
The essence of information systems (2nd ed.).
London: Prentice Hall.
51
Kelly, T. P., McDermid, J., Murdoch, J., & Wil- Mitsubishi. (2007). About Mitsubishi. Retrieved
son, S. (1998). The goal structuring notation: A September 1 2008, from http://www.mitsubishi.
means for capturing requirements, rationale and com/e/group/about.html
evidence. In A. J. Vickers & L. S. Brooks (Eds.),
Projects, T. C. C. I. (2004). The challenges of
Requirements engineering at the University of
complex IT projects. Retrieved September 1,
York: University of York.
2008, from http://www.bcs.org/server_process.
Kim, J. S., Park, S., & Sugumaran, V. (2006). php?show=conWebDoc.1167
Contextual problem detection and manage-
Robertson, P. (1997). Integrating legacy sys-
ment during software execution. Industrial
tems with modern corporate applications.
Management & Data Systems, 106, 540–561.
doi:10.1108/02635570610661615
doi:10.1145/253769.253785
Lamsweerde, A. v. (2001). Goal-oriented require-
Strong, D. M., & Volkoff, O. (2004). A roadmap
ments engineering: A guided tour. Paper presented
for Enterprise system implementation. IEEE
at the 5th IEEE International Symposium on
Computer Society, 37, 22–29.
Requirements Engineering (RE’01), Toronto,
Canada. Terry, P. (1975). Organisation behaviour. Indus-
trial & Commercial Training, 7(11), 462–466.
Lamsweerde, A. v. (2003). From system goals to
doi:10.1108/eb003504
software architecture. Formal methods for soft-
ware architectures ( . LNCS, 2804, 25–43. Weiss, D. M., Bennett, D., Payseur, J. Y., Tendick,
P., & Zhang, P. (2002). Goal-oriented software as-
Lamsweerde, A. V. (2004). Goal-oriented require-
sessment. Paper presented at the 24th International
ments engineering: A roundtrip from research
Conference on Software Engineering (ICSE ‘02),
to practice. Paper presented at the 12th IEEE
Orlando, FL.
Joint International Requirements Engineering
Conference(RE’04,), Kyoto, Japan.
Laudon, J. P., & Laudon, K. C. (2007). Manage-
ment information systems: Managing the digital
firm (10th ed.). Prentice Hall.
This work was previously published in Social, Managerial, and Organizational Dimensions of Enterprise Information Sys-
tems, edited by Maria Manuela Cruz-Cunha, pp. 415-432, copyright 2010 by Information Science Reference (an imprint of
IGI Global).
52
53
Chapter 1.4
Enterprise Systems in Small
and Medium-Sized Enterprises
Sanjay Mathrani
Massey University, New Zealand
Mohammad A. Rashid
Dennis Viehland
AbstrAct of ES vendors, ES consultants, and IT research

firms in a NZ context. Implications for practice in
The market for enterprise systems (ES), continues implementation processes, implementation models,
to grow in the post millennium era as businesses and organizational contexts are discussed.
become increasingly global, highly competitive, and
severely challenged. Although the large enterprise
space for ES implementation is quite stagnated, IntroductIon
now all of the ES vendors are focusing on the
small to medium-sized enterprise (SME) sector for Enterprise systems (ES), also known as enterprise re-
implementations. This study looks at the current source planning (ERP) systems, are large, complex,
ES implementation scenario in the SME sector. highly integrated information systems designed to
The purpose of the study is to gain insights into meet the information needs of organizations and
what is a typical case of ES implementation and are, in most cases, implemented to improve orga-
to understand how current implementations in the nizational effectiveness (Davenport, 2000; Hedman
SME sector differ from the earlier implementations & Borell, 2002; Markus & Tanis, 2000). These are
in the large enterprise sector through a perspective comprehensive, fully integrated software packages
supporting automation of most standard business
DOI: 10.4018/978-1-59904-859-8.ch013 processes in organizations including extended
Enterprise Systems in Small and Medium-Sized Enterprises
modules such as supply chain management (SCM) literature. This is a replication study following
or customer relationship management (CRM) a similar approach used by Shakir (2002), who
systems. ES applications connect and manage also investigated aspects of ES implementation
information flows across complex organizations, in the NZ vendor-and-consultant community. The
allowing managers to make decisions based on focus of that study was to identify key drivers in-
information that accurately reflects the current fluencing ES adoption and implementation (e.g.,
state of their business (Davenport & Harris, 2005; Shakir and Viehland, 2004) whereas the focus of
Davenport, Harris, & Cantrell, 2002). In a more in- the current study is to understand how current
tegrated and global market, extended ES offers new implementations in the SME sector differ from
functions and new ways of configuring systems, the earlier implementations in the large enterprise
as well as web-based technology to establish the sector. The current study extends and builds upon
integrated, extended business enterprise (Shanks, existing ES research.
Seddon, & Wilcocks, 2003). The market for ES Semi-structured interviews were conducted
continues to grow despite much speculation on with key players of ES implementations in New
its future in the post millennium era. Although Zealand including ES vendors, ES consultants,
most large enterprises have completed their ES and IT research firms to explore the current ES
implementations by now, the ES market continues implementation scenario. Several measures such
to grow in the small and medium-sized enterprises as the number of users, modules implemented,
(SME) sector. A number of research studies have cost of implementation, number of sites/locations
been conducted to establish and understand the where implemented, industry type, organization
critical success factors for ES implementations size, implementation phases, time to implement,
(e.g., Allen, Kern, & Havenhand, 2002; Bancroft, implementation partners, and levels of customiza-
Sep, & Sprengel, 1998; Holland & Light, 1999; tion were discussed to understand a typical case
Parr & Shanks, 2000; Plant & Willcocks, 2006; of ES implementation and current implementation
Sarker & Lee, 2000; Scott & Vessey, 2002; Skok practices. The empirical findings are analyzed
& Legge, 2001; Sumner, 1999; Yang & Seddon, and reported in this paper. This study has been
2004). However, there has been little research conducted in a New Zealand (NZ) context which
that examines ES implementation at the strategic can be extended to show current trends worldwide.
decision-making process level (Viehland & Shakir,
2005) and compares current implementations in
the SME sector with earlier implementations in reseArcH metHodoloGy
the large organizations. The purpose of this study
is to examine the current ES implementations Using a qualitative research methodology,
scenario in New Zealand. The main objectives data were collected by way of semi-structured
of this study are to explore what is a typical case interviews with ten key respondents in the ES
of ES implementation and to understand how implementation industry. The interviews were
current implementations in the SME sector dif- carried out between February and August 2006.
fer from the earlier implementations in the large The key respondents were senior ES consultants
enterprise sector. The study does so through a or senior managers in the organizations which are
practitioners’ perspective, with interview data key players in the field of ES in New Zealand,
collected from ES vendors, ES consultants, and principally major ES vendors, ES consultants,
IT research firms who are actively engaged in and IT research organizations (see Table 1). The
ES implementation. This approach differs from positions of the respondents included: director
the organizational approach usually found in the professional services, consulting manager, manag-
54
Table 1. Key respondents for the study
ES Vendors (Flagship ES products) ES Consultants IT Research

SAP NZ (SAP) PricewaterhouseCoopers NZ Gartner Limited NZ
Oracle NZ (Oracle, J.D. Edwards, PeopleSoft) Ernst & Young NZ IDC NZ
Microsoft NZ (Dynamics (earlier Navision)) KPMG Consulting NZ
Infor NZ (Mapics, SSA Global (earlier BaaN)) EMDA NZ
ing director, consulting practice director, partner tyPIcAl es

group manager, vice president, consulting partner, ImPlementAtIons In nZ
general manager, and business consultant.
The purpose of the interviews was to seek Findings in this study reveal that the ES market is
insights from experienced ES stakeholders and based on three different size segments -- the large
professionals in answering what is a typical ES enterprise segment, the medium-sized enterprise
implementation in New Zealand and what are and the small firm. Most respondents used the num-
the current ES implementation practices? Ques- ber of employees as a measure for organizational
tions were asked to extract information such as size, however some respondents used revenue.
the number of users, the modules implemented, Until recently, the focus of implementations was
cost of implementation, number of sites where on the large enterprise – businesses and govern-
ES were implemented, type of industry and the ment agencies with more than 500 employees
size of the organizations in terms of number of and revenue greater than $250M. But now the
employees and revenue. focus for new implementations has shifted to the
Contact was first established with the respon- SME sector. The higher end of medium-sized
dents through email and by phone. An introductory organizations in NZ employ between 100 to 299
letter explaining the study briefly and seeking staff and have revenue between $50M to $200M.
appointment for an interview was then sent to At the lower end of this segment, employees are
the respondents. On receipt of confirmation, the between 20 to 100 and revenue is between $10M
research information sheet along with the ques- to $50M. In the small organization segment in
tions was sent prior to the interview. The answers NZ, employees are less than 20 and revenue less
were then discussed during the interview. than $10M.
The respondents discussed ES implementa- Findings revealed that the large enterprises
tions based upon their perspective and experience in NZ could have 200 or more users in a typical
in terms of their ES applications, their clients, and implementation. SME-based implementations
their implementation methodologies. Ten face- could be between 20 and 200 users. A classifica-
to-face meetings took place at the respondent’s tion by consultancy firm IDC, provided as part of
organizations with one interview from each firm. the current study, shows the sizes of companies
The interviews lasted between 60 and 90 minutes in terms of number of users where ES is imple-
each. The interviews were tape recorded and mented as a percentage of companies in NZ. A
transcribed immediately after each interview. small organization with less than 20 users are 26%
The empirical findings were analyzed using the in a NZ context, a medium-sized organization
Nvivo 7.0 qualitative software tool and the infer- with 20 – 200 users at 49%, and any organiza-
ences reported. tion above 200 users are large at 25% in the NZ
market, as shown in Table 2. In another study by
55
Table 2. Number of users in NZ companies where

growth market in NZ is in the medium-sized mar-
ES is implemented
ket segment. The declining markets include large
Size of Organization Number of Users Percent in NZ companies which have already made their $10M
Large >200 25%
investment and are not looking to make another
Medium 20-200 49%
investment. “These companies might make a mil-
Small <20 26%
lion or half a million dollar investment but they
are not the growth area for software companies.”
It was evident from vendors’ perspective that
the large enterprise space is quite stagnated now
Parr and Shanks (2000), in an Australian context,
and all of the ES vendors are focusing on SME’s.
the number of users reported for small organiza-
Once an implementation is completed in an orga-
tions is less than 100, medium less than 200, and
nization it goes out for a recycle every 7-8 years,
large more than 200.
so these software vendors need to look for new
Most respondents agreed that typical imple-
customers. A general consensus among the soft-
mentations in NZ currently are predominantly in
ware vendors was that more opportunities exist
the medium-sized enterprise category with 20 - 200
in the SME space or even at the lower end of the
users whereas the range used to be 200 – 500 users
SME space. There are many small companies using
a few years ago. These findings also confirm the
home grown PC-based systems where their usage
Shakir (2002) study where the numbers of users
has outgrown their original requirements, but they
for majority implementations in NZ were found
have not been replaced. Or a hodgepodge of several
to be 100 in medium-sized organizations with
different types of disparate systems exist which are
revenue between $50M- $250M.
expensive to integrate and maintain. The software
Many small businesses below 100 users in NZ
vendors are trying to push into this space, although
are companies that are basically part of multina-
there are many challenges within that space such
tional corporations that implement solutions “but
as limited resources, lack of infrastructure or lack
are more governed out of the regional or global
of necessary in-house skills to cope with chang-
office and the NZ office really has no control”.
ing requirements associated with implementing
There are also many small businesses in NZ, the
an ES. The challenge facing software vendors is
usual 20 – 25 user organizations that find it hard
to figure out how they take the learning acquired
to justify an ES investment from a cost perspec-
in the large enterprise implementations down to
tive, and typically do not implement expensive
a smaller enterprise. And it is a fallacy to believe
enterprise solutions from large vendors. These
that large organizations require different informa-
businesses prefer smaller inexpensive fragmented
tion as compared to smaller organizations. They
solutions. The big players like SAP and Oracle
actually require quite the same information. It is a
have also moved their business model and are
different degree of how much they require. Since
now targeting smaller sites.
the SME space offered most opportunity, therefore
Microsoft revealed that there are about 3,500
larger players like SAP and Oracle are focusing
medium-sized companies in NZ with employees
on the SME sector which was overlooked earlier.
from 100 to 200, which have not used one of the
“Whenever any new technology evolves there are
traditional ES as their core technology and a large
leaders, followers, and the laggards. Typically the
proportion of these businesses did not have any
large enterprise organizations are the leaders but
technological solution to help them with their
now they are beyond the leader category and it’s
business problems. Microsoft believes they have
the SME now which are going up in that curve.”
400 of these as their current customer base. The
Traditionally SAP customers were the large
56
enterprises, companies like Fonterra, Telecom, may be a SME in NZ it may have some large
and Carter Holt Harvey in NZ. However, in the offshore subsidiaries and therefore the group may
last 5 years due to shift in focus on the SME’s, not necessarily be tiny. For example, Fisher and
SAP is now moving to smaller companies with Paykel might not be considered one of the largest
annual revenues $5M or more and are offering companies in NZ but if you consider its holdings
two different types of solutions. One is similar and offshore organizations, it is quite substantial
to a Microsoft based type solution called “Busi- from the NZ perspective of a large company
ness One”. The other is “All-in-One” and is the category that runs JDE. But typically, we apply
traditional SAP solution. The large enterprises the rule, an SME will be putting in JDE and if it
are probably old SAP or one of the Oracle family is a large enterprise it will be putting E-Business
member customers having implemented either Suite or PeopleSoft.”
PeopleSoft, JD Edwards, or Oracle systems. Now Most informants suggested that the ES
the focus of the software vendors is to compete implementations are typically divided into two
for winning upgrade or add-on contracts from “waves” or phases. The first wave or phase 1 is
such large enterprises. SAP recently bagged such the implementation of core ES modules such as
a contract from Fonterra which is an old loyal finance; materials management including purchas-
customer of SAP. But the main battlefield for new ing, warehousing and inventory; and operational
implementations is in the SME sector. modules including production management, pro-
The Microsoft respondent, who had earlier duction planning, logistics, sales, and distribution.
worked with other vendors (i.e. JD Edwards, Some companies also include HR and payroll in
PeopleSoft, and Oracle), remarked “there’s no the first phase. In the second wave or phase 2, the
such thing as a typical enterprise solution imple- companies implement supplementary modules
mentation in NZ”. According to him, the size of (also called ERP II by Gartner (Zrimsek, 2002)),
enterprises ranges from large with 500 or more which include collaborative scenarios such as
users through to SME with 5 to 10 users at the customer relationship management (CRM), supply
lower end and hence there is no typical orga- chain management (SCM), supplier relationship
nization. He suggested however, the size of an management (SRM), and management services
organization that may drive business intelligence applications such as business intelligence (BI).
(BI) and knowledge capital as information tools Companies consider extending applications to
are the medium to large organizations. Organiza- satellite businesses or international operations
tions that employ 100 plus staff and have an ES in the second phase. These findings differed to
with a user base of more than 40 are more mature the Shakir (2002) study which noted that phase
in the context of using information for business 1 implementations would always include the
benefit. Microsoft’s largest customer in NZ has core modules of financials plus one or two other
a user base of 130 while their smallest customer modules that may include SCM and CRM. This
has just two with majority of customers in the change can be attributed to the larger number of
range of 15 to 50 users in NZ. modules that ES vendors have put into the market
From an Oracle perspective, Oracle has three in the last five years and SME owners being more
ES product lines - Oracle E-Business Suite, ambitious in phase 2 implementations. Findings
PeopleSoft and JD Edwards (JDE) from their from the current study revealed that about 75%
acquisitions. The E-Business Suite and PeopleSoft organizations are still in phase 1 and comprise
are generally implemented in the larger enterprises mostly of new implementations in the SME sec-
and JDE in the SME’s. “Unfortunately, some of tor. Balance 25% organizations are large that have
these lines are blurred because while a company moved into phase 2.
57
Traditionally, the time for implementation due to the smaller size of organizations and their
used to be more. However, since the focus is limited funds for investment.
now towards SME’s, the implementation time One vendor explained the reason for the high
has decreased. Earlier, processes such as model- costs in large organizations compared to smaller
ling the organization and configuring the design ones is because large organizations have more
used to take time. Now, using accelerators such complex operations and require more customiza-
as preconfigured business processes, the time to tions. Their decision making processes may not
implement has been slashed, and so has the cost be as efficient, and require a lot of change man-
of implementation, which has made these systems agement procedures. On the other hand, smaller
much more affordable for the small businesses. companies are more agile, more decisive, and
In the Shakir (2002) study the length of time to especially more inclined to adopt a best practices
the first phase implementations varied from 2.5 implementation, greatly reducing customization
months to 2 years. However, in the current study and change management costs.
although the time for implementation varied be- Table 3 summarizes the time and costs esti-
tween different informants, the general consensus mates for ES implementations as suggested by
was that currently large projects take around 12 to various participants in the current study based
24 months and SME’s 3 to 12 months to complete. on the size of the project. The costs include soft-
The cost of implementation is related to the ware, number of licenses, hardware, consultancy,
number of modules, their types, the software pack- implementation, customization, and training costs.
age size and brand, the number of user licenses is- Locations are the one or more sites where the
sued, consulting costs, customization levels, along ES is implemented. Findings in this study show
with the vendor or third party implementation cost. that currently more implementations are multi-site
The hardware cost, and the cost of training and while in the earlier years implementations were
change management may be additional. Accord- more single site. This represents a continuing trend
ing to informants in this study, in the SME sector, to more multi-site implementations in NZ, first
cost is the most important factor in selecting an ES observed by Shakir (2002). Traditionally, in large
for implementation. This was attributed by them, organization implementations, each location had
Table 3. Time and cost for ES implementations
Key Participants Time for Implementation (Months) Cost Based on Project Size (NZ$)
SAP Large – 18 to 36 Very large > $10 M
SME – 6 to 18 Large - $2 M to 10 M
SME - $0.5 M to 2 M
Microsoft Large – 18 to 24 Very large – Multi-million $
Medium – 9 to 12 Mid-market – $0.5 M to 2 M
Small – 3 to 6 Small – $0.1 M to 0.5 M
Oracle Large – 18 to 24 Not answered in figures
SME – 6 to 12
EMDA Consulting Large – 10 to 12 Large > $1 M
SME – 3 to 6 SME – $0.2 M to 1 M
PricewaterhouseCoopers Large – 24 to 48 Very large – $10 M to 50 M
SME – 4 to 12 Large – $2 M to 10 M
SME – $0.5 M to 2 M
IDC Large – 9 to 18 Not answered in figures
SME – 3 to 9
58
its own implementation. However, companies in best value from a price perspective and from the
the SME sector are now optimizing by using one perspective of having the best experts involved
implementation at multi-locations because they in the project.
are finding it too hard to manage and maintain The post implementation and after sales sup-
separate implementations at all these locations. port from the software vendor or the implementer
“Organizations are realizing its no use having IT to the customer organization normally includes
administrators in all the locations doing a simi- three levels of support (see Figure 1). The first
lar task.” The growth in export markets of NZ level is at the customers’ end where the customers’
companies coupled with availability of Internet- super user (i.e. ES champion), determines whether
capable technology is also driving multi-site ES it is a “how to” question -- where the user does
implementations in NZ. These implementations not know how to use the system -- or something
now are single instance which as explained by SAP else. If the problem is related to the user or an
meant that only one installation of the software organizational issue it is resolved at the first level.
is made to run on one server but the software is If not an end-user problem, then is it a general
used at multiple locations. This environment is business requirement issue? If so, it is referred to
different to the earlier multi-instance where mul- the second level support which is the local vendor
tiple installations of the software were made to implementer or the implementation partner. The
run across the company in one or more locations. second level support determines whether it is a
Typical implementations today are single instance functionality or software performance setting issue
multi-site implementations. Organizations are that requires additional configuration to make it
now implementing ES into one site which is their work and meet the business requirement. Finally,
main manufacturing or business centre and this if the local implementation partner determines
single instance is used by all other subsidiary that the problem is a software bug or a product-
sites, distribution warehouses, and sales offices. related issue it is raised to level three which is
An implementation partner is mostly used for the support channel inside the software vendor.
managing the ES project. Findings revealed that So it is a typically a three tier support model as
while a third party or a consultant implementer shown in Figure 1.
was popular in the past for large organization Customization is the process in which changes
implementations, SME customers now prefer are made to the ES software during the implemen-
the software vendor’s direct involvement. This tation phase to suit the needs of the organization
finding again confirms the Shakir (2002) study where it is being implemented. This happens when
which also noted that vendor driven implemen- the best business practices embedded in the ES
tations were on the increase. A majority of the software do not satisfy the needs of the business,
participants in the current study suggested that and the software is changed to meet the require-
there has been a shift over the last five years. ments of the organization (Davenport & Prusak,
Customers traditionally preferring to work with 1998; Kumar & Van Hillegersberg, 2000).
the big 5 consulting companies for implementation There are two implementation strategies or
are now more inclined to work with the software models. The first is the “comprehensive custom-
vendor directly so that they have a one-stop shop. ization” type when many and sometimes major
Customers are starting to realize that the techni- changes to the software are performed during
cal skills a software vendor provides may not be the implementation to satisfy business require-
possible from consultants. One vendor explained ment. The second is “vanilla” or “out-of-the-box”
what customers feel is that unless they actually when the ES software application is implemented
talk to the software owners, they may not get the without any changes to the software and the
59
Figure 1. Three tier post implementation support model
business processes within the organization are encounter difficulties, suffer on cost, schedule
changed to suit the functionality of the software. and performance metrics. The company will also
SAP explained that there is a potential source experience difficulties when attempting to upgrade
of confusion about the extent of customization to a later package release. But on the other hand
as every project needs some form of customer the organizational adaptation to the ES will be
specific reports, customer specific interfaces, and easy and the system will meet the needs of the
customer specific data conversion programs. ES business. The current study revealed that vanilla
software is designed to meet most customization implementations are more common in the SME
requirements by adjusting parameter settings. All category whereas the large organizations are more
modern software vendors now have a software likely to have comprehensive implementations.
architecture that does not require modification to Shakir and Viehland (2004) noted cost as a driver
the core software statements to achieve results. towards vanilla implementations as the two ap-
The user access can be built through parameter proaches has major implications on the change
settings to accommodate specific requirements. management strategy. When the best practice
However, this should not be confused with what is chosen, people issues become top priorities
is called true customization in which the core whereas when the implementation strategy is
software is actually modified. geared towards customization, it is more of a
Findings in this study revealed that organiza- technical challenge. Parr and Shanks (2000) have
tions now view the ES software not as a bunch of reported in their study on different ES implementa-
statements but as pre-defined business processes. tion approaches that vanilla implementations are
These organizations prefer to adhere to the pre- usually single site and comprehensive multi-site.
defined business processes in the software and However, the current study suggests that vanilla
change their own processes to the software’s implementations could be single or even multi-site
requirements. The companies doing this are more and currently more implementations are multi-site.
likely to be successful in capturing the benefits and An implementation is considered new when
controlling the cost of the implementation as this it is implemented in an organization for the first
also helps in future upgrades, and the overall cost time. An upgrade is when a revised version of
of ownership gets reduced. These findings also the software with some additional functionality
confirm the Brehm, Heinzl, and Markus (2001) is implemented to upgrade the existing software
study in which they have estimated that greater the in the current implementation (Dalal, Kamath,
customization, the more will the implementation Kolarik, & Sivaraman, 2004). Add-ons, also called
60
bolt-ons, include adding new modules to the exist- revealed that typically best-of-breed model was
ing implementation. Replace means change the adopted by many organizations in the past. For
existing implementation with a different vendor’s example, an organization might install HRM
software. The Shakir (2002) study observed that module from PeopleSoft, financials from Oracle
while new implementations were happening in and manufacturing from SAP in the first phase.
SME’s, large organizations were focusing only Then subsequently install bolt-on modules such as
on upgrades and add-ons comprising of 10-15% CRM from Microsoft, SCM from SAP or Oracle,
of the total implementations. However, findings or BI from Cognos in the second phase.
in this study suggest an equal split between new However, both customers and vendors are now
implementations vs. upgrades, add-ons, and removing towards single vendor implementation.
placements in NZ organizations. This is because the additional benefit received
In the current study, SAP suggested a 50-50 from a best-of-breed implementation is vastly
split. “We’re definitely focusing on new imple- outweighed by the cost of implementing, main-
mentations because that’s where our goal is. taining, and managing those disparate systems.
However, we have to look after our existing cus- Organizations have realized that although there are
tomer base and as their requirements change, the really good benefits in PeopleSoft, SAP, and Oracle
presentation of our software in their business may for the different modules, the cost to implement
also need change.” In the case of replacements, and maintain is enormous. While they may only
Oracle noted that an organization will replace an get 85% to 90% of the best-of-breed benefit in a
ES only if there is a need to satisfy some major single vendor implementation that is preferable
benefit which remains unsatisfied in their exist- especially since it can save three times the cost
ing system, because it is expensive to replace. It of implementation and maintenance. Vendors
is not just the cost of the software, but it is the also do not release new versions of the software
huge organizational change that the organization exactly at the same time, therefore managing the
has to go through to replace an enterprise system. upgrade path becomes difficult, the investment
Oracle also revealed that in the past this cost was depreciates faster than expected, and organizations
underestimated, but “replacement cost is three are unable to take advantage of the new features
times the cost of upgrade”. Oracle also revealed of the software. This aspect differed from the
the maintenance aspect which included the cost findings of the Shakir (2002) study which noted
of up-grading the ES. “Typically in every five- that while traditional ERP implementations still
year period, companies spend up to four times dominate new implementations, upgrades, add-
the initial purchasing implementation cost, just ons, and replacements appear to favor the best-
to maintain the ES. That is why IT budgets in of-breed model. The best-of-breed model is also a
organizations allocate substantially for upgrade consideration for new implementations, especially
support as opposed to new requirements.” for organizations that operate in niche industries.
Another model used during ES implementa- The application service provider (ASP) imple-
tion is the best-of-breed, as opposed to a single mentation model is one in which a service provider
vendor implementation. The best-of-breed model provides an ES application software as a service
includes implementation of a mix of different ven- or hosting to organizations at a fixed cost for a
dor modules which the vendor specializes in, to specific period (Malcolm, 2002; Pamatatau, 2002).
have the best of everything (James & Wolf, 2000; There was a mixed response from respondents on
Pender, 2000). A single vendor implementation this model. One vendor noted, “ASP pops up every
includes all the modules from a single vendor 5 years and was a bit like an economy that came
as an integrated package. Findings in this study and went and nothing really happened. I’m not
61
too sure it is addressing a real market requirement provided specific examples in highlighting trends.
need”. Another vendor responded “I don’t think SAP explained that traditionally, over the last 10
this model has picked up at all”. However, yet years, there have been many implementations in
another vendor confirmed that the ASP model is the consumer packaging goods, manufacturing,
used in NZ. “We use it. Customers are happy with forestry, and pulp and paper industries. However,
it. We have time and resources for providing the in the last two years there has been a slight shift
service. There’s a huge market there. These are in the ES market in NZ with several implementa-
small companies that do want an ES and they don’t tions in the retail and utilities industry, and this
mind paying sixty to seventy thousand dollars a trend is likely to continue for the next two years.
year but are not able to spend half a million to one ES maturity in an organization depends upon
million dollars for buying the software. It is not the number of year’s experience the organization
too difficult for these companies to put up a few has had with ES and the stage of ES implemen-
servers each with the latest operating system of tation (Hawking, Stein, & Foster, 2004). This
windows. We’ve got the people and it’s not much concept of ES maturity and the different stage of
of their time, so we can provide this service. There ES implementation is reinforced by the Nolan and
is no trend, but there is a huge market out there if Norton Institute (2000) classification that groups
marketed properly.” Except for this one vendor, implementations into levels of maturity such as
the overall response was not very positive for the beginning where ES is implemented in the past 12
use of the ASP implementation model, either in months, consolidating where ES is implemented
the current ES environment or in the recent past between 1 and 3 years, and mature where ES is
(Shakir, 2002). implemented for more than 3 years.
Another model referred to as the business pro- Findings revealed that most NZ organiza-
cess outsourcing (BPO) or the managed service tions are reasonably mature with ES technology
model was cited by respondents as a growing and IT in general. Most large organizations and
implementation model in a NZ context. In this many SMEs in NZ have been using some form
model outsourcers run a customized managed of ES technology for more than a decade and are
service of ES implementation for customers where at a fairly advanced level of maturity. This also
effectively a single solution is sold to a customer. confirms the Shakir (2002) study which noted
One consultant explained that this is a low cost that although NZ is a small country, technology
commodity solution where the customer prefers is mature and on par with what’s happening in
not to manage the ES. This model positions itself the U.S.
very much in the SME market for example, in out- However, as per the respondents, there are a
sourcing transaction services or specific functions couple of issues in managing ES projects which
like finance, or payroll, or a similar function to a do highlight the slower pace of ES maturity within
third party supplier. “An organization may have the NZ industry. First, many NZ organizations do
implemented Oracle financials or SAP finance for not conduct a proper business justification of their
example, but may be paying, say to IBM, to run implementation. Although some improvement has
the technology and specific functions for them”. been seen in the last couple of years, most NZ
Despite its risks, ES implementation is perva- organizations produce little or no value assess-
sive in many different types of industries (Kumar ments that often lead to weak business cases and
& Van Hillegersberg, 2000; Mabert, Soni, & Ven- insufficient benefit models which cannot be used
kataraman, 2000). A majority of the respondents for benefit tracking. Plant and Willcocks (2006)
noted that ES implementations are covered in most in their study on critical success factors for ES
industry sectors in NZ; however, some respondents implementations have also found an increased
62
emphasis upon the determination of clear goals now started realizing the value of technology and
and objectives at the project outset as one of the its use to stay ahead of competition.
important factors for ES implementation suc- Findings revealed that the mix between national
cess. Second, many organizations in NZ believe and international ES implementations is a 50-50
implementation of ES is a technology challenge. split in NZ. Respondents noted that on several oc-
However, according to most respondents, it is more casions the implementations started as a national
about change management, people, and processes implementation but quite quickly reached out to
and less about technology. With better business countries like Fiji, Australia, Europe, Singapore
case development, these are two areas in which or wherever the sales and distribution offices
many NZ companies are struggling. are located. Although many NZ companies are
Respondents also revealed that typically when a based in NZ and the reach is national, there is a
new system is implemented, productivity drops for growing trend in NZ organizations to expand to
a period and then goes up again. Oracle suggested global markets therefore now the reach is becom-
the depth of the drop depends upon how well the ing global. This is also supported by the growing
system is implemented, how well the change is export-oriented market of NZ organizations.
managed, how well the business case is defined, Global implementations are also part of multi-
and how well the managers are measuring and national organizations that implement ES within
managing benefits before the organization starts their NZ companies. These implementations are
seeing the benefits starting to flow through. normally “roll-outs” based on a global template
Until a few years ago the majority of orga- that includes standard business processes. The
nizations did not use the ES in its true capacity. “roll-out”, as explained by one consultant, is
ES was used as a financial system, as a central an implementation generated from a template
repository for HRM records, or as a method for customized for an overseas location. The roll-
raising purchase orders. This was because the out starts with a massive data set prepared by
organizations had not thought about what they the first implementation followed by the addi-
were trying to optimize, what benefits they were tion of country specific and localized data. For
trying to bring into the organization, what they example, GST or VAT percentages are different,
were trying to change, how they were trying to the states as part of the addresses are different,
manage the business, and whether they could and therefore a couple of master files which are
actually get the information to manage the busi- country specific are implemented on top of the
ness. However, recently the software vendors have local customer and vendor base that is created. The
started to see several companies trying to find data-set roll-out is established using country spe-
ways to get more value out of their investment. cific data where new country settings override the
Companies have recently started asking how to template settings. A separate dedicated warehouse
establish analytical processes to optimize and for these locations is also included for tracking
realize business value from their ES investment. transactions. However, many NZ companies are
Many NZ organizations have already completed governed by their parent organizations; hence all
their first phase of ES needs and are now extend- the decision making for the ES implementation
ing into the second phase with CRM, SCM, or BI. is done offshore by the parent company, without
Most respondents agreed that the slower pace of much control from the NZ businesses. This is
ES maturity within the NZ organizations is due to nothing new. Implementations based on global
the limited spending power, which is attributable templates and critical decision-making being
to the comparatively small NZ economy. However, made offshore were observed in Shakir’s 2002
this trend is now changing. NZ organizations have study. In summary, ES implementations in NZ
63
organizations are moving from national towards and SME organization segments. Typical imple-
a global reach either by expansion into overseas mentations in the large organization segment
markets or off-shore ownership. with revenues more than $250M are currently in
phase 2 and the organizations are fairly mature
with their ES. These organizations are likely to be
conclusIon And in the phase of acquiring collaborative scenarios
FurtHer reseArcH like SCM or CRM, or management services
applications such as BI. These can be single or
The main objective of this study was to understand multi-site implementations. The number of users
typical ES implementations and practices in NZ is estimated to be 200 or above and the cost of
and how current implementations in the SME sec- the project is likely to be more than $2M. In the
tor differ from the earlier implementations in the SME segment, typical implementations are in
large enterprise sector. The findings are analyzed organizations with revenue between $50 - 200M.
and summarized in table 4, based on different These implementations are likely to be new with
organizational size segments of the ES market. two or more core ES modules. These can be single
Table 4 suggests the different ES implementation or multi-site implementations with number of
determinants for both large and SME organiza- users in the range of 20 – 200 and the cost of the
tions and explains the relationships between the implementation between $100,000 – 1M.
organization size and the implementation process Many ES implementations in New Zealand
variables. are several years old now however, these com-
It is evident from the table that typical cases panies have only recently started asking how to
of ES implementation in NZ exist in both large actually optimize processes and realize business
Table 4. Determinants for describing typical ES implementation based on organizational size segment
Organizational Organization Size Large SME

Characteristics Revenue in Million ($NZ) 250M and Over Small 10-50 M
SME 50-250 M
ES implementation Phases of ES implementation Phase 2 Phase 1
process variables
Modules Supplementary modules Core modules
(HR, SCM, CRM, data ware- (Finance, manufacturing,
housing and BI) distribution)
Time for implementation 12 to 24 months 3 to 12 months
Locations Single or multi-site Multi-site
Cost of implementation Above $NZ 2 M $NZ 100,000 – $NZ 1 M
$NZ 1 M – $NZ 2 M
Number of users Above 200 Below 20
20 – 200
ES implementation Implementation partners Third party Vendor
models - Vendor vs. Third party
Customization Comprehensive Vanilla
- Vanilla vs. Comprehensive
Implementation Upgrades/add-ons/replace New
- New vs. Upgrades/add-ons/replace
Implementation Best-of-breed Single vendor
- Single vendor vs. Best-of-breed
64
value from their ES investments. Organizations Dalal, N. P., Kamath, M., Kolarik, W. J., &
are establishing analytical processes for tracking Sivaraman, E. (2004). Toward an integrated
benefits continuously improving in taking advan- approach for modeling Enterprise processes.
tage of the technology. Communications of the ACM, 47, 83–87.
The findings of this study are limited to the doi:10.1145/971617.971620
views of professionals from different ES vendors,
Davenport, T. H. (2000). Transforming the Practice
ES consultants, and IT research organizations and
of Management with Enterprise Systems. In Mis-
are limited by a small sample size. There may also
sion Critical (p. 203-235). Boston, MA: Harvard
have been some influence on the responses by the
Business School Press.
commercial interests of the firm the participant
worked for. However, the study has achieved its Davenport, T. H., & Harris, J. G. (2005). Au-
objectives. This is achieved due to the seniority tomated Decision Making Comes of Age. MIT
and experience of the respondents within the or- Sloan Management Review, Summer 2005 46(4),
ganizations interviewed, and the position of these 83-89.
organizations as key players in the ES industry
Davenport, T. H., Harris, J. G., & Cantrell, S.
in New Zealand. Further research is in progress
(2002). The Return of Enterprise Systems: The
to analyze the current practices and the critical
Director’s Cut. Accenture Institute for Strategic
effectiveness constructs of ES in New Zealand
Change.
from the practitioners’ perspectives identified
by this study. Davenport, T. H., & Prusak, L. (1998). Working
Knowledge. Boston, Harvard Business School
Press.
reFerences
Hawking, P., Stein, A., & Foster, S. (2004). Re-
Allen, D., Kern, T., & Havenhand, M. (2002). visiting ERP systems: Benefit Realisation. Paper
ERP Critical Success Factors: An Exploration of presented at the Proceedings of the 37th Hawaii
the Contextual Factors in Public Sector Institu- International Conference on System Sciences,
tions. Paper presented at the Proceedings of the Hawaii.
35th Annual Hawaii International Conference on Hedman, J., & Borell, A. (2002). The impact of
System Sciences, 9/02, Hawaii. Enterprise Resource Planning Systems on Organi-
Bancroft, N. H., Sep, H., & Sprengel, A. (1998). zational Effectiveness: An Artifact Evaluation. In
Implementing SAP R/3 (2nd Edition ed.). Green- F. F.-H. Nah (Ed.), Enterprise Resource Planning
wich, USA: Manning Publications. Solutions & Management (p. 125-142). Hershey,
London: IRM Press.
Brehm, L., Heinzl, A., & Markus, M. L. (2001).
Tailoring ERP systems: A spectrum of choices Holland, C., & Light, B. (1999). A critical success
and their implications. Paper presented at the factors model for ERP implementation. IEEE Soft-
34th Hawaii International Conference on System ware, (May/June): 30–36. doi:10.1109/52.765784
Sciences, Hawaii. James, D., & Wolf, M. L. (2000). A Second Wind
for ERP. McKinsey Quarterly, Issue 2, 100-107.
Kumar, K., & Van Hillegersberg, J. (2000). ERP
Experiences and Evolution. Communications of the
ACM, 43(4), 23–26. doi:10.1145/332051.332063
65
Mabert, A. M., Soni, A., & Venkataraman, M. A. Scott, J. E., & Vessey, I. (2002). Managing Risks
(2000). Enterprise Resource Planning Survey of in Enterprise Systems Implementations. Commu-
US Manufacturing Firms. Production and Inven- nications of ACM, April 2002, 45(4).
tory Management Journal, 41(2), 52–58.
Shakir, M. (2002). Current Issues of ERP Imple-
Malcolm, A. (2002). Fonterra Rents its Accounting mentations in New Zealand. [Massey University,
Application. Computerworld IDG Communica- Auckland, New Zealand.]. Research Letters in
tion Ltd., 11 July, 2002, Web page: http://www. Information and Mathematical Science, 4(1),
idg.net.nz/webhome.nsf/UNID/8433B6BCB6B 151–172.
E15FECC256BF1007BF560
Shakir, M., & Viehland, D. (2004). Business Driv-
Markus, M., & Tanis, C. (2000). The Enterprise ers in Contemporary Enterprise System Imple-
Systems Experience - From Adoption to Success. mentations. Proceedings of the Tenth Americas
In R. W. Zmud (Ed.), In Framing the Domains of IT Conference on Information Systems, New York,
Research Glimpsing the Future Through the Past 103-112.
(p. 173-207). Cincinnati: Pinnaflex Educational
Shanks, G., Seddon, P. B., & Wilcocks, L. P.
Resources, Cincinnati, USA.
(2003). Second-Wave Enterprise Resource Plan-
Nolan and Norton Institute. (2000). SAP Bench- ning Systems: Implementing for Effectiveness.
marking Report 2000. Melbourne. Cambridge University Press.
Pamatatau, R. (2002, June). The Warehouse Skok, W., & Legge, M. (2001). Evaluating En-
outsources Oracle Management. NZ Infotech terprise Resource Planning (ERP) Systems Using
Weekly, 24, 3. an Interpretive Approach. Paper presented at the
Proceedings of The 2001 ACM SIGCPR Confer-
Parr, A., & Shanks, G. (2000). A Model of ERP
ence on Computer Personnel Research, April, p.
Project Management. Journal of Information Tech-
189-197.
nology, 15(4). doi:10.1080/02683960010009051
Sumner, M. (1999). Critical Success Factors in
Pender, L. (2000). Damned If You Do: Will Inte-
Enterprisewide Information Management Systems
gration Tools Patch the Holes Left By An Unsat-
Projects. Paper presented at the 5th America’s
isfactory ERP Implementation? CIO Magazine,
Conference on Information Systems, Milwaukee,
September 15, 2000. Retrieved from http://www.
Wisconsin, USA.
cio.com/archive/091500_erp.html
Viehland, D., & Shakir, M. (2005). Making Sense
Plant, R., & Willcocks, L. (2006). Critical Success
of Enterprise Systems Implementation. Business
Factors in International ERP Implementations:
Review . University of Auckland, 7(2), 28–36.
A Case Research Approach. Working Paper Se-
ries - 145, London: Department of Information Yang, S., & Seddon, P. B. (2004). Benefits and
Systems, London School of Economics and Key Success Factors from Enterprise Systems
Political Science. Implementations: Lessons from Sapphire 2003.
Paper presented at the 35th Australasian Confer-
Sarker, S., & Lee, A. S. (2000, 13 November 2000).
ence in Information Systems. Hobart, Australia.
Using a case study to test the role of three key social
enabales in ERP implementation. Paper presented Zrimsek, B. (2002). ERPII: The Boxed Set. Re-
at the ICIS 2000, 13 November 2000 http://www. trieved Mar. 4, 2002, from www3.gartner.com/
commerce.uq.edu.au/icis/ICIS2000.html pages/story.php.id.2376.s.8.jsp
66
key terms Customization: Altering a system’s software

code to include functionality specifically wanted
Application Service Provider (ASP): An by an organization, although not originally in-
ASP is a business that provides computer-based cluded in the package itself.
services of specialized software to customers Extended ERP: Extends the foundation ERP
over a network. system’s functionalities such as finances, distri-
Business Intelligence (BI): Software tools that bution, manufacturing, human resources, and
use the ERP database to generate customizable payroll to customer relationship management,
reports and provide meaningful information and supply chain management, sales-force automa-
analysis to employees, customers, suppliers, and tion, and Internet-enabled integrated e-commerce
partners for more effective decision making at and e-business
the organization. Small and Medium-Size Enterprise (SME):
Business Process Outsourcing (BPO): A business enterprise independently owned by
Contracting of specific business task (s), such contributing most of the operating capital and
as payroll, marketing, billing etc, to a third party managed by the owners or managers, having
service provider as a cost saving measure fewer than 250 employees and a small-to-medium
Enterprise Resource Planning (ERP): market share. This number differs in different
Software systems for business management that regions or countries (in some countries it is less
integrates functional areas such as planning, than 500 while in others the number may be less
manufacturing, sales, marketing, distribution, than 100). The number may also vary depending
accounting, finances, human resource manage- on the type of business.
ment, project management, inventory manage- Supply Chain Management (SCM): Soft-
ment, service and maintenance, transportation, ware systems for procurement of materials,
and e-business. transformation of the materials into products, and
Customer Relationship Management distribution of products to customers, allowing the
(CRM): Software systems that help companies enterprise to anticipate demand and deliver the
to acquire knowledge about customers and deploy right product to the right place at the right time at
strategic information systems to optimize revenue, the lowest possible cost to satisfy its customers.
profitability and customer satisfaction.
Critical Success Factors (CSF): Are the
factors which are critical for an organisation or
a project and which must go right to achieve the
defined mission.
This work was previously published in Handbook of Research on Enterprise Systems, edited by Jatinder N. D. Gupta, Sushil
Sharma and Mohammad A. Rashid, pp. 170-184, copyright 2009 by Information Science Reference (an imprint of IGI Global).
67
68
Chapter 1.5
A Conceptual Framework for
Developing and Evaluating ERP
Implementation Strategies in
Multinational Organizations
Kai Kelzenberg
RWTH Aachen University, Germany
Thomas Wagner
Kai Reimers
AbstrAct IntroductIon
The chapter develops generic strategies for the Market demands are becoming more and more
specification and implementation of an enterprise dynamic, forcing organizations to be flexible
resource planning (ERP) system in a multinational in order to satisfy the needs of their customers
company. After the presentation of a framework (Mabert, Soni, & Venkataramanan, 2001). At the
for categorizing companies by their global busi- same time, organizations face an ever increasing
ness orientation, ERP strategies corresponding to competition through globalization. As a result
each category are derived. Subsequently, various of both phenomena, business organizations tend
implementation strategies are developed for each to act in networks of tightly or loosely coupled
type of ERP strategy; they provide decision mak- productive units. Bartlett and Ghoshal (1998)
ers with a high degree of freedom in specifying identify four different business orientations which
an implementation strategy in accordance with a can be used to describe the structures of multi-
company’s strategic goals. The results are sum- national companies (MNCs). Starting from these
marized in a phase model; the overall approach four business orientations, this chapter presents a
is illustrated by two polar cases. conceptual framework for deriving enterprise re-
source planning (ERP) implementation strategies
DOI: 10.4018/978-1-59904-531-3.ch003
A Conceptual Framework for Developing and Evaluating ERP Implementation Strategies
in multinational organizations. This is motivated bAckGround

by the findings of prior research that shows the
importance of aligning the IT strategy with a firm’s The implementation of an enterprise resource
business strategy (Ward, Griffiths, & Whithmore, planning system in a company can have different
1990; Earl, 1993). degrees of complexity which will be conditioned
However, organizational contingencies are by the following items (this list is limited by the
seldom considered in the literature on ERP imple- scope of the chapter; several more perspectives
mentations which focuses on critical success fac- could be added in future work):
tors (CSFs) (Holland & Light, 1999; Akkermanns
& van Helden, 2003) or critical issues and risk 1. ERP definition
factors in general (Bingi, Sharma, & Godla, 1999; 2. ERP strategy
Sumner, 2000; Scott, 1999; Hong & Kim, 2002; 3. Implementation strategy
Gosh & Gosh, 2003). The framework of Bartlett
and Ghoshal has previously been applied in the Referring to O’Leary (2000), ERP systems
IS field. Reimers (1997) shows how IT can be “provide firms with transaction processing mod-
managed in the transnational organization, and els that are integrated with other activities of the
Madapusi and D´Souza (2005) have used the firm” (p. 7). Moreover, they can reduce informa-
framework to develop recommendations regard- tion asymmetries and help to create one view on
ing the way ERP systems should be configured in all relevant data which can be shared across the
multinational companies. While these authors also whole organization. This concept is based on a
discuss the issue of appropriate implementation single database that contains all data of several
strategies, this discussion focuses on the issue of a functional and/or local areas. Bancroft, Sprengel,
‘big bang’ vs. a phased implementation approach, and Seip (1996) offer a similar definition of ERP
which we deem too narrow. Rather, we propose systems focusing on SAP/R3. For them, an ERP
that the configuration of an ERP system should fol- system consists of “one database for the entire
low an appropriate ERP implementation strategy corporation without any data redundancy and
which comprises many more issues than that of a with a clear definition of each [data] field” (p.
big bang vs. a phased implementation approach. 17). Firestone (2002) adds another perspective on
In this chapter, we offer a framework which helps ERP as he mentions that customers want ERP for
to conceptually organize the issues that should be decision-making support, although there are other
considered in deriving an ERP implementation (software) systems that are more specialized in this
strategy for multinational companies and which area. Markus and Tanis (2000) add the opinion of
also helps to fine-tune the implementation strategy some ERP vendors who state that their software
as the implementation process unfolds. “met all the information-processing needs of the
The remainder of this chapter is organized companies that adopted them” (p. 174). This in-
as follows: First we will give a short review of cludes an automatic data transfer facility between
different views on ERP before we derive ERP several functions within the system as well as a
strategies from an organization’s business ori- shared database for all applications.
entation. Subsequently we will discuss different These conditions could be satisfied fairly eas-
ERP implementation strategies. Afterwards, our ily if one was dealing with a single-site company,
framework is presented and illustrated by two but the more interesting question is what happens
examples. The chapter ends with a discussion. if a company is composed of different sites with
different ranges of functionality, for example, a
69
large producer with several distribution centers a general ERP strategy they will provide important
around the world or a federation of several pro- insights regarding the questions of when one could
ducing and distributing companies that form one speak of a global ERP strategy and how important
major company. Is it possible, referring to the it is to have one organizational entity for the whole
above mentioned authors, to call the installed company that can develop and promote the ERP
system across all members of such a unit one big strategy for all sites. In this chapter, we define an
ERP system? ERP strategy as containing the range of system
Following such a scenario, what would have to components that have to be installed in each
be done to standardize the data of each business site, as well as the interfaces and data formats in
unit in such an organization and which benefits which data transfer should be done. The chapter
would result from this effort? How far should will go on to show different generic ways how
standardization go and which areas could be to realize an ERP system according to the global
standardized in a useful manner? Moreover, who ERP strategy (ERP implementation strategy). In
should be in charge of choosing and designing an this context, the problems of tailoring an ERP
ERP system fitting to a company’s requirements system fitting the company’s requirements as
and of keeping the system up and running? Which described by Lehm, Heinzl, and Markus (2001)
organizational units have to be established to lead will be outlined. Since not all strategies require
this venture to success? different implementation strategies, some can be
Many of these questions could not be answered discussed jointly.
in a general way but they should help to consider This chapter focuses mainly on the first
the different aspects of the design process of such phase(s) of an ERP life-cycle (cf. Esteves &
a project. The presented framework tries to support Pastor, 1999, or Markus, Axline, & Tanis, 2000).
the decision makers with a structured approach Given its academic and practical attention, the
facing this task. implementation phase seems to be a crucial one
If the ERP system has been designed in ac- that motivates our focus. The consideration of all
cordance with the company’s requirements, an phases is beyond the scope of this chapter; for a
IT strategy focusing on implementation has to be comprehensive literature review of all phases, we
developed. Karimi and Konsynski (1991) state refer to Esteves and Pastor (2001).
that an IT strategy has to be derived from the
business strategy. Similarly, an ERP strategy has
to be developed which shows the long-term goals busIness orIentAtIon
to be reached by the use of an ERP system. Clem-
mons and Simon (2001) state that a misalignment In 1998, Bartlett and Ghoshal published a book
between business and ERP strategy often causes in which they describe a couple of companies and
ERP implementation delays and failures. (As will how their way of doing business differs. Based
be shown later, there is a need for a centralized on an analysis of these differences, the authors
authority which is in charge of deciding whether conclude why one firm is more successful with
a global ERP strategy is to be developed, and, if its strategy than another. They develop a frame-
so, how to enforce it within the whole company.) work consisting of four types of multinational
In the subsequent sections the work of Bartlett companies. Each type is characterized by a dis-
and Ghoshal (1998) helps to identify four differ- tinct strategy and business orientation that deals
ent types of business orientations for companies. with the allocation and interconnection of the
These are generic views which—in reality—will company’s resources. The term “resource” is not
be mixed in a number of ways, but for developing described in detail but refers to skills, personnel,
70
money, knowledge, and so on. Applying these system, and the interconnection of all functional
insights to the topic of this chapter, ERP systems areas. In this context it is negligible how many
should optimize the use of such resources. The functions one site will implement. That means,
different business orientations will be explained one site could just implement one module for sales
in the subsequent sections and provide the basis and distribution, while another needs production
for deriving appropriate global ERP strategies and purchasing too. The integration of the func-
which are discussed in the same sections. The tions does not differ in both cases. On the other
designer of such a system must take into account hand, HQ has to consider all local requirements
that there might be not only positive influences regarding the single sites such as local laws, lo-
on the company’s learning capabilities, but that cal accounting rules, and possibly environmental
side effects could cause changes in organizational stipulations (Krumbholz, Galliers, Coulianos,
values and norms (Butler & Pyke, 2003). Thus, & Maiden, 2000). As a consequence, the ‘how’
ERP implementation goes beyond an IT project: seems to be less complex than the ‘what’. Argu-
it needs the support of top management to enable ably, the planning process for such a system takes
the organizational rebuilding processes. longer than in other company constellations while
the implementation process could be shortened.
Global orientation From a strategic point of view, the MNC with a
global orientation could easily add new plants to
This business orientation is composed of a strong its network because it has a predefined and tested
center that makes all important decisions for the system that just has to be implemented. Besides,
whole company. The decisions made will be the MNC could realize learning curve effects
communicated to the single sites and have to be from previous implementations which should
implemented without any adjustment to local ensure the success of an implementation process
requirements. Through this process the whole at a single site.
company looks—from the customers’ point of Thus, the ERP strategy implies that all sites,
view—alike all over the globe. One implica- including HQ, get one system. In accordance
tion of this orientation is that the whole world is with this ERP strategy, a single implementation
seen as one market which will be supplied with strategy can be developed including estimated
identical products. To ensure the loyalty of each project duration, costs, and scope for the single
site, the sites’ general managers are handpicked sites (the term scope is used to mean the range of
by headquarters (HQ) to control all activities modules that should be implemented in one site).
and to implement the envisioned processes. New HQ determines the design of the system and the
processes have to be developed just once and can rollout sequence for all sites, and defines the pilot
be implemented worldwide in a short time accord- site where the system should be introduced and
ingly. The global orientation is a hub-and-spoke tested for the first time.
kind of network with strong ties of the spokes to
the hub, which is represented by HQ. International orientation
In this setting, HQ will make the decision
to implement an ERP system and what kind of In this orientation HQ partly delegates control to
software will be used. Since the idea is that all the individual sites. Local adjustments are possible
processes should be equal in all sites, the system but the main decisions regarding product policies
has to be developed just once. It can thus easily and strategies are made by HQ. Unlike the global
meet all the above mentioned requirements of orientation, the sites have the opportunity to reject
an ERP system such as one shared database, one HQ’s decisions which strengthens their position.
71
As a consequence, core competencies remain IT landscape, implying that the above definition
in the HQ while individual sites could develop of an overall ERP strategy is not applicable in
important business competencies too. such a context. There will be no shared database,
The ERP strategy for this business orientation neither guidelines for one system nor for the range
is similar to that of the global orientation. The of modules that should be established. Regarding
major activities are initiated and coordinated by the whole company, there is no global resource
HQ. The whole company gets one system that—in planning at all.
contrast to the global orientation—is open to ad- Another case occurs if HQ decides to imple-
justments to local requirements. For this purpose ment several modules in order to manage its
the guidelines of the whole implementation have business and satisfy its data needs. Following
to be softened. The core system is developed by that idea, interfaces, including data formats and
HQ but can be reconfigured by local sites. The forms, between the individual sites and HQ have
system has one shared database which helps to be defined. However, this scenario will still not
to collect data centrally and supports decision- meet the requirements of the definition of a global
making processes at the highest management level. ERP strategy because it focuses on HQ only and
Rollout sequence must be harmonized with regard not on the entire organization.
to the interests of the individual sites, which also A third scenario emerges when HQ can de-
includes determination of a pilot site. Due to the velop an ERP template and also has the power to
possibility of local adjustments, the complexity of implement it in the several sites. This scenario
the implementation at a single site as well as of the causes other problems regarding Bartlett and
overall process will increase. Therefore, neither Ghoshal’s definition of a multinational orientation.
the ‘what’ nor the ‘how’ seem to be that simple. Moreover, there will be no need for one shared
If we follow the idea of Huber, Alt, and Österle database or integration of all functional areas
(2000), the use of templates could simplify the because every site manages its own operations.
implementation process and allow the creation of The production line, customers, and suppliers
a standardized way to adjust the system locally. differ from each other. A single database has no
additional benefit for the organization. The only
multinational orientation part of an ERP system that could unite the several
sites is finance, which has to be done in some way
In this orientation the single subsidiaries are by HQ as it acts as a kind of portfolio manager.
loosely coupled to the center. Following the ex- Apparently there will be no fit between a mul-
planations of Bartlett and Ghoshal (1998), every tinational orientation and a global ERP strategy
site is doing business under its own responsibility. if there are no exceptions allowed regarding our
The main strategic focus is differentiation and definition. At that point, the decision makers in
adjustment to local requirements. Because of this kind of MNC have to think about the need for
the absent link between several company sites, an ERP system. Who will benefit from a shared
each site creates its own data and knowledge. database? The answer to that question is fairly
Regarding the definition of a global ERP strategy easy: no one, because each site acts indepen-
given above, the question whether a company dently and reacts to different countries’ markets
has a global ERP strategy or not will have to be and customers, and deals with different products.
assessed based on the power and position of HQ. Possibly, there is no need for some subsidiaries
One possibility is that HQ has no influence over to implement a fully integrated software package
the individual sites. This can mean no influence because of their small size. A globally defined
over any guidelines at all, especially regarding the ERP system will not meet the local requirements
72
if there are no major adjustments to the system. Sites that have few local resources or capabilities
Thus, the ERP strategy could consist of just a (competencies) but operate in an important local
minimum—that is, the data connection between market (environment) are labeled black holes. In
the individual sites and HQ in the second scenario contrast, a site that has strong competencies but
described above. Another ERP strategy could be operates in a rather unimportant market is called
that HQ is in charge of supporting implementation contributor (Bartlett & Ghoshal, 1998).
of ERP systems in each site that should connect Because of the specific structure of a trans-
with HQ. Sites will retain their independence national organization, it is unclear who can start
because only the use of any system is dictated, but the initiative for an ERP strategy. People work
neither the system vendor nor the way it will be together on a voluntary basis and the subsidiar-
implemented is centrally prescribed. According to ies can implement and use business practices as
Markus, Tanis, & van Fenema (2000), this strategy they want. Regarding the roles described above,
could lead to a complete disaster for individual it may be argued that the strategic leader has the
sites because there are no learning curve effects, competencies to develop an ERP strategy and
while the possibility of repetition of errors that can might convince other firms in the organization
be made within the ERP implementation is real. to take part in this venture. As soon as some sites
The implementation strategy can be simple implement a jointly developed system, a kind of
because only data formats and forms have to be domino effect could start so that ultimately every
defined which are used for data exchange with HQ. site in the network will join in the ERP project.
Obviously, sites labeled black holes should have
transnational orientation a major interest in advancing their abilities. A
common and integrated database across all sub-
The transnational orientation describes an in- sidiaries—that is the accumulated knowledge of
tegrated network of business units. No site nor individual sites stored in a shared database—will
center has overall control of decisions and strate- be a source for weaker companies to learn in an
gies. Moreover, each site’s general manager has easy and inexpensive way from the experience of
the opportunity to cooperate on each level or in other network participants.
each functional area with other sites. The initiative Thus, the ERP strategy includes the freedom
for such actions can come from the needs of the of module choice and implementation approach
firm at that moment. Bartlett and Ghoshal (1998) in the single sites of the network. The initiative
write about cooperation in research areas and in could be started by a strategic leader with core
developing common business processes in order to competencies in one area, while other design
illustrate the principle of the transnational orienta- processes could be delegated to other sites which
tion. Moreover, they identify different roles which have competencies in a special area. However,
a single site can play. A role depends on the site’s all activities have to be coordinated within this
position along two dimensions: (1) how important network which is a difficult task because there is,
is the site’s national or local environment to the by definition, no HQ or center within this business
firm’s strategy, and (2) what competencies does orientation. The strategy must contain the pos-
the site have? If both dimensions indicate high sibility to use individual parts of the ERP system
levels, then the site will act as a strategic leader in any way and in any combination at each site.
in the network of business units. Otherwise, for
example when a site scores low on each dimen-
sion, the site will be a so-called implementer.
73
erP ImPlementAtIon On the other hand, data collection must be done

strAteGIes just once. After a pilot implementation and the
elimination of teething troubles, the implementa-
An ERP implementation strategy should consist tion process will be quickly finished. The idea of
of three major parts: implementation orientation, the presented global rollout is closely connected
scope of standardization, and implementation to centralization within the whole IT landscape.
procedure. The global rollout orientation seems to be most
appropriate in an MNC with a global orientation,
Implementation orientation strong top management support which formulates
clear goals and which supports sustained change
Referring to its business orientation, an organiza- management (Brown & Vessey, 2000).
tion may take one of the following three imple-
mentation orientations to plan the implementation Local Implementation
of a global ERP system in its several sites: global
rollout, local implementation, or mixed-mode Local implementation is the direct opposite to
implementation. global rollout orientation. Every site is enabled
to customize any kind of ERP system to the
Global Rollout company’s demands. Therefore, the idea of de-
centralization is realized within this approach. It
Global rollout means the creation of a kind of is assumed that some of the restrictions mentioned
global template which will be implemented in with regard to the global rollout have to be observed
each site without any customization. This tem- by local implementations as well because a decen-
plate could also be the vendor’s off-the-shelf tralized approach, as noted in the descriptions of
solution. The main attribute is the unchangeable the multinational orientation, must have a shared
configuration of the system. To ensure the suc- base which affords management of the portfolio
cess of this approach, it is necessary to collect of sites and defines the existing organization. But
the essential data on unique business processes in the common basis could be very small and may
the organization. This information is the basis for just concern specific interfaces or number calcula-
the general adjustment of the kernel of the ERP tion metrics (Light, 1999). The responsibility for
system, which means a single customization of the the right choice, configuration, administration,
system once before the first global rollout starts. and maintenance remains in each site. If there is
One might argue that this single customization is any grouping of site activities, the appropriate
not necessary if an organization decides to change initiative comes from the sites themselves. This
its processes according to pre-built processes in implementation orientation is most appropriate in
the chosen ERP system. However, we assume that, MNCs with a multinational orientation; neither
in reality, no organization will follow this kind of of the three critical success factors mentioned
implementation approach, due to a wide range above (top management support, clear goals, and
of problems that can occur in such an approach sustained change management) is required on a
(Light, 2005). We propose that organizations tend global scale.
to customize their system in order to increase the Both implementation orientations represent
business performance (Holland, Light, & Gib- the extreme points on a continuum, and in re-
son, 1998). In accordance with the requirements ality there will be different kinds of mixtures.
mentioned above, all sites will be assimilated and The combination of these two implementation
will lose any kind of procedural differentiation. orientations might be an adequate approach for
74
transnational organizations or international busi- is done should the standardization process start.
ness orientations. Otherwise existing business advantages could
disappear because of a misalignment between stan-
Mixed-Mode Implementation dardized processes and enterprise requirements.
Standardization could aim at the system/processes,
Mixed-mode implementation dilutes the global the data/data formats, and/or the organizational
rollout and local implementation orientations. structure of sites. The degree of standardization
On the one hand, data must be harmonized to depends on the combination of the three items
generate global economies of scale. On the other mentioned and their characteristics.
hand, there will be a high degree of freedom to If the ERP software is specified, this also
choose an ERP system according to local demands includes business processes according to the
which fits the individual site. These two aspects vendor’s experiences with other ERP implemen-
result most likely in standardization of interfaces tations. Using the same system thus means to
among sites, while hosting and administration introduce similar processes which could be cus-
of the systems remains the responsibility of the tomized by the purchaser. A major advantage of
individual sites. Assuming the transnational this approach lies in the pre-specified interfaces
organization operates in a single industry, there which interconnect the individual system modules.
might be similar processes and data which could Data exchange should thus be easily established.
be joined within a kind of template. By generat- On the other side, there might not be an ideal
ing best-practice solutions, the single sites could alignment between system and site in the case
upgrade their position within the organization and of heterogeneous business processes among the
their market. The template has to grow and evolve different sites (Mabert et al., 2001). The tailoring
over the whole process and so will the degree of of an ERP system and its consequences for the
complexity. This includes the monitoring of new implementation should be considered within the
processes added to the template which might have whole project plan—that is, right at the outset of
side effects or offers new opportunities to already the system design process. There are standard-
finished ERP implementations. This could mean ized process manuals developed by IEEE that
the need to conduct a further project to restructure can be used for this process (Fitzgerald, Russo,
the site or its ERP system to participate in new & O’Kane, 2003).
system settings. Thereby, boundaries of a single Data and data format standardization imply that
ERP blur, implying that the allocation of costs there will be interface specifications that enable
regarding a single ERP implementation becomes data flow between different systems. Besides, the
more difficult for finance and cost accounting. prescribed data formats guarantee an integrated
The mixed-mode orientation would also fit view on all sites on a common basis. As mentioned
a holding company that has to cope with a net- before, that might be limited to data in finance
work of different sites or groups of sites which and accounting departments.
constitute one or more of the above mentioned If the system/processes and/or the data/data
business directions. formats are standardized, some sites in an orga-
nization have to change their processes according
scope of standardization to these specifications. Therefore, departments,
mode of operations, and hierarchical power may
Davenport (1998) advised that an organization have to be changed too. This has to be taken into
must analyze its business to find out about its consideration regarding the expected implemen-
unique business processes. Not until this step tation duration.
75
Which type of standardization is most ap- A phased implementation can be site-by-site,

propriate depends on the overall implementation department-by-department, or module-by-mod-
orientation; for example, a global rollout imple- ule. Site-by-site implementation is similar to the
mentation orientation will aim at standardization latter case of the big bang approach but will be
of site organizational structures while a local discussed here. All these cases enable an organiza-
implementation orientation will focus on data tion to test single configurations of the ERP system
standards, preferably in the financial area. before implementing them. Meanwhile, the old
Finally, deriving an ERP strategy from the legacy system could handle market demands in
business strategy must consider costs that arise its usual way. On the other hand, this procedure
during the entire process. There will be a trade- extends the duration of the whole implementation
off between the degree of system standardization process which may increase costs and slow down
which additionally includes costs for administra- the speed with which increased efficiency can be
tion, maintenance, and (user) support and the realized. The phased implementation approach
created benefits of using it. For example, Light offers the possibility to arrange a sequence for
(1999) mentioned that a shared database—which the ERP implementations which can be adjusted
is stated as a major advantage of an ERP—does according to project experiences, urgency, and
not need to be installed to meet the widespread strategic importance of the sites. It must be pointed
market demands of an MNC. It therefore seems out, however, that HQ of an organization may not
to be important to invest more resources and time be free to arrange the sequence of implementa-
into defining the standardization scope to find the tions without obtaining support and consent of the
right fit between the degree of standardization and sites affected. In multinational and transnational
the costs of implementation. orientations, the individual sites have high degrees
of autonomy which might put them in a position
Implementation Procedure to resist the organizational change process.
Neither a big bang nor phased implementation
In the literature two procedures for ERP imple- procedure can realize learning effects if there is
mentations are mentioned: these are big bang not a sequential implementation procedure of
and phased implementation (cf. Umble, Haft, & any kind. So the company has to install devices
Umble, 2003). to collect knowledge from finished implementa-
In a big bang approach, every site introduces tions to use it by project management in follow-up
the system at the same time or one site implements projects. This knowledge, for example, contains
all modules at the same time. In the first case, the not only difficulties and solutions with system
risk of failure might be at the maximum. If there adjustments but also how the company reacted
is some sort of template and if this is insufficient to the new system, and how people have to be
for the individual sites, a successful implementa- trained on new functionalities or issues regarding
tion ends up in a disaster for the whole organiza- the local legal requirements.
tion. If there is no template, every site faces the Another very important dimension of the
problem of customizing its system according to implementation procedure is business process
the organization’s specifications. There might redesign (BPR). Depending on the decisions of
be redundant system configurations without any which kind of implementation orientation and
learning effects from previous implementations. scope of standardization is chosen by the com-
Depending on project management and site ca- pany, changes in existing business processes or
pabilities, the project could be successful or not. the system processes are implied. For example, the
76
implementation with a global template in a globally multinational orientation to an international one,

oriented organization demands the reconfiguration all the separate data will have to be cleaned up
of business processes in every site in accordance and consolidated in one database. This means a
with the template. For the sake of completeness, long-lasting process of consensus building with
this topic is mentioned here briefly, but due to its regard to appearance, form, and quantity of data
complexity and the implied problems, for example (cf. Oliver & Romm, 2002). Moreover, both steps
resistance to change or maintenance of the sys- presuppose a solution to the above mentioned
tem, the topic is beyond the chapter’s scope. For problem regarding the need for a central authority
a deeper insight into BPR and ERP, we refer to that defines and enforces the global ERP strategy,
the literature (Ng, Ip, & Lee, 1999; Koch, 2001; because the possibility of initiating a change of
Scheer & Habermann, 2000; Estevez et al., 2002; business orientation implies a sufficiently power-
Gunson & de Blais, 2002; Luo & Strong, 2004). ful center in the organization. A change in business
orientation could be necessary to stay competitive
or if HQ wants to change its position itself. In the
cHAnGInG tHe busIness case where HQ wants to act more like a portfolio
orIentAtIon manager, a company must be reorganized if its
form does not fit this intention. This could mean
If we follow the framework of business orientations more decentralization of power and decoupling
described above, companies’ ERP decision makers of sites. Because of this, the ERP implementation
have to bear another aspect in mind: it could be an process could become more complicated since
organization’s strategy to use the implementation decision-making power and the commitment of
of an ERP system in order to change the business top management is delegated one step down in the
orientation of the organization, thereby exploiting organizational hierarchy. For example, if in the
the ERP implementation process as a tool. Whether original business orientation all decisions are made
or not this use is feasible (see Future Trends and by HQ, they have access to all global resources
Research Directions), it complicates the task for of the organization. In case that one division or
ERP planners. New authorization procedures may business unit (site) general manager is given the
have to be created and new modules may have responsibility for an ERP implementation process
to be established within the different sites which for parts of the whole company, access to these
may result in yet another implementation. While resources is hampered or impossible for HQ. This
three of the four orientations have similar impli- will complicate the implementation of an ERP
cations regarding the implementation procedure, system in part of the company.
the change towards or away from a multinational
orientation seems to be a special case. The reason
for this is the centralized vs. decentralized struc- A model For monItorInG tHe
ture of data management. In the case of creating erP ImPlementAtIon Process
a multinational orientation, the individual sites
must be enabled to do business on their own. To The discussion so far is summarized in Figure 1.
do so, they will need an ERP system with all func- The business orientation leads to an ERP
tions, and the knowledge of how to handle such strategy based on a centralized or decentralized
a system to configure, maintain, and expand the approach in general. Adding the required manage-
system on their own (e.g., Luo & Strong, 2004). ment conduct, which is a precondition for a suc-
On the other hand, if an organization is to be cessful realization of an ERP strategy, results in
more integrated, for example by moving from a the constraints for an ERP implementation strat-
77
Figure 1. Steps in an ERP implementation strategy
egy. In order to evaluate the ERP implementation In the following, the three phases are—by way
strategy after an implementation orientation is of example—illustrated by the two polar cases
chosen, we distinguish three phases: intelligence global rollout and local implementation.
phase, implementation phase, and evaluation
phase. In the first phase, the scope of standardiza- Global rollout
tion is set by HQ which could have major effects
on project duration and costs. If this phase is Intelligence Phase
planned and executed well, it just has to be done
once in the ERP project. In the second phase, there First, there is a scanning phase which comprises
is the choice among a big bang or phased imple- the collection of important information regarding
mentation approach. To analyze this phase in more established business processes, needed data/data
detail, there are other models such as those by formats, interfaces, subsystems, national regula-
Markus and Tanis (2000) or Somers and Nelson tion, and so on. Once this information is gathered,
(2004) who divided the single implementation it is used by a strong moderator (most likely HQ)
into several other phases. Due to the disregard of to generate a template. The moderator identifies
an organization’s business orientation, this view core processes, adjusts these processes to unify
seems to be too limited because the effects on the them, and therefore defines the template’s core
ERP strategy are not considered. The data that processes. The template covers approximately
were collected in the implementation phase must 80-90% of all business processes (Umble et al.,
be analyzed in the third, evaluation phase. After 2003). This could be a time-consuming and ex-
the experiences of prior implementations, the pensive process depending on the number and the
sequence of future implementations could be heterogeneity of sites. Furthermore, this process
rearranged. Depending on the outcomes of this becomes more complicated if the sites are located
phase, the intelligence phase might have to be in different cultural areas (Davison, 2002; Soh,
revised. This model should enable an organization Sia, & Tay-Yap, 2000). Subject to their strategic
to plan its ERP strategy and incorporate learning meaning or role, not all sites have the same stake
effects. in this process.
78
Implementation Phase Implementation Phase
Once the template is generated, the actual global In this phase each site has to be examined by
rollout can begin. The usage of the template may local management. The result could be a fully
cause a business process redesign within the sites, customized ERP system, only obeying the con-
but still grants an adaptation to local requirements straints mentioned above. Of course, every single
in the range of 10-20%. Resistance to change implementation bears the same risks of failure.
should be low because all sites were participants Meeting the requirements of the single site will
of the process producing the template. As there increase this phase’s time and money require-
is only limited potential for customization, the ments. Afterwards the system is well tailored to
duration of the implementation phase should be the individual site, but each implementation will
significantly shorter than of the intelligence phase. repeat the whole process over and over again.
Evaluation Phase Evaluation Phase
The subsequent evaluation phase helps the or- The outcome of the evaluation phase depends on
ganization to improve its project management the kind of implementation procedure. If there
skills and therefore to reduce the time needed to was a bin bang (all sites at one time) approach,
introduce the template into further sites. Mean- no learning effects would be realized and the
while, the sequence of ERP implementations is result of the implementation process might be
checked, for example with regard to excellent unpredictable. If it was a phased implementation,
project management techniques and a well-built learning effects can be realized through “reuse” of
template, so that strategically important sites are processes or parts which could shorten the time
served first. for the next implementation phase.
Figure 2 shows a schematic demonstration
local Implementation of how the allocation of time and money could
differ in the two approaches. We propose these
Intelligence Phase curve progressions because of the above men-
tioned arguments that a global rollout starts with
In contrast to global rollout, only the information a high budget in the intelligence phase to create a
needs of HQ are relevant. After identifying the template, but could reduce implementation costs
important data which every site has to deliver, HQ because of the repetition and learning effects from
determines the data, data formats, and interfaces using the same process over and over again. On the
for interconnection within the structure. This dif- other hand, the costs of the local implementation
fers from the template because it is not a finished start on a low level and increase over the course
system that could be easily implemented in each of the project.
site; instead it creates constraints and demands to
them for their implementation or data support. It
is imaginable that HQ uses the collected data for conclusIon
performance comparison of each site. To reduce
information asymmetries or simplify compari- Implementation of an ERP system possibly takes
son, channels of data interchange may have to years and cost millions (Mabert et al., 2001).
be standardized. Therefore, organizations must manage this process
very carefully. Some approaches try to forecast
79
Figure 2. Allocation of time/money during project phases
the success of such a step (Magnusson & Nilsson, on the structure of a corporation. The structure of
2004) and distinguish between several dimen- an organization or the orientation of an MNC was
sions. Only rarely, the implied ERP implementa- considered as fixed and given. However, as we
tion strategy of such an organization method is mentioned in the section ‘Changing the Business
mentioned or examined. The exploration of the Orientation’, it could be one of top management’s
relationship between business strategy and—as objectives to use an ERP implementation process
recommended—the derived ERP implementation strategically. This strategic use would imply that
strategy can help: the implementation (process) of ERP systems
has a strong effect on organizations. It would
a. Management to detect and identify the ef- also raise the question whether ERP systems can
fects of the choice of an ERP strategy, and achieve such changes in structure by themselves
b. IT departments (e.g., represented by a CIO) or whether they act as a facilitator only. In the first
to explain the complexity and thus the re- case, the ERP project becomes more complex and
quired support by top management to lead challenging for both top management and project
a (successful) ERP implementation in their management. In the latter one, top management
organization. must consider additional measures which enable
the ERP project to reach the intended change or
The framework developed in this chapter measures that are supported by the ERP project.
gives hints regarding important considerations Top management as well as project management
and possible pitfalls in planning a global ERP must bear in mind that those additional measures
implementation process. Based on these insights can also have negative side effects on each other.
the presented model can help to monitor, control, What kind of measures would be appropriate,
and adjust the derived ERP strategy in order to meet how they support ERP projects and how they are
management’s expectations of the ERP system. supported by such projects, and in what useful
combinations they can be applied are interesting
questions for future research. Furthermore, if
Future trends And one allows for intentional effects of ERP imple-
reseArcH dIrectIons mentation processes on corporate structure, it
becomes necessary to address the question of
In this chapter we implicitly assumed that the unintended effects. These will not only regard
implementation of an ERP system has no effects ERP implementation processes in corporations
80
that are actively trying to change their structures, Butler, T., & Pyke, A. (2003). Examining the
but also ERP projects in fixed structures which influence of ERP systems on firm-specific knowl-
were discussed in this chapter. However, we did edge and core capabilities: A case study of SAP
not discuss the implications of unintended effects implementation and use. In Proceedings of the 11th
on ERP implementation processes due to limitation European Conference on Information Systems.
of space and reasons of complexity. The model
Clemmons, S., & Simon, J. S. (2001). Control
presented in this chapter allows for enhancement
and coordination in global ERP configuration.
by future work and research.
Business Process Management Journal, 7(3),
205–215. doi:10.1108/14637150110392665
AcknoWledGment Davenport, T. H. (1998). Putting the enterprise into
the enterprise system. Harvard Business Review,
The authors would like to thank the editors and 76(4), 121–131.
anonymous reviewers for their useful comments
Davison, R. (2002). Cultural complications
and suggestions. Furthermore, the authors would
of ERP. Communications of the ACM, 45(7),
like to thank Guido Schryen of RWTH Aachen
109–111. doi:10.1145/514236.514267
University for his help in editing this chapter.
Earl, M. J. (1993). Experience in strategic infor-
mation systems planning. MIS Quarterly, 17(1),
reFerences 1–24. doi:10.2307/249507
Akkermanns, H., & van Helden, K. (2003). Vi- Esteves, J., & Pastor, J. (1999). An ERP lifecy-
cious and virtuous cycles in ERP implementation. cle-based research agenda. Proceedings of the
European Journal of Information Systems, (11): International Workshop on Enterprise Manage-
35–46. ment Resource Planning Systems (EMRPS) (pp.
359-371), Venice, Italy.
Bancroft, N., Sprengel, A., & Seip, H. (1996).
Implementing SAP R/3: How to introduce a large Esteves, J., & Pastor, J. (2001). Enterprise resource
system into a large organization. Englewood planning systems research: An annotated bibliog-
Cliffs, NJ: Prentice Hall. raphy. Communications of the AIS, 7(8), 1–52.
Bartlett, C. A., & Ghoshal, S. (1998). Manag- Esteves, J., Pastor, J., & Casanovas, J. (2002).
ing across borders—The transnational solution. Monitoring business process redesign in ERP
Boston: Harvard Business School Press. implementation projects. Retrieved April 13, 2007,
from http://coblitz.codeen.org:3125/citeseer.ist.
Bingi, P., Sharma, M. K., & Godla, J. K. (1999). psu.edu/cache/papers/cs/31804/http:zSzzSzwww.
Critical issues affecting an ERP implementation. lsi.upc.eszSz~jesteveszSzBPR_amcis2002.PDF/
Information Systems Management, 16(3), 7–14. esteves02monitoring.pdf
doi:10.1201/1078/43197.16.3.19990601/31310.2
Firestone, J. M. (2002). Enterprise information
Brown, C. V., & Vessey, I. (2000). NIBCO’S portals and knowledge management. Boston:
“big bang.” In Proceedings of the International Butterworth-Heinemann.
Conference on Information Systems.
81
Fitzgerald, B., Russo, N. L., & O’Kane, T. (2003). Koch, C. (2001). BPR and ERP: Realizing
Software development method tailoring at Mo- a vision of process with IT. Business Pro-
torola. Communications of the ACM, 46(4), 65–70. cess Management Journal, 7(3), 258–265.
doi:10.1145/641205.641206 doi:10.1108/14637150110392755
Ghosh, S., & Ghosh, S. (2003). Global implemen- Krumbholz, M., Galliers, J., Coulianos, N.,
tation of ERP software—Critical success factors on & Maiden, N. A. M. (2000). Implementing
upgrading technical infrastructure. In Proceedings enterprise resource planning packages in dif-
of the International Engineering Management ferent corporate and national cultures. Journal
Conference—Managing Technologically Driven of Information Technology, 15(4), 267–279.
Organizations: The Human Side of Innovation doi:10.1080/02683960010008962
and Change (pp. 320-324).
Lehm, L., Heinzl, A., & Markus, L. M. (2001).
Gunson, J., & de Blais, J.-P. (2002). Implement- Tailoring ERP systems: A spectrum of choices and
ing ERP in multinational companies: Their their implications. In Proceedings of the 34th Ha-
effects on the organization and individuals at waii International Conference on Systems Science.
work. Retrieved April 13, 2007, from http://
Light, B. (1999). Realizing the potential of ERP
www.hec.unige.ch/recherches_publications/
systems: The strategic implications of imple-
cahiers/2002/2002.07.pdf
menting an ERP strategy: The case of global
Holland, C. P., & Light, B. (1999). A critical suc- petroleum. Electronic Markets, 9(4), 238–241.
cess factors model for ERP implementation. IEEE doi:10.1080/101967899358914
Software, 16(3), 30–36. doi:10.1109/52.765784
Light, B. (2005). Potential pitfalls in packaged
Holland, C. P., Light, B., & Gibson, N. (1998). software adoption. Communications of the ACM,
Global enterprise resource planning implementa- 48(5), 119–121. doi:10.1145/1060710.1060742
tion. In Proceedings of the Americas Conference
Luo, W., & Strong, D. M. (2004). A framework
on Information Systems.
for evaluating ERP implementation choices. IEEE
Hong, K.-K., & Kim, Y.-G. (2002). The critical Transactions on Engineering Management, 51(3),
success factors for ERP implementation: An 322–333. doi:10.1109/TEM.2004.830862
organizational fit perspective. Information &
Mabert, V. A., Soni, A., & Venkataramanan, M.
Management, 40, 25–40. doi:10.1016/S0378-
A. (2001). Enterprise resource planning: Com-
7206(01)00134-3
mon myths versus evolving reality. Business
Huber, T., Alt, R., & Österle, H. (2000). Tem- Horizons, 44(3), 69–76. doi:10.1016/S0007-
plates—Instruments for standardizing ERP sys- 6813(01)80037-9
tems. In Proceedings of the 33rd Annual Hawaii
Madapusi, A., & D’Souza, D. (2005). Aligning
International Conference on System Sciences.
ERP systems with international strategies. Infor-
Karimi, J., & Konsynski, B. (1991). Globalization mation Systems Management, 22(1), 7–17. doi:1
and information management strategies. Journal 0.1201/1078/44912.22.1.20051201/85734.2
of Management Information Systems, 7(4), 7–26.
82
Magnusson, J., & Nilsson, A. (2004). A conceptual Scheer, A.-W., & Habermann, F. (2000). Making
framework for forecasting ERP implementation ERP a success. Communications of the ACM,
success. In Proceedings of the International 43(4), 57–61. doi:10.1145/332051.332073
Conference on Enterprise Information Systems
Scott, J. E. (1999). The FoxMeyer Drug’s bank-
(pp. 447-453).
ruptcy: Was it a failure of ERP? In Proceedings
Markus, L. M., Axline, S., Petrie, D., & Tanis, C. of the Association for Information Systems 5th
(2000). Learning from adopters’ experiences with Americas Conference on Information Systems.
ERP: Problems encountered and success achieved.
Soh, C., Sia, S. K., & Tay-Yap, J. (2000). Cultural
Journal of Information Technology, 15(4), 245–
fits and misfits: Is ERP a universal solution?
265. doi:10.1080/02683960010008944
Markus, L. M., Tanis, C., & van Fenema, P. C. doi:10.1145/332051.332070
(2000). Enterprise resource planning: Multisite
Somers, T. M., & Nelson, K. G. (2004). A taxono-
ERP implementations. Communications of the
my of players and activities across the ERP project
ACM, 43(4). doi:10.1145/332051.332068
life cycle. Information & Management, 41(3),
Markus, M. L., & Tanis, C. (2000). The enterprise 257–278. doi:10.1016/S0378-7206(03)00023-5
systems experience—From adoption to success.
Sumner, M. (2000). Risk factors in en-
In R.W. Zmud (Ed.), Framing the domains of IT
terprise-wide/ERP projects. Journal
research: Glimpsing the future through the past
of Information Technology, 15, 317–327.
(pp.173-207). Cincinnati, OH: Pinnaflex Educa-
doi:10.1080/02683960010009079
tional Resources.
Umble, E. J., Haft, R. R., & Umble, M. M. (2003).
Ng, J. K. C., Ip, W. H., & Lee, T. C. (1999). A
Enterprise resource planning: Implementation
paradigm for ERP and BPR integration. Inter-
procedures and critical success factors. Euro-
national Journal of Production Research, 37(9),
pean Journal of Operational Research, 146(2),
2093–2108. doi:10.1080/002075499190923
241–257. doi:10.1016/S0377-2217(02)00547-7
O’Leary, D. E. (2000). Enterprise resource
Ward, J., Griffiths, P., & Whithmore, P. (1990).
planning systems: Systems, life cycle, electronic
Strategic planning for information systems.
commerce and risk. Cambridge, UK: Cambridge
Chicester: John Wiley & Sons.
University Press.
Oliver, D., & Romm, C. (2002). Justifying en-
terprise resource planning adoption. Journal
AddItIonAl reAdInG
of Information Technology, 17(4), 199–213.
doi:10.1080/0268396022000017761 Adam, F., & O’Doherty, P. (2000). Lessons from
Reimers, K. (1997). Managing information tech- enterprise resource planning implementations in
nology in the transnational organization: The po- Ireland—Towards smaller and shorter ERP proj-
tential of multifactor productivity. In Proceedings ects. Journal of Information Technology, 15(4),
of the 30th Annual Hawaii Conference on System 305–316. doi:10.1080/02683960010008953
Sciences. Bakos, Y. (1986). Information technology and
corporate strategy: A research perspective. MIS
Quarterly, 10(2), 106–119. doi:10.2307/249029
83
Biehl, M. (2007). Success factors for imple- Hanseth, O., & Braa, K. (1998). Technology as a
menting global information systems. Com- traitor: Emergent SAP infrastructure in a global
munications of the ACM, 50(1), 53–58. organization. In R. Hirscheim, M. Newman, &
doi:10.1145/1188913.1188917 J.I. DeGross (Eds.), Proceedings of the 19th In-
ternational Conference on Information systems
Chien, S.-W., Hu, C., Reimers, K., & Lin, J.-S.
(pp. 188-197).
(2007). The influence of centrifugal and centrip-
etal forces on ERP project success in small and Henderson, J., & Venkatraman, N. (1999). Stra-
medium-sized enterprises in China and Taiwan. tegic alignment: Leveraging information technol-
International Journal of Production Economics, ogy for transforming organizations. IBM Systems
107(2), 380–396. doi:10.1016/j.ijpe.2006.10.002 Journal, 38(2&3), 472–484.
Dowlatshahi, S. (2005). Strategic success fac- Huang, S. M., Hung, Y. C., Chen, H. G., & Ku,
tors in enterprise resource planning design and C. Y. (2004). Transplanting the best practice for
implementation: A case-study approach. Inter- implementation of an ERP system: A structured
national Journal of Production Research, 43(18), inductive study of an international company.
3745–3771. doi:10.1080/00207540500140864 Journal of Computer Information Systems, 44(4),
101–110.
Fleisch, E., Österle, H., & Powell, S. (2004). Rapid
implementation of enterprise resource planning Ioannou, G., & Papadoyiannis, C. (2004). Theory
systems. Journal of Organizational Comput- of constraints-based methodology for effective
ing and Electronic Commerce, 14(2), 107–126. ERP implementations. International Journal of
doi:10.1207/s15327744joce1402_02 Production Research, 42(23), 4927–4954. doi:1
0.1080/00207540410001721718
Gattiker, T., & Goodhue, D. (2005). What hap-
pens after ERP implementation: Understanding King, W. (2006). Ensuring ERP implementation
the impact of inter-dependence and differentiation success. Information Systems Management, 22(3),
on plant-levels outcome. MIS Quarterly, 29(3), 83–84. doi:10.1201/1078/45317.22.3.20050601/
559–585. 88749.11
Ghoshal, S., & Bartlett, C. (2001). The multi- Krumbholz, M., & Maiden, N. (2001). The
national corporation as an interorganizational implementation of enterprise resource planning
network. Academy of Management Review, 15(4), packages in different organisational and national
603–625. doi:10.2307/258684 cultures. Information Systems, 26(3), 185–204.
doi:10.1016/S0306-4379(01)00016-3
Gosain, S. (2004). Enterprise information systems
as objects and carriers of institutional forces: The Lall, V., & Teyarachakul, S. (2006). Enterprise
new iron cage? Journal of the Association for resource planning (ERP) system selection: A data
Information Systems, 5(4), 151–182. envelopment analysis (DEA) approach. Journal of
Computer Information Systems, 47(1), 123–127.
Grossman, T., & Walsh, J. (2004). Avoiding the
pitfalls of ERP system implementation. Informa- Luftman, J., Lewis, P., & Oldach, S. (1993).
tion Systems Management, 21(2), 38–42. doi:10. Transforming the enterprise: The alignment of
1201/1078/44118.21.2.20040301/80420.6 business and information technology strategies.
IBM Systems Journal, 32(1), 198–221.
84
Mahato, S., Jain, A., & Balasubramanian, V. Reimers, K. (2003). International examples of
(2006). Enterprise systems consolidation. Infor- large-scale systems—Theory and practice: Imple-
mation Systems Management, 23(4), 7–19. doi: menting ERP systems in China. Communications
10.1201/1078.10580530/46352.23.4.20060901 of the AIS, 11, 335–356.
/95108.2
Ribbers, P., & Schoo, K.-C. (2002). Program man-
Ng, C., Gable, G., & Chan, T. (2003). An ERP agement and complexity of ERP implementations.
maintenance model. In Proceedings of the 36th Ha- Engineering Management Journal, 14(2), 45–52.
waii International Conference on System Sciences.
Somers, T., & Nelson, K. (2003). The impact of
Parr, A., & Shanks, G. (2000). A model of strategy and integration mechanisms on enterprise
ERP project implementation. Journal of system value: Empirical evidence from manufac-
Information Technology, 15(4), 289–303. turing firms. European Journal of Operational
doi:10.1080/02683960010009051 Research, 146(2), 315–338. doi:10.1016/S0377-
2217(02)00552-0
Rathnam, R. G., Johnsen, J., & Wen, H. J.
(2004/05). Alignment of business strategy and IT Teltumbde, A. (2000). A framework for evalu-
strategy: A case study of a Fortune 50 financial ating ERP projects. International Journal of
service company. Journal of Computer Informa- Production Research, 38(17), 4507–4520.
tion Systems, 45(2), 1–8. doi:10.1080/00207540050205262
Reich, B., & Benbasat, I. (1996). Measuring the Voordijk, H., Van Leuven, A., & Laan, A. (2003).
linkage between business and information tech- Enterprise resource planning in a large construc-
nology objectives. MIS Quarterly, 20(1), 55–81. tion firm: Implementation analysis. Construction
doi:10.2307/249542 Management and Economics, 21(5), 511–521.
doi:10.1080/0144619032000072155
This work was previously published in Enterprise Resource Planning for Global Economies: Managerial Issues and Chal-
lenges, edited by Carlos Ferran and Ricardo Salim, pp. 37-54, copyright 2008 by Information Science Reference (an imprint
of IGI Global).
85
86
Chapter 1.6
Integrated Research and
Training in Enterprise
Information Systems
Chen-Yang Cheng
Penn State University, USA
Vamsi Salaka
Vittal Prabhu
AbstrAct is trained by this plan has an acquaintance with

the typical information systems in an organiza-
The success of implementing Enterprise Infor- tion. Further, for training and research purposes
mation System (EIS) depends on exploring and the authors developed prototype information
improving the EIS software, and EIS software systems that emulate the ones usually found in
training. However, the synthesis of the EIS imple- organizations. This ensures the EIS software logic
mentation approach has not been investigated. is consistent with the business logic. This chapter
In this chapter, the authors propose an integrated also discuss some of the case studies conducted
research and training approach for students and with the prototype systems.
employees about enterprise information systems
(EIS) that are encountered in an organization. Our
integrated approach follows the different stages IntroductIon
of a typical EIS project from inception to comple-
tion. These stages, as identified, are modeling, Enterprise Information Systems (EIS) constitute
planning, simulation, transaction, integration, the spectrum of information technology solutions
and control. This ensures that an employee who used by an organization and influence nearly all
aspects of operations of an organization. Typical
DOI: 10.4018/978-1-60566-146-9.ch011
Integrated Research and Training in Enterprise Information Systems
EIS systems such as Enterprise Resource Plan- tion (CMEI) at the Penn State University focuses
ning (ERP), Customer and Supplier Relationship on training students and working professional
Management (CRM and SRM), and Manufac- in EIS and related enterprise integration issues.
turing Execution Systems (MES). It is widely Further, as part of research at CMEI, projects
accepted that EIS deliver great rewards, but the are undertaken to study the information system
risks these systems carry are equally important. infrastructure for planning in Small Manufactur-
If an organization rolls out an EIS without ana- ing Enterprise (SME). These projects were aimed
lyzing the business implications, the logic of the at improving the operations management for
system may conflict with the logic of the business SMEs with the help of Information Technology
(Subramanian & Hoffer, 2005). This may result (IT). We found many avenues for improvement
in failure of implementation, wasting vast sums both in operations management and information
of money, and weakening important sources of systems, but there were barriers in implementing
organization’s competitive advantage. such projects. These barriers vary from capital
Prior research has investigated and identified and complexity of the systems to human inertia
the critical factors that influence the successful to change. But the key barrier we noticed among
implementation of EIS. Education and training SMEs is the lack of tools or expertise for handling
of employees on EIS is one of the most widely the EIS software. Most of the EIS software avail-
recognized factors (Umble & Haft, 2003) (H. able requires extensive training for its use and
Hutchins, 1998; Ptak & Schragenheim, 2000) maintenance. SMEs cannot afford this training
(Laughlin, 1999). EIS implementation requires the as it involves hiring specialized engineers and
knowledge of the employees to smoothly carry on consultants. Based on our experience with SMEs
the business process and further solve problems and to reduce training costs for EIS deployment
that arise from the new system. Even with good in large manufacturing enterprises, we developed
technical assistance, the complete potential of EIS a rollout plan which encompasses the different
cannot be realized without employees having an settings in an organization that employees use
appreciation of the capabilities and limitations of EIS. The important stages that we identified are
the system (Somers & Nelson, 2001). To make the the Modeling, Planning, Simulation, Integration,
employee training successful, it is agreed that it Execution and Control. We developed prototype
should start early, preferably well before the rollout software for each stage that emulates the func-
of EIS begins. The upper management in large tionalities of typical organizational software.
manufacturing enterprises often underestimate In the following sections, we present an overall
the level of education and training necessary to view of our rollout plan and then elaborate on
implement EIS and as well as its associated costs. the individual stages. In this work, our focus has
It has been suggested that reserving 10–15% of been mostly centered on information systems that
the total EIS implementation budget is a good are typical in manufacturing organizations, but it
practice to ensure the employees receive enough can be readily extend to other industry segments.
training (McCaskey & D. Okrent, 1999) (Umble
& Haft, 2003). With the estimated budgets for
implementing EIS in billions of dollars the cost rollout PlAn For
of training the employees on these systems is a trAInInG In enterPrIse
very sizeable portion (Hong & Kim, 2002). These InFormAtIon systems
costs can be brought down if the employees have a
prior education and training on EIS. Hence, at the The rollout plan developed for training in EIS was
Center for Manufacturing and Enterprise Integra- modeled to follow the different stages that a new
87
project undergoes from inception to completion poses, we developed Sensors to Suppliers

as shown in Figure 1. Following this procedure it (S2S) planning and simulation tool (Mehta,
is envisioned that training employees or students 2003).
would introduce them to the spectrum of EIS • The integration phase is where we focus
software that are used in an organization. We have our discussion on the need and advantages
developed prototype software that would emulate of seamless integration of various informa-
the working of industry standard EIS software. It tion systems within the organization. To
is important to note that the purpose of develop- train and demonstrate how data level inte-
ing prototype software for different stages in the gration works, we have developed Schema
rollout plan is to reduce the complexity while Lens software, which extracts the schema
maintaining the fundamental concepts found in from a source database and has the capa-
industrial strength software. The different stages bilities to map it with transformations to a
that are part of our rollout plan are as follows: destination.
• The transaction phase is where organiza-
• In the modeling phase the requirements tions execute the business process to cre-
and objectives of a project are established. ate customer value. In general, the trans-
In general the requirements are captured action system in an enterprise constitutes
by enterprise models using modeling tech- the Customer Relationship Management
niques such as data flow diagrams (DFD), (CRM), and Enterprise Resource Planning
and unified modeling language (UML). (ERP) system. CRM provide a single view
The typical software used in industry for of all customer interactions, and campaign
developing enterprise models are Microsoft management for personalized services.
Visio® and Rational Rose®. For training ERP systems execute a fixed sequence of
purposes we developed a tool customizing defined functions such as purchasing or
Microsoft Visio® to provide user friendly selling. They lack the flexibility to imple-
interfaces supporting a wide variety of ment varying business process. Hence, we
modeling techniques (Mathew, 2003). developed the automated workflow system
• The planning phase is where organizations which integrates the ERP system with web
evaluate alternatives, do what-if analysis services.
for designing and to come up with a plan • The control phase in manufacturing sys-
of execution. In general the planning phase tems is where the data collected from the
is strongly tied with the simulation phase shop floor is fed back to business systems to
where the feasibility and the success of the make informed decisions. Here we discuss
proposed plans are evaluated by the devel- the Radio Frequency Identification (RFID)
oped simulation models. For training pur- data collection techniques with distributed
Figure 1. Rollout plan for training in EIS
88
computing to solve computational prob- important to train the users of an EIS system in
lems in manufacturing. For training pur- enterprise modeling to familiarize them with the
poses, we implemented the RFID-based organization’s business process and provide an
Manufacturing Execution System (MES) understanding of how EIS software drives the
for updating the cutting tool information business process. For the purpose of training
in real-time and built a library for distrib- and conducting related research we developed
uted computing (Cheng & Prabhu, 2007a,, an Integrated Decision Support (IDS) framework
2007b,, 2007c). by enriching the interfaces in Microsoft Visio®
for more specific data collection and modeling
requirements. The customized environment is
enterPrIse modelInG shown in Figure 2.
There exists a number of enterprise modeling
There is a general lack of documentation and stan- techniques employed in the industry and are
dardization of business processes in organizations. discussed by (Kamath, Dalal, Chaugule, Sivara-
This lack of documentation leads to inconsistent man, & Kolarik, 2003). We support a wide range
formats and data. The flow of documents is often of them in IDS for the following reasons: A par-
not clear, leading to gaps in these processes. For ticular modeling tool can be better suitable for
implementation of information systems, it is very expressing a part of the problem; and different
important that an organization has standardized users can have their own preference and comfort
business processes. Enterprise modeling facili- with a particular modeling tool. In one of the case
tates the understanding of business processes and studies we conducted with this tool, we were able
relations that exist within and across the various to work with a defense and aerospace organization
departments of an organization (Kamath, Dalal, to capture their business process and IT infra-
Chaugule, Sivaraman, & Kolarik, 2003). It is structure to identify the requirements for enterprise
Figure 2. Enterprise modeling environment customized in Microsoft Visio®
89
integration. Further, we were able to evaluate the business model as quickly as possible and in
various available integration choices collabora- as much detail as desired by the user. Once the
tively based on the estimates of cost and complex- business model is captured, it can be simulated
ity. The detailed functioning of IDS along with and desired metrics can be analyzed. The user
the case study is presented in a related work can then iteratively change planning policies
(Salaka & Prabhu, 2006). and quantitative variables of the model, add and
remove components to gain insights under dif-
ferent scenarios. S2S is implemented in object
PlAnnInG And sImulAtIon oriented programming language on the Microsoft
.Net platform. The simulation is executed by hav-
One of the most common information system ing objects, representing business components
deployments in manufacturing is for identifying interacting with each other. Using the develop-
planning requirements. These systems maintain ment interface, business models can be created,
order status and explode it into material and op- modified, saved and merged.
erational requirements based on bill of materials. S2S is currently being used in an academic
Further, these systems act as a bookkeeping tool setting to educate and train students in simulation
and are linked to accounting. Simulation allows for evaluating various planning policies in manu-
the quantitative analysis of the operation policies facturing systems. Further, it has been an attractive
made in the planning phase. As computing power tool for SMEs who lack the capital, experience,
becomes inexpensive, using software simulation and expertise in using other commercial simula-
for analysis is becoming much widely acceptable. tion tools. We demonstrated the use of S2S in
In general the planning phase is strongly tied with SMEs by a case study performed in a wood
simulation where the feasibility and the success of manufacturing enterprise. The details of the case
the proposed plans are evaluated by the developed study are presented in a related work (Salaka,
simulation models. We developed a Sensors to Mehta, & Prabhu, June 2005).
Supplier (S2S) tool that enables both planning
and simulation in manufacturing systems with
their complex structure and behavior. A brief enterPrIse InteGrAtIon
description of the S2S tool is presented in the
following section. Enterprise integration (EI) is creating new business
solutions by effectively utilizing the capabilities
sensor to supplier Planning of existing software applications by allowing
and simulation tool rapid movement of information between them.
EI is strategically important for improved busi-
The S2S simulation tool contains a rich library of ness process management. The market for EI was
components that can be configured to represent valued at $8.3 Billion for the year 2004-2005 and
a particular business. The components support was expected to grow 20% during the year 2005-
different policies and, the user controls decision- 2006 (Salaka & Prabhu, 2006), which indicates
making rules and the quantitative variables. Apart the emphasis on integration in the industry. The
from that, the tool provides an interactive and knowledge of enterprise integration is an essential
intuitive user interface for model development part of training on EIS software for students and
and performance analysis. The elements of the organizational employees to make them realize
modeling tool and their function are shown in the potential of integrated enterprises. As part of
Figure 3. The objective of the tool is to capture our rollout plan, we train students on the con-
90
Figure 3. S2S Modeling tool
temporary standards and techniques in EI. The from the source to the tables and columns of the
training plan on integration includes familiarizing destination database. Currently, Schema Lens is
the undergraduate students with typical XML being used as a test bed for a PhD student’s work.
features that are supported currently by most of
the industry software for data integration. For
this, we demonstrate the XML features that are control oF mAnuFActurInG
in Microsoft Office® software and how these systems
features can be used in migrating data between
software applications. Figure 4 shows one of The general idea of control is to use feedback
the demonstrations to the students as part of the mechanism from sensor to adjust a system to
training. For research purposes and acquaint obtain desired results. In a manufacturing sys-
students industry standard integration software, tem, the feedback from shop floor can improve
we developed Schema Lens data level integration business decisions. We developed RFID-based
software, which is described next. Manufacturing Execution System for the purpose
of training and related research in latest advances
schema lens of manufacturing control with the advent of RFID
technology. The current system is modeled to ad-
Figure 5 shows the framework of the software. dress the tool management issues in machining.
Schema Lens transfers the data from a legacy Further, to demonstrate the importance of having
system to a production system extracting schema faster solution frameworks that can take advantage
and data as an XML file from the source database of large data generated from RFID technology
using a Java® application. The front end for the in manufacturing control, we built PennDiCon,
application is a web browser. The browser con- a library for distributed computing. In the fol-
nects to a web application residing in a web server. lowing sections, we detail the functioning of the
Further the web server contacts the destination above system.
database and enables the mapping XML data
91
Figure 4. Demonstration of XML features in Microsoft Office®
rFId-based manufacturing tool is attached with a RFID tag that records the
execution system tool’s usage time, preset data (projection, edge
radius, and effective diameter), and operation
Figure 6 illustrates the RFID-based Manufacturing data (speeds, feeds and depth of cut). When the
Execution System framework for tool manage- tool is attached to the spindle, the RFID reader
ment. Each Computer Numerical Control (CNC) on the CNC machine is configured to read the
Figure 5. Framework of Schema Lens
92
data from the RFID tags. The purpose of this trAnsActIon In enterPrIse
avoids all sources of potential data entry error, InFormAtIon system
tool offsetting errors and possible tool crashes.
After the completion of each operation, the RFID A business transaction is a process that has two
encoder on the CNC machine is programmed to or more partners communicating to synchronize
write the usage time of the tool onto the RFID the related activity states in all information sys-
tag and further update the Quality Performance tems for the purpose of creating customer value.
Management (QPM) system. Based on the demand The transaction system of the outbound business
for tool replacement as scheduled by QPM, a pur- comprises CRM system. CRM enables business to
chase order can be created from ERP in advance track and manage all of their customers’ interac-
to avoid stock shortages and reduce downtime. tions over the lifetime, from first contact through
to purchase and post-sales. On the other hand, the
distributed computing -Penndiconn transaction system in an enterprise constitutes ERP
system. ERP systems execute a fixed sequence of
PennDiCon is a distributed computing library that defined functions such as purchasing or selling.
provides several communication functionalities However, existing CRM or ERP systems fail to
such as peer-to-peer, multicast and broadcast, adapt to frequent requests for changing business
which can be called from C/C++ programs. In processes. Therefore, we developed a workflow
distributed computing environment, any entity system, Process Automation with Web Services
can communicate with another entity or a group of (PAWS), which support modification of business
entities using PennDiCon, allowing it to dynami- logics at runtime. A brief description of the training
cally discover resources in the system. The detailed and research in transaction system is presented in
architecture of PennDiConn along with the case the following section.
study is presented in a related work (Sharma &
Prabhu, 2005). outbound transaction system
Outbound transaction covers the types of retailing,

channels, tools for data collection, and transaction
standard. The adopting of Information technolo-
gies in the outbound transaction system includes
Figure 6. RFID-base MES System
93
CRM, point of sale systems, barcode, RFID/EPC, class on manufacturing systems. Typically, this
global data synchronization, EDI, XML, and web training incorporates a day or two of lectures on
services. Students learn the hands-on experimen- how business processes need to work in tandem
tation/demonstrations such as Microsoft CRM, to make a business successful and how an ERP
PC-based POS terminals, scanners, RFID tags software such as SAP® are used in the industry
and readers, and integrated software for store and to accomplish the same. For a small to medium
home office using Microsoft Retail Management size company, ERP software could be in-house
System (RMS). Training course includes a group development software such as TEAM (Total Enter-
project in which students will design and construct prise Application Manger). TEAM was developed
an IT application focused on retail. A single view in Microsoft Access® by Georgia Technology
of all customer interactions, and campaign man- Economic Development Institute, shown in Figure
agement for personalized services using MSCRM 7. By using Visual Basic Application, (VBA),
software is adopted in training. It helps student to students can learn to develop or extend the ERP
learn how to target new customers, and manage software, and also get exposure to the software
marketing campaigns, enabling closer relation- through hands-on exercises.
ships with customers.
Process Automation with Web
Inbound transaction system- erP services (PAWs)
ERP constitutes the typical transaction system PAWS is an automatic workflow system for tool
within an organization. This system provides infor- quotation process within and between the supply
mation systems support for planning, production, chains. It extends the RFID-based MES discussed
marketing, and finance. Hence, it is important to in the previous section. Figure 8 illustrates the
educate students and employees about this soft- framework of PAWS. In PAWS, ERP broadcasts
ware. To learn the prevalent ERP software, we the tool requirements to the web services of cer-
currently have the training on Production Planning tificated suppliers registered with the Universal
module of SAP® part of a senior undergraduate Description Discovery and Integration (UDDI)
Figure 7. Mastering Scheduling in the TEAM ERP
94
server. The web services in each supplier’s server industry practitioners benefited from this training.
verify the inventory and price in its own database The following is the summary of training experi-
and respond to the quotation based on its current ences from various stages of the rollout plan:
status. The business logic of the Request for
Quotation (RFQ) process in PAWS decides the • Microsoft Visio has been used to train un-
candidate supplier and starts another RFQ process dergraduate students in enterprise mod-
for logistic provider. This automation workflow eling techniques such as UML and IDS
system not only integrates the inner-business was used as a test bed in a master’s thesis
process including QPM, ERP, and mailing sys- (Salaka, 2004). Further, IDS has been used
tem, but also combines the business-to-business in an industry project to demonstrate its
(B2B) process in the supply chain. Figure 9 shows capabilities to analyze an enterprise’s busi-
the quotation process which broadcasts the RFQ ness process and identify requirements for
synchronous to each supplier. Thus PAWS dem- integration (Salaka & Prabhu, 2006).
onstrates to students the use of work flow systems • In simulation and planning, Sensors to
to automate a transaction process to shrink time Suppliers Chassis (S2S) is being primarily
consuming RFQ process(Cheng & Prabhu, 2007a,, used in two ways. One way to show indus-
2007b,, 2007c). try managers the cost Vs benefit analysis
of various policies in manufacturing sys-
tems. The other use of S2S is in academic
student exPerIences settings where it is being used to educate
students in evaluating various manufactur-
Given the breadth of the proposed rollout plan, ing control policies and as an example to
there is no one course or an example that can be train students in object-oriented software.
used to train the students in the rollout plan. Even Also S2S is being used in an undergradu-
then, different categories of students including ate honors thesis to study the advantages
graduate students, undergraduate students, and
Figure 8. The framework of PAWS execution
95
Figure 9. Quotation process
of the tie-in between a simulation environ- tandem with ERP and MES systems to au-
ment and an ERP system. tomate an organization’s business process.
• In the integration part of the rollout plan, • The control software developed is being
undergraduate students are trained on used as a test bed in a graduate course to
XML features in software applications study various control algorithms for manu-
useful in integrating software applications. facturing systems. It is also being used by
Further, they are familiarized with web a PhD student for research.
services and the impact of service oriented
architectures on an enterprise’s business
process. Further, the software tool Schema conclusIon
Lens developed is being used as a test bed
for a PhD student’s research. In this work, we identified the need for a rollout
• In transaction, students are trained on map for education and training in implementing en-
SAP® modules to illustrate how busi- terprise information systems (EIS). In accordance,
ness processes need to work in tandem to we proposed a rollout map that encompassed
make a business successful. The students different settings in an organization where em-
are provided login to the SAP® systems to ployees use EIS. Further, we developed prototype
learn some of the software functions and information systems that emulate the ones that are
get familiarized with the typical data flow typical in organizations. This software is currently
among various modules of the SAP® sys- being used in academic settings to educate and
tems that support various business func- train students in using EIS. The current authors
tionalities encountered in an organization. are working to incorporate the rollout plan part of
Further, PAWS, the workflow software de- the courses taught in graduate and undergraduate
veloped, helps to illustrate to graduate stu- curriculum. The student response so far to the
dents how a workflow system can work in training has been very positive. The important
96
takeaway for other researchers and academicians Laughlin, S. (1999). An ERP game plan. The
is that following this rollout plan the students can Journal of Business Strategy, 20(1), 32–37.
be familiarized with the broad spectrum of EIS doi:10.1108/eb039981
systems in a systematic way. In the future, we
Mathew, S. (2003). Quantitative Models for Total
plan to extend this training to small manufacturing
Cost of Ownership of Integrated Enterprise Sys-
enterprises that lack the experience in using EIS.
tems. Pennsylvania State University, University
Park.
reFerences McCaskey, & Okrent, M. D. (1999). Catching
the ERP second wave. APICS––The Performance
Cheng, C. Y., & Prabhu, V. (2007a). Applying Advantage, 34–38:
RFID for Cutting Tool Supply Chain Management.
Proceedings of the 2007 Industrial Engineering Mehta, R. (2003). Software Modeling Tool for
Research Conference, 637-642 Analysis of Manufacturing and Supply Networks.
Pennsylvania State University, University Park.
Cheng, C. Y., & Prabhu, V. (2007b). Complexity
Metrics for Business Process Enabled by RFID N/A. (1998). 7 key elements of a successful
and Web Services. Proceedings of 17th Interna- implementation, and 8 mistakes you will make
tional Conference on Flexible Automation and anyway. APICS 1998 International Conference
Intelligent Manufacturing (FAIM 2007), 812-819 Proceedings. Falls Church, VA, 356–358
Cheng, C. Y., & Prabhu, V. (2007c). Performance Ptak, C., & Schragenheim, E. (2000). ERP: Tools,
Modeling of Business Processes Enabled by RFID Techniques, and Applications for Integrating the
and Web Services. Proceedings of 6th IEEE/ACIS Supply Chain. Boca Raton, FL: St. Lucie Press.
International Conference on Computer and Infor- Salaka, V., Mehta, R., & Prabhu, V. V. (June 2005).
mation Science (ICIS 2007), 718-723 Sensors-to-Suppliers Simulation Modeling of
Hong, K. K., & Kim, Y. G. (2002). The criti- Manufacturing Supply Chains. Proceedings of
cal success factors for ERP implementation: an the 15th International Conference on Flexible
organizational fit perspective. Information & Automation and Intelligent Manufacturing (FAIM
Management, 40, 25–40. doi:10.1016/S0378- 2005), Bilbao, Spain
7206(01)00134-3 Salaka, V., & Prabhu, V. V. (2006). Project Man-
Hutchins, H. (1998). APICS 1998 International agement for Enterprise Integration. The tenth
Conference Proceedings, Falls Church, VA, 1998, IEEE conference on enterprise computing (EDOC
pp. 356–358. 2006), Hong Kong
Kamath, M., Dalal, N., Chaugule, A., Sivaraman, Sharma, A., & Prabhu, V. V. (2005). Computing
E., & Kolarik, W. (2003). A review of enterprise and Communication Quality of Service for Dis-
process modeling techniques. In V. V. Prabhu, tributed Time-scaled Simulation in Heterarchical
S. Kumara & M. Kamath (Eds.), In Scalable Manufacturing Control. International Journal of
Enterprise Systems: An Introduction to Recent Modelling and Simulation.
Advances (pp. 1–32). Boston, MA: Kluwer Aca-
demic Publishers.
97
Somers, T. M., & Nelson, K. (2001). The impact Umble, E. J., & Haft, R. R. (2003). Enterprise
of critical success factors across the stages of resource planning: implementation procedures
enterprise resource planning implementations. and critical success factors. European Journal
Proceedings of the 34th Hawaii International Con- of Operational Research, 146(2), 241–257.
ference on System Sciences, Hawaii, USA, 1–10 doi:10.1016/S0377-2217(02)00547-7
This work was previously published in Encyclopedia of Global Implications of Modern Enterprise Information Systems: Tech-
nologies and Applications, edited by Angappa Gunasekaran, pp. 195-208, copyright 2009 by Information Science Reference
98
99
Chapter 1.7
Free and Open Source
Enterprise Resources Planning
Rogerio Atem de Carvalho
Federal Center for Technological Education of Campos, Brazil
AbstrAct rently focused only in cost aspects, bringing more

attention to other aspects and pointing out their
This chapter introduces the key aspects of Free/ innovative potential.
Open Source Enterprise Resources Planning sys-
tems (FOS-ERP). Starting by related work carried
out by researchers and practitioners, it argues in IntroductIon
favor of the growing acceptance of this category of
enterprise systems while showing how this subject Free/Open Source1 ERP (FOS-ERP) systems are
is not yet well explored, especially by researchers. gaining a growing acceptance and consequently
The goals of this chapter are to highlight the dif- improving their market share. According to a recent
ferences between FOS-ERP and their proprietary market study, FOS-ERP related services would hit
equivalents (P-ERP) in terms of business models, about US$ 36 billion by 2008 (LeClaire, 2006).
selection, customization, and evolution; and show- The reasons for this phenomenon are basically two:
ing the challenges and opportunities that they offer lower costs and free access to application’s source
to adopters, vendors, researchers, and individual code. On the cost side, they impose reduced or no
collaborators. Therefore, this chapter tries to broaden investment in licensing in general. On the access to
the discussion around the FOS-ERP subject, cur- code side stands the perception that if customization
is inevitable, why not adopt a solution that exposes
its code to the client company, which can freely
DOI: 10.4018/978-1-59904-859-8.ch003
Free and Open Source Enterprise Resources Planning
adapt the system to its needs? Maybe this second work has been done on FOS-ERP, this subject is
reason is more complex and much less studied and still a new one, with many topics to be explored
is addressed in many topics later in the chapter. and tendencies to be confirmed, since the num-
Given this raising on FOS-ERP deployment, ber of adopters and the operation times are still
and the relative small number of references to this small in relation to the P-ERP figures. In fact,
subject, instead of simply comparing functional- FOS-ERP is a barely explored research subject.
ities of various different solutions, this chapter As said before, the first academic paper on this
aims to a) present tendencies on open source specific subject was Smets-Solanes & De Carvalho
software in general and open source enterprise (2003); the first paper on evaluating FOS-ERP is
systems that directly influence on FOS-ERP, b) De Carvalho (2006); and the first international
highlight the differences between FOS-ERP and event on FOS-ERP was held in Vienna, Austria,
proprietary ERP (P-ERP) in terms of business also in 20062. These facts show how FOS-ERP is
models, selection, customization and maintenance, a young research area, with relatively very little
and c) identify the challenges and opportunities academic effort put on it until now.
that they offer to stakeholders and developer However, some good work on related topics
communities. can be found. Currently the most in-depth analy-
sis of the economic impact of Free/Open Source
Software (FOSS) in enterprise systems was the
relAted Work one conducted by Dreiling and colleagues (Dreil-
ing, Klaus, Rosemann & Wyssusek, 2005). The
While increasing in market importance, FOS-ERP authors argue that “standards that supposedly open
is still poorly analyzed by academy, where large development by ensuring interoperability tend to
quantities of articles put their research efforts on be interpreted by enterprise systems global players
P-ERP deployment, project management, and according to their interest”. The authors follow
economic aspects (Botta-Genoulaz, Millet & this reasoning showing the deeper consequences of
Grabot, 2005). Research on FOS-ERP software is this: “[global players interests] might be incongru-
rather deficient, and, therefore, a series of relevant ent with the interests of the software industry at
aspects of FOS-ERP, which differentiate them large, those of users organizations, and may also
from P-ERP, are still not well understood. As an have effects on local and national economies.” And
example of this situation, a research conducted more: “despite control of interfaces and standards
on the FOS-ERP evaluation subject, has shown by few software developers, even integration of
how evaluating FOS-ERP brings more concerns the information infrastructure of one single com-
than evaluating P-ERP (De Carvalho, 2006). One pany with one brand of enterprise system cannot
indication that FOS-ERP seems to be another be consolidated over time [citing many other
situation where technology has outstripped the authors].” On the open standards subject, they
conceptual hawsers, is the fact that, according conclude, “software engineering principles and
to Kim and Boldyreff (Kim & Boldyreff, 2005), open standards are necessary but not sufficient
“by September 2005 only one paper about Open condition for enterprise software development
Source ERP (Smets-Solanes & De Carvalho, becoming less constrained by the politics of
2003) has been published in the whole of ACM global players, responsive to user interests, and
and IEEE Computer Society journals and pro- for ensuring a healthy software industry that can
ceedings, whereas more numerous articles have cater for regional market.”
been published in non-academic industrial trade On the innovation side, Dreiling and colleagues
magazines.” Although nowadays more research state that many economists agree to the point of
100
dominant companies – like the ERP global play- instance is a subject with only a few works on it.
ers – are less disposed to respond to articulated Herzog (2006) presents a very comprehensive
customer requirements, and monopolies as well approach that identifies three different methods
oligopolies tend to stifle product and service inno- for implementing a FOS-ERP solution - select a
vation. Furthermore, “controlling architectures by package, develop one by itself, and integrate best
means of proprietary software and open standards of breed solutions – and five criteria for evaluat-
in the enterprise application industry appears to ing alternatives: functional fit, flexibility, support,
actually preclude innovation that could be of continuity, and maturity. This method introduces
benefit for many users of enterprise systems”, the interesting possibility, not yet well explored
which includes less developed economies. This in practice, of integrating solutions from different
seems to be a serious problem, since adapting is vendors through Enterprise Application Integra-
a crucial point to ERP, which by nature must be tion (EAI) techniques. A successful case study on
adapted to the adopter needs. This conclusion mixing P-ERP solutions is described by Alshawi
reinforces a positive consequence of the freedom and colleagues (Alshawi, Themistocleous & Al-
of manipulating the source code by itself in FOS- madani, 2004) – but the literature lacks examples
ERP: if the vendor changes its contract terms, the on doing the same with FOS-ERP.
client company is not locked in to a particular De Carvalho (2006) also presents an FOS-ERP
solution supplier (Kooch, 2004). Additionally, evaluation method, named PIRCS, that holds some
can two competing companies derive a strategic similarity with Herzog’s3, however stressing more
differential using the same ERP? Although this on risk evaluation, given the strategic nature of
problem can also happen with FOS-ERP, it seems ERP. In fact, according to Caulliraux and col-
to be bigger for P-ERP, since, due to the tightly leagues (Caulliraux, Proença & Prado, 2000) ERP
control over source code, adaptations are limited is strategic, given that “it is a major commitment
to parameterization or high-cost functionality of money, and thus with long range implications
changes through Application Program Interfaces even if only from a financial point of view”, and
(APIs) or proprietary languages, restricting real ERP systems are also important not only as a
differentiation and raising customization costs tangible asset, but “as a catalyst through their
(De Carvalho, 2006). Therefore, if the fact that implementation in the formation of intangible as-
integration among processes can by itself becomes sets and the company’s self-knowledge.” Aiming
a source of competitive advantage (Caulliraux, to include risk considerations, the PIRCS method
Proença & Prado, 2000), this can be extrapolated seeks to identify weaknesses in the FOS-ERP’s
to the possibility of changing source code to drive development environment during its evaluation
an even better advantage. process phases. These phases name the process,
If on one hand FOS-ERP can foster innova- and are summarized as Prepare the evaluation
tion and give more power to adopters, on the process, Identify the alternatives, Rate alterna-
other some important questions are yet to be tives’ attributes, Compare alternatives’ results,
answered, given that this type of FOSS is still a and Select the one that best fits the adopter needs.
newcomer to the enterprise systems landscape. Also related to the strategic nature of ERP,
Even some enthusiasts recognize that FOS-ERP during PIRCS’ Preparation phase, the adopter
vendors service level have much to improve and must define its strategic positioning in relation to
gain experience, while in contrast, P-ERP have the product, behaving as a simple consumer, only
a mature network of consulting partners and a getting the solution from the vendor, or becom-
long history of success and failures (Serrano & ing a prosumer (Xu, 2003), by mixing passively
Sarrieri, 2006). In fact, evaluating FOS-ERP for purchasing commodity parts of the system with
101
actively developing or customizing strategic ones and the ability to reach more customers through
by itself. Of course, choosing how to behave is not improved pricing flexibility”. Other aspect on
a simple decision in these cases, since it involves the vendor side identified by these authors is that
a series of demands like expertise on the FOS- opening the code can reduce the costs on software
ERP platform and architecture, dealing with the testing and Research & Development tasks. These
developer community – which can mean manag- advantages are stimulating a better market accep-
ing demands of disparate stakeholders (West & tance, according to a Zinnov (2006) study, which,
O’Mahony, 2005), and allocating resources for among other things, shows figures on the raising
development. It is a question of weighting the of venture capitalism participation and the higher
direct and indirect gains of developing parts of penetration in the US enterprise systems market
the system with the shortcomings of doing so. by open source solutions. Furthermore, Goth
Many other subjects related to open software in (2005) affirms that the open software market is
general that affect FOS-ERP should be addressed on a “second wave” towards enterprise software,
to better understand their dynamics. Crowston and and that FOSS business models are finally ready
Howison (2006) assess the health of Open Source for facing this new market challenge. These two
communities as a way of helping checking if an last references points to a general improvement
FOSS is suitable for the adopter or contributor on the relation between FOSS communities and
needs – this kind of assessment can be one of the enterprise systems users.
tools to check a specific FOS-ERP project maturity. Despite the differences, FOS-ERP and P-ERP
Assessing FOS-ERP communities means under- certainly have one thing in common: both have
standing other organizations’ behavior towards the a company behind their deployment activities.
project: since ERP in general are not for individual Although there exist FOS-ERP maintained al-
use, contributors most of times are companies’ most solely by communities formed basically
employees, not free-lancers. Hence, to understand by individuals, like GNU Enterprise, it seems
the differences between this and other types of that only company-sponsored FOS-ERP, such as
open software, it is necessary to understand how Compiere, ERP5, OpenMFG, and SQL Ledger,
commercially sponsored and community built are really successful. In other words, FOS-ERP
FOSS projects behave. According to West and are typically of the commercial open source kind,
O’Mahony (2005), one of the key moments of which “a for-profit entity owns and develops”, ac-
commercial FOSS is the start-up phase: when the cording to Riehle (2007) classification. The next
project code is opened, “an infant community is topics show how FOS-ERP differs from P-ERP,
presented with a large complex system that may present opportunities and challenges that this kind
be harder to decipher”, thus the FOS-ERP creator of software offer to developers and adopters, and
may have to wait until contributions from other finally drawn some conclusions on the subject.
firms become viable and also advantageous - the
main economic incentive for firm participation
is the emancipation from the price and license dIFFerences betWeen
conditions imposed by large software companies Fos-erP And P-erP
(Wang and Chen, 2005), but the potential for doing
such substitution must be only latent in the project. The fact that FOS-ERP expose their code forces
The same type of incentive is identified by Riehle vendors and adopters processes to accommodate
(2007), who stats that solution providers can take the consequence that customization and mainte-
advantage from open source software “because nance can be done by other than the vendor. This
they increase profits through direct costs savings fact means that the adopter is free to choose the
102
participation level of the vendor in the different defines seven life-cycle phases for any enterprise
phases of the ERP life cycle – meaning that, at entity that are pertinent during its life. These phases,
some extent, vendor participation can be also presented on Figure 1, can be summarized as follows:
customized4. Analyzing the differences between
open source and proprietary ERP depend on side a. Identification: identifies the particular
of the commercial relation the organization is: enterprise entity in terms of its domain and
adopter or vendor. environment.
b. Concept: conceptualizes an entity’s mission,
differences for the Adopter vision, values, strategies, and objectives.
c. Requirements: comprise a set of human,
Selecting an ERP for adoption is a complex pro- process, and technology oriented aspects and
cess, because, besides the size of the task, it is activities needed to describe the operational
an important enterprise component that impacts requirements of the enterprise.
the adopter organization in financial and in self- d. Design: models the enterprise entity
knowledge terms. Therefore, it is important to and helps to understand the system
use a framework to understand how open source functionalities.
alternatives can impact this kind of project. e. Implementation: the design is transformed
The Generalized Enterprise Reference Architec- into real components. After tested and ap-
ture and Methodology (GERAM) is a well-known proved the system is released into operation.
standard that provides a description of all elements f. Operation: is the actual use of the system,
recommended in enterprise engineering and a col- and includes user feedback that can drive to
lection of tools and methods to perform enterprise a new entity life cycle.
design and change with success (IFIP– IFAC, 1999), g. Decommission: represents the disposal of
providing a template life cycle to analyze FOS-ERP parts of the whole entity, after its successful
selection, deployment, and evolution. GERAM use.
Figure 1. GERAM life cycle phases. The design phase is subdivided into preliminary and detailed de-
sign.
103
Except for decommission and identification, criteria. For FOS-ERP these criteria and others
which are not influenced by licensing models, related specifically to FOSS must be also taken
these phases can be used to better understand into account– remembering that even if the imple-
how FOS-ERP differs from P-ERP, providing mentation represents a smaller financial impact,
key aspects for evaluating alternatives and suc- in terms of a company’s self-knowledge it can
cessively refining objectives, requirements and assume a much bigger importance, since it holds
models, as next subtopics address. not only a inventory of records and procedures,
but also how those records and procedures are
concept realized in technological form – through source
code.
During this phase, high-level objectives are estab- In other words, a FOS-ERP can have a smaller
lished, such as the acquisition strategy, preliminary financial impact but a much bigger knowledge and
time and cost baselines, and the expected impact of innovation impact. Although P-ERP are also highly
ERP adoption. In the case of FOS-ERP, the level parameterized, and adaptable through APIs and/
of involvement of the adopter in development or dedicated programming languages, the access
can be established. In other words, at this point to the source code in FOS-ERP can drive much
the adopter can start considering the possibility better exploration of the ERP’s capabilities, thus
of actively contributing to an open source project, allowing a better implementation of differenti-
becoming a prosumer. Of course, this decision ated solutions.
will be possible only during the more advanced From this standpoint, the strategic positioning
phases, when the adopter better knows the solution of an adopter in relation to a FOS-ERP seems
requisites and the decision alternatives. to be of greatest importance, given the possibil-
ity of deriving competitive advantage from the
requirements and Preliminary source code. Therefore, the adopter must decide
to behave as a simple consumer, only getting the
design solution from the vendor, or become a prosumer,
by mixing passively purchasing commodity parts
Taking as a principle that most software develop- of the system with actively developing strate-
ment (and customization) today is done through gic ones by itself. Thus it is clear that when an
interactive and incremental life cycles, it can be adopter considers FOS-ERP as an alternative, it
considered that there is no clear borderline between should also consider developing parts of it to fit
the requirements and preliminary design phases its requirements – taking into account that, as said
and between the detailed design and implementa- before, this kind of positioning involves allocating
tion phases, thus they are considered together in managerial and technical resources for develop-
this analysis. ment tasks in a FOSS environment.
The requirements phase deals with system’s
functional and non-functional requirements. detailed design and Implementation
The adopter may model some main business
processes – part of the Preliminary Design – as The detailed design phase focus on refining
a way to check how the alternatives fit to them. models, and is associated to business process
At this point FOS-ERP starts to differ more from modeling and parameter identification and value
P-ERP. Evaluating P-ERP involves comparing definition. The implementation phase concentrates
alternatives under the light of functionality, Total on validating, integrating modules, and releasing
Cost of Ownership (TCO), and technological its modules for initial use.
104
If the adopter decided to participate actively A very interesting point is the openness of
in the selected FOS-ERP project, deeper design parts customized for and sponsored by a specific
decisions are involved, such as creating entire adopter. Maybe the adopter doesn’t want to become
new modules or extending the basic framework. a developer at all – which is most likely to hap-
A consequence of assuming a more active role is pen, but it still wants to keep some tailored parts
to invest more human and financial resources for of the system in secret. In these cases, the vendor
learning the FOS-ERP platform and framework, must adapt the licensing terms for its solution, so
developing and maintaining parts of it, and manag- that general openness of the code is guaranteed,
ing the relationship with the project community. while some client-sponsored customized parts
In that case, customization and maintenance con- can be kept closed5.
tracts must define responsibilities of each part on
the deployment process. For instance, what the operation
vendor should do if the adopter finds a bug in the
original code, written by the first, which is being During the operation phase the resources of the
adapted by the second? What is the priority that entity are managed and controlled so as to carry
the vendor must follow for correcting this bug? out the processes necessary for the entity to fulfill
Actually, is the vendor responsible for correcting its mission. Deviations from goals and objectives
this bug, since for this part the adopter decided or feedbacks from the environment may lead to
to take advantage of the solution’s free license, requests for change; therefore during this phase
therefore exempting the vendor of responsibility system maintenance and evolution occur. During
for the bug? operation the adopter can decide at any moment,
The adopter has the option of assuming dif- unless contractual clauses hinders, to shift to anoth-
ferent grades of involvement for each phase. For er vendor or to assume the system’s maintenance
ordinary modules, like payroll, the adopter can let by itself. Minor changes can also be conducted by
the vendor do the work. However, for strategic the own adopter or even by community individuals
modules, where the adopter believes that it holds that may help on specific matters.
competitive advantage in the related business As a conclusive remark on the differences of
processes, it can take an active role from detailed FOS-ERP on the adopter side, experience has
design to implementation and maintenance, to be shown that most of the times the adopter will not
sure that the business knowledge, or at least the get involved on customization or even maintenance
more precious details that keep the competitive tasks. Still, FOS-ERP can be a good choice, since
advantage, will be kept in the adopter company. it reduces vendor dependency. Moreover, the open-
In that situation the vendor is limited to act as a ness of code on FOS-ERP also makes adapting
kind of advisor to the adopter. One can think that it it to specific needs easier, thus reducing costs in
is possible to keep secrecy on parts of the system customization and further evolution of the software.
by properly contracting a P-ERP vendor, which In other words, the central points to consider are
is true, but the adopter will become dependent cost reduction and freedom of choice. Last but not
of the vendor in a strategic part of the system. least, as a general rule, FOS-ERP also rely on other
Becoming dependent means to wait for other open technologies. For instance, while most P-ERP
vendor’s priorities or pay a high price to become systems export and import data to and from MS-
the priority when changes are needed. Even if Office, FOS-ERP, like ERP5, interact with the also
the P-ERP adopter decides to develop these high- free Open Office. The same is truth for databases
strategic parts, it will have to deal with licensing and operational systems – thus reducing licensing
costs anyway. costs on ERP supportive software too.
105
differences for the vendor open the code of its own system improvements,
following the general FOSS code of ethics6.
The FOS-ERP vendor business models are a Following the common reasoning about FOSS
consequence of the customer’s freedom of choice pricing, FOS-ERP vendors can take advantage
and of the general open source market character- from open source software because, according
istics. Like in other types of FOSS, if on one hand to Riehle (2007), FOSS “increase profits through
vendors benefit from the community improve- direct costs savings and the ability to reach more
ments and testing work, on the other hand they customers through improved pricing flexibility”,
face the competition of this community when as shown on Figure 2.
dealing with deployment and maintenance. In Figure 2 shows a situation that maybe is more
fact, as previously shown, even an adopter can applicable to partners and free-lance vendors of
become a competitor, at some extent of course. FOS-ERP, which can switch from more expensive
It is important to note that there are three types of proprietary software to less expensive open source
vendor: the original system creator, its partners, software, thus potentially increasing their profit
and free-lance vendors. In the case of partners, margin. The creator organization must know how
a formal, most of times contractual, agreement to manage the community around the project,
is set between them and the system creator. This finding prospective partners, hiring individuals
agreement involves some responsibilities to the that become highly productive on the project
partner, in special, following creator’s deployment whenever possible, and even trying to transform
practices, communicating new business generated free-lancers into partners. Like their proprietary
by the system, opening the source code of new counterparts, FOS-ERP needs a network of part-
and improved parts of the system, and helping in ners that can help on deployment projects where
development tasks managed by the creator. Free- the creator has no conditions to be the main con-
lance vendors are free of these obligations, and as tractor, and finding new markets and customers
a consequence, have no special treatment by the for the system. However, gathering contributors
creator, that expects that the free-lancer at least for a starting FOS-ERP project can be a hard task.
Figure 2. Sales margins and number of customers; (a) The lower price limit determines the customers
the system integrator takes on; (b) Switching from closed source software to open source software can
result in more customers and higher profits (Source: Riehle, 2007, with permission)
106
As said before, ERP users are organizations, not knowledge on the system platform and
individuals, and therefore the creator must learn framework, and, as customers to the solu-
how to attract partner firms that are willing to tion appears, more money can be invested
contributing to the project without becoming on a growing commitment to the project.
competitors. As a main conclusion, FOS-ERP Additionally, with the raising of venture
vendors must fight hard to form a community capitalism investment on FOSS startups, a
around the project and to retain customers. This smaller firm can even get financed in a way
seems to be a big difference between the open that would be very unlikely to happen if it
and proprietary licensing models, since the risk worked on top of a P-ERP solution, given the
of the vendor loose a client after deployment is restrictions imposed by the global players.
almost inexistent in the current P-ERP dominated b. Challenges: If on one hand it is easier to
market landscape, where global players dictate enter the market, on the other it is harder to
market rules in practice. retain clients: a broader consultancy basis
The differences between FOS-ERP and P-ERP empowers the demand side, making custom-
can led to a shift from the vendor-dominated ers more demanding.
perspective of P-ERP to a more customer-driven
FOS-ERP perspective. These differences in con- Keeping quality level among a heterogeneous
ducting selection, adoption, and selling also bring network of consulting services providers is also
a series of opportunities and challenges for both a major challenge. FOS-ERP in general lack
vendors and adopters, which are addressed in the certification and quality assurance programs that
following topics. guarantee service levels to clients. Moreover,
FOS-ERP skeptics argue that few reliable con-
sulting firms have experience on implementing
oPPortunItIes And them. But exactly those programs keep smaller
cHAllenGes consulting firms way from P-ERP, pushing them
towards FOS-ERP. For a small consulting firm, a
FOS-ERP offer a series of opportunities for actors possible solution to this deadlock is to start with
that are currently out or ill inserted into the ERP smaller, less demanding projects, and then go
market. These opportunities come together with towards bigger ones, as the deployment processes
a series of challenges, as listed below. and related activities gain maturity. This maturity
For smaller consulting firms: will become the competitive advantage of the firm
on a high competitive FOS-ERP market.
a. Opportunities: P-ERP vendors generally For smaller adopters:
impose high costs and a rigid set of rules
for firms that desire to enter their partner a. Opportunities: lower costs open new oppor-
network, raising the difficulties for smaller tunities for Small and Medium Enterprises
firms to become players in this market. In (SME) to become ERP adopters. With
contrast, smaller consulting firms can enter globalization, small firms suffer more and
the FOS-ERP market in an incremental way, more with competition, and when they try to
increasing their commitment to a project as modernize their processes, they hit the wall
new business opportunities appear and bring of global players’ high costs, or have to adopt
more financial income. In other words, firms smaller off-the-shelf (and also proprietary)
can start contributing with small improve- solutions that ties them to a single supplier
ments to the project as a way of gaining that normally doesn’t have a partner network.
107
In contrast, FOS-ERP are less expensive information on their internal features and
and support can be found in different ways, development processes, on the other hand
including individuals and other SME. it is harder to get information from a dis-
tributed set of partners that sometimes carry
This is also truth for local governments and informal agreements. Social and economical
countries in development in general. FOS-ERP aspects, like reward structures, must be taken
reduce costs, thus helping local governments focus into account to understand the dynamics
on their core business – directly taking care of of FOS-ERP, like in every FOSS, bringing
citizens – and reducing technological dependency more components to be analyzed.
from global players. In fact, FOSS in general is an
opportunity for countries in development to shift For individuals:
from buyers to players in the software industry
(Ouédraogo, 2005). a. Opportunities: FOS-ERP represent an
unique opportunity for an individual to install
b. Challenges: lower costs can also mean that an ERP framework and understand its inter-
adopters have to deal with lower service lev- nals. It is the chance of participating in a big
els, then stressing the necessity of carefully software development project without being
evaluating FOS-ERP options and the matu- an employee of a big company (Spinellis,
rity of their supportive services. Actually, as 2006). Also, the developer can incrementally
said before, consulting certification is yet on gain knowledge of the system, and get free
the early stages for FOS-ERP, thus quality of support from the community, without the
service must be carefully addressed during necessary investment on the P-ERP high
contract negotiation. cost training and certification programs. In
that way an individual can improve his/her
For researchers: employability without investing too much
money on courses, books and certifications.
a. Opportunities: The author has been con- In the future, these advantages can make
tributing to a FOS-ERP project7 since its more free-lance developers enter FOS-ERP
conception. During this time it was possible communities, currently formed mostly by
to know deeply, and sometimes take part of, companies’ employees.
all the process that compose an ERP solu- b. Challenges: learning the internals of a FOSS
tion, from conception and development, to in general means to spend considerable time
business models, deployment, operation in understanding system architecture, design
and maintenance, and evolution. This is a decisions, and specific features. Moreover,
really good opportunity, since most research FOS-ERP in special currently lack books and
papers on ERP are related to deployment courseware in general to help accelerating
and operation, given that P-ERP companies the learning process, and many times the
don’t usually open their projects’ internals individual must count on Web sites, mailing
for researchers. Smaller research groups can lists, discussion forums, and the good will
find their way in this area by getting associ- of community members to acquire deeper
ated to a FOS-ERP project, and contributing knowledge on the framework.
to specific parts of it.
b. Challenges8: If on one hand the openness
of FOS-ERP may give researchers more
108
conclusIon reFerences
In this chapter Free/Open Source ERP particu- Alshawi, S., Themistocleous, M., & Almadani,
larities, opportunities and challenges were briefly R. (2004). Integrating diverse ERP systems: A
presented. It is important to note that this type of case study. The Journal of Enterprise Information
software inherits all advantages and shortcom- Management, 17(6). Emerald Group Publishing
ings of open source software in general and have Limited, pp.454-462.
some more of both. As a matter of fact, FOS-ERP Botta-Genoulaz, V., Millet, P.-A., & Grabot, B. A.
two main advantages are both directly related (2005). Survey on the recent research literature on
to FOSS: lower TCO given the reduced or non- ERP systems. Computers in Industry, 56, 510–522.
existent licensing costs - including of supportive doi:10.1016/j.compind.2005.02.004
software, like spreadsheets, databases, networking
and operational systems; and the possibility of Caulliraux, H. M., Proença, A., & Prado, C. A. S.
having direct access to the source code whenever (2000). ERP Systems from a Strategic Perspec-
is needed. tive. Sixth International Conference on Indus-
Nevertheless, despite the growing interest on trial Engineering and Operations Management,
this subject, it still has many topics to be explored Niteroi, Brazil.
by researchers and practitioners, given the short
Crowston, K., & Howison, J. (2006). Assessing
period of time that passed since this kind of soft-
the Health of Open Source Communities. IEEE
ware appeared in the market and the relatively Computer, May, 89-91.
small number of users, which indicates that the
list of opportunities and challenges aforesaid is De Carvalho, R. A. (2006). Issues on Evaluat-
a reflection of current tendencies that must be ing Free/Open Source ERP Systems. Research
confirmed and better scrutinized as new deploy- and Practical Issues of Enterprise Information
ments occur. For instance, currently there are no Systems, 667-676. Springer-Verlag
research figures on FOS-ERP success rates. In
Dreiling, A., Klaus, H., Rosemann, M., & Wys-
other words, in this arena, it is necessary that more
susek, B. (2005). Open Source Enterprise Systems:
data on deployment, customization, operation, and
Towards a Viable Alternative. 38th Annual Hawaii
evolution become available so that tendencies may
International Conference on System Sciences,
be confirmed and can become facts and figures.
Hawaii.
Hence, as a relatively new kind of software,
FOS-ERP has a potential to be realized, but many Goth, G. (2005). Open Source Business Mod-
questions about it are yet to be answered. Nev- els: Ready for Prime Time. IEEE Software,
ertheless, their growing commercial acceptance (November/December): 98–100. doi:10.1109/
is a fact, and their lower costs, easier adaptation, MS.2005.157
and potentially more competitive supplier market
Herzog, T. (2006). A Comparison of Open Source
can slowly force a shift in the ERP market from
ERP Systems. Master thesis, Vienna University of
the current vendor perspective to a customer
Economics and Business Administration, Vienna,
perspective.
Austria.
IFIP – IFAC Task Force on Architectures for
Enterprise Integration. (1999). GERAM: Gen-
eralized Enterprise Reference Architecture and
Methodology, 31
109
Kim, H., & Boldyreff, C. (2005). Open Source Xu, N. (2003). An Exploratory Study of Open
ERP for SME. Third International Conference on Source Software Based on Public Archives. Master
Manufacturing Research, Cranfield, U.K. Thesis, John Molson School of Business, Concor-
dia University, Montreal, Canada.
Kooch, C. (February 01, 2004). Open-Source
ERP Gains Users; http://www.cio.com/ar- Zinnov Research and Consulting. (2006). Penetra-
chive/020104/tl_open.html tion of Open Source in US Enterprise software
market – An Overview. 37.
LeClaire, J. (December 30, 2006). Open Source,
BI and ERP: The Perfect Match?http://www.
linuxinsider.com/story/LjdZlB0x0j04cM/Open-
Source-BI-and-ERP-The-Perfect-Match.xhtml key terms
Ouédraogo, L.-D. (2005). Policies of United Na- ERP: Enterprise Resources Planning, a kind
tions System Organizations Towards the Use of of software which main goal is to integrate all
Open Source Software (OSS) in the Secretariats. data and processes of an organization into a uni-
Geneva, 43p. fied system.
Riehle, D. (2007). The Economic Motivation of ERP Business Models: Broad range of infor-
Open Source Software: Stakeholder Perspectives. mal and formal models that are used by a vendor
IEEE Computer, 40(4), 25–32. to make profit from an ERP system deployment,
customization, and maintenance.
Serrano, N., & Sarrieri, J. M. (2006). Open ERP Evaluation: Process of selecting an
Source ERPs: A New Alternative for an Old Need. ERP package, among various alternatives and in
IEEE Software, (May/June): 94–97. doi:10.1109/ accordance to business processes, information,
MS.2006.78 technology, and strategic requirements.
Smets-Solanes, J., & De Carvalho, R. A. (2003). Free Software: According to the Free Software
ERP5: A Next-Generation, Open-Source ERP Foundation, is a Software that gives to the user
Architecture. IEEE IT Professional, 5(4), 38–44. the freedom to run the program for any purpose,
doi:10.1109/MITP.2003.1216231 study how the program works and adapt it to his/her
needs, redistribute copies, improve the program,
Spinellis, D. (2006). Open Source and Profes- and release his/her improvements to the public,
sional Advancement. IEEE Software, (September/ so that the whole community benefits.
October): 70–71. doi:10.1109/MS.2006.136 Free/Open Source ERP: ERP systems that
Wang, F.-R. He, D., & Chen, J. (2005). Motiva- are released as Free Software or Open Source
tions of Individuals and Firms Participating in Software.
Open Source Communities. Fourth International Free/Open Source Software Adopter Types:
Conference on Machine Learning and Cybernet- According to Xu (2003) it is possible to classify
ics, 309-314. a software user company in accordance to its
positioning in relation to a FOSS: Consumer: a
West, J., & O’Mahony, S. (2005). Contrasting passive role where the adopter will just use the
Community Building in Sponsored and Com- software as it is, with no intention or capability
munity Founded Open Source Projects. 38th An- of modifying or distributing the codes, Prosumer
nual Hawaii International Conference on System an active role where the adopter will report bugs,
Sciences, Hawaii. submit feature requests, post messages to lists.
A more capable Prosumer will also provide bug
110
fixes, patches, and new features. Profitor: a pas- Movement has also some political connota-
sive role where the adopter will not participate in tions for instance - for the goals of this work
the development process but simply will use the the two terms will be treated as synonyms.
software as a source of profits. Partner: an active 2
The IFIP First International Workshop on
role where the adopter will actively participate in Free/Open Source Enterprise Information
the whole open source development process for Systems/ERP, held during the First IFIP TC8
the purpose of earning profits. International Conference on Research and
Open Source Software: According to the Practical Issues of Enterprise Information
Open Source Initiative, licenses must meet ten Systems – CONFENIS 2006.
conditions in order to be considered open source 3
Despite the fact that these methods hold some
licenses: 1. The software can be freely given similarities, they were developed without
away or sold. 2. The source code must either be having knowledge of each other and were
included or freely obtainable. 3. Redistribution published in a very close range of time:
of modifications must be allowed. 4. Licenses Carvalho’s method was published in April
may require that modifications be redistributed and Herzog’s in June, 2006.
only as patches. 5. No Discrimination Against 4
This is a generic assumption, since in practice
Persons or Groups. 6. No Discrimination Against the vendor can impose specific license terms
Fields of Endeavor. 7. The rights attached to the that keep the software open, but constrain
program must apply to all to whom the program deployment to terms that keep vendor control
is redistributed without the need for execution on it.
of an additional license by those parties. 8. The 5
Although this seems to be nonsense in FOSS
program cannot be licensed only as part of a larger terms, it is a common real life situation in
distribution. 9. The license cannot insist that any FOS-ERP. In fact, the author knows a case
other software it is distributed with must also be where an adopter company sponsored the
open source. 10. License Must Be Technology- whole development of an FOS-ERP dur-
Neutral The official definition of Open Source ing a three-year period, without becoming
Software is very close to the definition of Free a prosumer, and keeping only a specific
Software, however, it allows in practice more algorithm, related to its product pricing
restrictive licenses, creating a category of “semi- schedule, in secret. The original license had
free” software. to be changed to fit this customer demand.
6
In practice, this return from free-lancers
doesn’t happen all the times.
endnotes 7
ERP5 http://www.erp5.com
8
The author considers that these challenges
1
The precise definitions of Free Software and represent, in fact, new research opportuni-
Open Source Software are on the chapter’s ties.
Key Terms list. Although there are differ-
ences between them – the Free Software
111
112
Chapter 1.8
E-Government and ERP:
Challenges and Strategies
Gita A. Kumta
SVKM’s NMIMS University, School of Business Management, Mumbai, India
AbstrAct expects that researchers, governments and solution

providers will be able to appreciate the underlying
The chapter introduces the essence of ERP in constraints and issues in implementation of ERP
government as a tool for integration of govern- and hopes that the learning from industry would be
ment functions which provides the basis for citi- useful to plan implementation of ES in government
zen services. It discusses the challenges faced in using emerging technologies.
modernization of government “businesses” and
discusses strategies for implementation. The basis
of Enterprise Resource Planning (ERP) solutions IntroductIon
is integration of functions which capture basic data
through transactions to support critical adminis- ERP provides an enterprise-wide view of an organi-
trative functions such as budgeting and financial zation and integrates various silos of activity. Such
management, revenue management, supply chain an integrated approach has a tremendous payback
management and human resources management. if implemented properly. Most ERP systems were
Today, Enterprise solutions (ES) go beyond ERP to designed to be used by manufacturing companies
automate citizen-facing processes. The integration to track men, machines and material so as to im-
of data sources with each contact point is essential prove productivity and reduce inventory. Viewing it
to ensure a consistent level of service. The author from a business perspective, ERP systems are now
known as Enterprise Solutions (ES) which takes a
DOI: 10.4018/978-1-59904-859-8.ch025 customer order and provides a software road map
E-Government and ERP
for automating the different steps along the path use of Information & Communication Technology
to fulfilling the order. The major reasons why (ICT) in the right perspective.
companies look at ES can be summarized as:
• Most governments have not changed their
• Integrate financial information processes in any way, and instead have au-
• Integrate customer order information tomated flawed processes.
• Standardize and speed up operational • Government budgets and administration
processes tends to be in departmental silos, but e-
• Reduce inventory governance cuts across departments.
• Standardize HR information • Too much attention to “citizen portals” has
taken attention away from internal gov-
Governments worldwide have been making ernment functioning. There is a big gap
efforts to use information and communications between a web site and integrated service
technologies (ICT) as an instrument of change to delivery.
provide better services to citizens, facilitate work • Governments often underestimate the se-
flow, and provide better governance and transpar- curity, infrastructure, and scalability re-
ency. Popularly known as E-Government, the focus quirements of their applications which im-
has initially been on information dissemination pact the quality of service. (Khalil, Lanvin,
which has now moved on to transactions. What Chaudhry, 2002)
is required is a transformation of the public ad-
ministration which takes a citizen service request Learning from the experiences of the corporates,
and provides a software road map for automating governments today understand the need for a con-
the different steps along the path to fulfilling the sistent and flexible information infrastructure that
request. This cuts across various departments can support organizational change, cost-effective
and it is therefore critical to lay down suitable service delivery and regulatory compliance. ERP is
policies, guidelines and specifications and also therefore needed to meet organizational objectives
redefine processes to facilitate faster proliferation and outcomes by better allocating resources - its
of ICT applications. people, finances, capital, materials, and facilities.
E-government does not happen with more com- Modernization programs however involve a broad
puters or a website. While online service delivery range of activities and require a wide array of skills
can be more efficient and less costly than other and experiences, as these programs affect every-
channels, cost savings and service improvements thing from computers to culture. The objective is
are not automatic. E-government has therefore to reduce administrative overhead and improve
to focus on planning, sustained allocation of core product/service delivery.
budgets, dedication of manpower resources and
above all, the political will. The e-government
field, like most young fields, lacks a strong body essence oF erP In Government
of well-developed theory. One strategy for cop-
ing with theoretical immaturity is to import and Before moving on to the ERP discussion it is
adapt theories from other, more mature fields. necessary to dwell a little on various aspects of
(Flak, Rose, 2005) E-government which is about transforming the
Literature survey on implementations of way government interacts with the governed.
e-governance has brought out the following ob- The E-Government handbook for developing
servations which would help us in redefining the countries identifies three major phases –Publish,
113
Interact and Transact. These however are not government services, available at any time.
sequential phases and hence can be considered Some of the cases are
as major aspects of e-government. In short, e- ◦ The Bhoomi Project. Delivery of land
government utilizes technology to accomplish titles online in Karnataka, India...
reform by fostering transparency, eliminating http://www.revdept-01.kar.nic.in/
distance and other divides, and empowering people Bhoomi/Home.htm
to participate in the political processes that affect ◦ The Government E-Procurement
their lives. (Khalil, Lanvin, Chaudhry, 2002). System in Chile...http://www.com-
Each of the phases is briefly summarized below. praschile.cl/Publico/entrada_publico.
asp
• Publish: Governments generate and also
publish in print large amount of informa- To achieve this it is essential to look at the
tion which can be disseminated to the pub- internal processes of the government, the rela-
lic using ICT. Some of these cases are tionships between various departments, sharing
◦ E-Government Portal of Canada is of information between departments and the IT
considered as one of the best gov- infrastructure required to support these aspects.
ernment portals in the world. http:// The following section indicates the characteris-
www.canada.gc.ca. tics of content, process, people and technology
◦ The JUDIS (Judgment Information required by government to implement the various
System) in India posts court records, initiatives, thereby identifying the essence of ERP
case information and judicial deci- in government (refer to Table 1).
sions. http://indiancourts.nic.in/itin- The E-Government handbook for develop-
jud.htm ing countries has many more case studies listed
• Interact: Interactive e-government in- which highlight these phases. (Khalil, Lanvin,
volves two-way communications, starting Chaudhry, 2002).
with basic functions like email contact Enterprise modernization is therefore a com-
information for government officials or plex, ongoing evolutionary process that involves
feedback forms that allow users to submit the integrated transformation of strategies,
comments on legislative or policy propos- policies, organization and governance structures,
als. Some of the cases are business processes and systems, and underlying
◦ Citizen Space. A section of the British technologies. Only by aligning these elements
Government’s web portal allowing with its business goals can an agency achieve
citizens to comment on government a successful modernization program. (Kirwan,
policy. http://www.ukonline.gov.uk Sawyer, and Sparrow, 2003).
◦ The Central Vigilance Commission The very essence of enterprise modernization
in India allows citizens to file online is integration of processes and connectivity of
complaints about corruption. http:// stakeholders. It therefore involves connecting:
www.cvc.nic.in/vscvc/htm
• Transact: Just as the private sector in de- • Government to government
veloping countries is beginning to make (G2G):Departmental integration of pro-
use of the Internet to offer e-commerce cesses ERP
services, governments will be expect- • Government to businesses (G2B):
ed to do the same with their services. A Information to suppliers and procurement
transact website will offer a direct link to SCM
114
Table 1. Requirements for e-government initiatives
Phase Content Process People Technology

Publish Existing documents Simple process Small project team with Basic IT infrastructure,
Focus: centralization in terms of rules and of capture and skill focus on IT. storage and web services –
of content regulations, documents, monitoring. Involvement of depart- Portal.
and forms ment staff minimal. Batch mode.
Minimal technical Can be totally outsourced.
support for usage.
Interact Grievances, Process of coordina- Small internal teams E-mail facility and
Focus: suggestions, feedback tion with departments with skill focus on com- collaborative systems.
Involvement of and communication on munication and people On-line and batch mode
citizens status of content. management. Can be partially outsourced.
Moderate technical sup-
port for usage.
Transact Integration of functional Process of data capture Large Project teams. Large and robust IT infra-
Focus: Direct link of processes through transaction, Skills in communication, structure. Storage, internal
citizens to govern- Service flow from validation, processing people management. And network, ERP solution, web
ment services. application to service using defined logic. functional knowledge. services and e-commerce.
delivery. Focus on managing Extensive technical Online, real-time mode.
Covers forms, logic of business processes. support for usage. Cannot be outsourced. Re-
computation, controls, quires technology
operational and legal partnerships ICT policies.
policies.
• Government to citizens (G2C): informa- citizens which no longer need to be restricted to

tion and service to citizens CRM the structure of the government.
• Government to employees (G2E): infor- Tremendous potential exists for rethinking
mation and service to employees ERP of the business of government to reduce cost
and improve the quality of government/constitu-
As citizens grow in awareness, governments ent interactions. Most so-called “e-governance”
today are under increasing pressure to deliver a initiatives have been simply focused on internet-
range of services – from ration cards, motor li- enabling old processes and systems which have
censes and land records to health, education and resulted in a series of costly, overlapping, and
municipal services – in a manner that is timely, uncoordinated projects. Transformation is possible
efficient, economical, equitable, transparent and only when one examines the inter-relationships
corruption-free. For any government that is keen between government agencies – both processes
to respond to this demand and to hasten the pace and systems which would result in true efficien-
of development –information technology comes as cies of E-Business.
an excellent tool. However technocratic responses “The benefits come from changing your busi-
in themselves are not a solution. They are a tool. A ness processes, not from installing ERP,” says
solution is one that is holistic and describes how Bill Swanton, a vice president at AMR Research
the tool can be feasibly deployed. (Carter, 2005). in Boston. Adds Buzz Adams, president of Peak
ERP solutions in government should therefore Value Consulting that specializes in process im-
provide horizontal components, which are relevant provement, “The technology will work the way
to every part of a government and should support you implement it, so what’s important is how you
vertical integration needed for the delivery of improve the processes — the way you do things.”
specific services. Such solutions would facilitate (Bartholomew, 2004).
governments to offer cross-functional services to
115
FunctIonAl model oF registration, tax assessment, management of online

PublIc AdmInIstrAtIon or offline revenue collection and reimbursements,
secure processing, and payment. Tax case man-
Enterprise resource planning (ERP) systems agement involves investigation and enforcement,
are becoming pervasive throughout the public revenue auditing, and analysis.
sector for their support of critical administrative Supply chain management (SCM) functions
functions such as budgeting and financial man- cover cataloguing, bidding and negotiation, and
agement, procurement, human resources (HR), tendering with workflow support for prepara-
business process, and customer relationship tion of tender requests, invoicing and payment,
management (CRM). purchasing and reporting. Logistics support in
ERP solutions for financial management cover the areas of material and equipment planning,
accounts payable, accounts receivable, fixed asset inventory, warehouse management, transportation
accounting, cash management, activity-based and and distribution, and the maintenance, repair, and
project accounting, cost-centre planning and analy- disposal of assets is another major area of public
sis, financial consolidation, financial reporting, and administration.
legal compliance and reporting. Budget management Human resource management covers civil
solutions support economic policy, collection and service recruitment, individual and collective
analysis of budget and economic data, economic training, performance assessment, career devel-
forecasting, revenue allocation to government opment, payroll and benefits, and travel and the
agencies, budget approval, and budget monitoring. movement of public employees.
The Public Administration Functional Model Finally citizen-facing processes within an or-
can be diagrammatically depicted as shown in ganization through Web-portals and call centers
Figure 1. as well as traditional over-the-counter services
Tax and revenue management also fall under require integration of data sources with each con-
the financial and budget management umbrella. tact point to ensure a consistent level of service.
This includes both tax filing and tax case manage- Traditionally processes were people driven and
ment. Tax filing covers taxpayer identification and the emphasis of ERP was on automation. Having
Figure 1. Public Administration Functional Model
116
achieved the basic requirements, organizations port, the challenging task is creating an IT orga-
expected to achieve efficiency and control. Today nization structure to support this process within
the expectation from an ERP system is strategic the government structure.
support adding value to business. Collectively, all
the functionalities of publishing, interacting and High expectation from citizens
transacting now need to be aligned to manage
the entire government business. ERP combines Citizens themselves are becoming increasingly
them all together into a single, integrated software demanding as they compare government services
solution that operates on a single database so that with other services like banking. Citizens do
the various departments can more easily share not want to be transferred from department to
information, communicate with each other and department for an answer to a simple telephone
provide end-to-end service to its stakeholders. query and be forced to queue at public offices to
make a straightforward transaction. They demand
facilities to be offered round the clock. They ex-
cHAllenGes In ImPlementAtIon pect high standards of service, instant access to
oF Ict solutIons information, efficient transactions, and support,
whenever and wherever they need it.
E-governance initiatives are common in most
countries as they promise a more citizen-centric Integration of various Functions
government and reduce operational cost. Unfortu-
nately most of these initiatives have not been able Today it is the era of packaged software which in
to achieve the benefits claimed. Often the reason most cases is exhaustively comprehensive. The or-
for this failure is a techno-centric focus rather ganization therefore needs to change its processes
than a governance-centric focus (Saxena, 2005). to align it to ‘best practices’ incorporated in the
The following challenges faced by organizations software package. However in the excitement to
are equally applicable to governments. implement such solutions one often loses sight of
the fact that each department has its own policies
Acquisition Process and rules that make it unique. “For integration
to work, your internal systems must be working
All government ICT and modernization programs properly,” says Mary Kay DeVillier, director/c-
face major challenges, one of which is the acquisi- business and information resources at Albemarle
tion process. This process requires a holistic view (Mullin, 2000).
of the public administration for identifying and
evaluating the right solution, taking decisions requirement of clean data
regarding outsourcing and managing the whole
implementation as a project. It’s not very difficult Organizations embarking on an ERP initiative
to buy software and hardware that fulfill the needs must take care not to underestimate the amount
of a single division without considering the needs of work needed to develop a clean set of master
of the entire organization. However, if each divi- data. The chart of accounts, citizen data, policies
sion of an organization develops its own business & norms and other mission-critical information
processes and IT infrastructure, the result will be has to be accurate from the start or mistakes will
lack of interoperability, duplicated components, multiply exponentially throughout the system.
functional gaps, and inability to share information. Many an ERP project has been scuttled or gone
Though external consultants do provide sup- south altogether because companies failed to do
117
this kind of basic blocking and tackling early on the government to use the software to improve
(Bartholomew, 2004). the ways they do their jobs is by far the most
critical challenge. Most ERP implementations
standards for Interoperability fail due to resistance to change and poor project
management (Koch). The whole implementation
In order to support these demands the internal has to be taken up as a project clearly identifying
functions of supply chain management assume the scope, cost, time and resource requirements
greater importance. There is tremendous impor- with clear definition of milestones.
tance for rules and procedures in the government.
Unless the records are kept properly, accessing Hidden costs
information and tracing the precedents becomes
time consuming and this is one of the reasons Certain costs which are more commonly over-
for the delays in government administration. The looked or underestimated than others are training,
rules and procedures can be made transparent to integration, customization, testing, data conver-
the citizens, and traceability can be incorporated sion and data analysis which are not adequately
which would improve the pace of effectiveness factored into the budgets. This leads to delay in
of governance by using Information Technology implementation which results in apathy and dis-
(Budhiraja, 2003). satisfaction.
With multiple players and departments increas-
ingly becoming involved in the e-Government
initiatives, standards for e-Government have ImPlementAtIon strAteGIes
become an urgent imperative for interoperabil-
ity. Enterprise solutions would therefore help An ERP system helps the different parts of the
in streamlining operations, be flexible and agile organization share data and knowledge, reduce
enough to respond to the demands of public sec- costs, and improve management of business
tor reforms. processes. In spite of their benefits, many ERP
systems fail (Stratman and Roth, 1999). Many ERP
Information security systems face implementation difficulties because
of workers’ resistance (Aladwani, 2001). Effec-
Managing a secure environment in an era of in- tive implementation of ERP requires establishing
tegrated and seamless service delivery presents five core competencies, among which is the use
an increasing challenge for governments as it can of change management strategies to promote the
be overwhelming and costly without the right infusion of ERP in the workplace (Al-Mashari
infrastructure. While processing sensitive data and Zairi, 2000).
such as citizen and financial information stored During the last few years there have been major
on ERP systems, government organizations want initiatives among different Governments towards
to ensure they take every possible measure to ushering in ICT and its tools in the functioning of
maximize security. Government. The emphasis has been on providing
better services to citizens and in improving the
Project management internal productivity. Cases listed in the Report for
the President’s Management Council in December
To get the most from the software, the people inside 2005 is testimony to this fact (Evans, 2005).
the government have to adopt the work methods The strategies get more focused when we view
outlined in the software. Getting people inside the organization as an open system composed of
118
interdependent components that include Strategy, for the implementation was that all state agencies
Processes, Structure, Technology and Culture as had a say in the project from its earliest stages.
mapped by Scott-Morton in the Management (Douglas, 2002).
of the 1990s equilibrium model (Scott-Morton, Developing a strategic plan and bringing all
1991). This model is depicted in Figure 2. The stakeholders on board are crucial early steps in
salient aspect of this open systems model is that implementing ERP”, said Ken Munson, senior
an impact of change on one component is im- principal in the State and Local Solutions Division
mediately felt on the other components either of AMS Inc., Missouri’s ERP vendor. “The plan
directly or indirectly. has to be well communicated and well bought in,
The ERP implementation strategies can there- not just by the different branches of government,
fore be classified into organizational, technical, but by all levels of each of the branches,” he said.
and people strategies. Organizational strategies (Douglas, 2002).
would cover strategic planning, business align- The Department of Information Technology,
ment, project management and change manage- Government of India, has felt it necessary to create
ment. The technology strategies would cover a rational framework for assessing e-Governance
enterprise architecture, business process mapping, projects on various dimensions. It is desirable that a
data capture and information security. People set of instruments is available to the administrators
strategies would cover communication, managing of those projects to appreciate the various attributes
resistance to change and ongoing support. of a good e-governance project, apply midcourse
corrections, where needed, and steer these projects
strategic Planning in the right direction.(Rao, etal, 2004).
Many problems related to ERP implementa-
The state of Missouri in USA was the first to tion are related to a misfit of the system with the
implement one of the largest government ERP characteristics of the organization. (Markus et
systems which was implemented in phases and al., 2000). ERP ‘tends to impose its own logic
operational by 2001. The finance, budgeting and on a company’s strategy, culture, and organiza-
purchasing functions serve 6,000 end users, and tion’ which may or may not fit with the existing
its human resources and payroll modules serve organizational arrangements. (Davenport, 1998),
9,000. One of the most important keys to success
Figure 2. Management of the 1990s Equilibrium Model
119
effective Project management tional change process, rather than as the replace-
ment of a piece of technology. (Boonstra, 2005).
Success depends on a well designed implemen- Though change management strategies facilitate
tation plan. An ERP system is the information the success of ERP implementation, many ERP
backbone and reaches into all areas of the busi- systems still face resistance, and ultimately, failure.
ness and value-chain. It is necessary to consider (Aladwani, 2001).
implementation as a project and plan the activities For the reform to succeed, governments need
involving people, processes and technology. to be culturally and technically prepared to un-
SAP’s Accelerated SAP Implementation Meth- derstand and implement change. Procedural and
odology provides a guideline for implementation. legal changes in the decision making and delivery
(Dejpongsarn, 2005). It focuses on five broad processes as well as internal functioning are re-
phases covering quired to make a success of ERP implementation.
• Project Preparation: This covers the re- business Process mapping

quirement of resources- people, technol-
ogy and budget Administrative reforms will have to precede at-
• Business Blueprint: This covers the docu- tempts at implementation of an ERP. The emphasis
mentation of current processes (As is) and will have to be on simplifying procedures, ratio-
comparison to the solution which has best nalizing processes and restructuring Government
practices. A gap analysis provides a start- functions. However, this is an ideal situation but
ing point to identify the changes required in practice changes and ERP implementation go
by the departments and new development. on parallel as one cannot expect reforms to take
• Realization: Set-up (configuration), test- place in a short period.
ing, data migration and development The whole government reform agenda has a
• Final Preparation: Training, final function- profound effect on government financial manage-
al testing ment. A shift of emphasis from inputs to outcomes
• Go-live & Support: Different approaches is a key driver behind many reform initiatives.
are followed for going live (all modules / (Microsoft). These include public / private part-
related modules), support through internal nerships, contracting, and decentralization of
or external teams. service delivery to semi-autonomous agencies,
competition between service providers, cost-
management of change recovery, revenue generation, and the many other
alternatives for service provision.
Governments are using organizational change
management techniques that have worked in the data capture
commercial world within their own moderniza-
tion programs. (Kauzlarich, 2003). The basis of an ERP system is the data captured
Modernization strategies, such as ERP imple- through transactions which integrates the business
mentation, commonly involve change. Hence, processes to achieve a task. It is therefore necessary
responsiveness to internal customers is critical to have data standards for use across government
for an organization to avoid the difficulties as- to enable easier, more efficient exchanging and
sociated with this change (Al-Mashari and Zairi, processing of data. It will also remove ambigui-
2000; Aladwani, 1999; Aladwani, 1998). ERP ties and inconsistencies in the use of data. The
implementation should be viewed as an organiza- United States Government is one of the largest
120
users and acquirers of data, information and sup- aligned with the Lines of Business) toward
porting technology systems in the world. The E- common ground and common language to
Government program continuously identifies IT facilitate improved information sharing
opportunities for collaboration and consolidation • Provides guidance for implementing
using the Federal Enterprise Architecture (FEA) repeatable processes for sharing data
framework. The framework is a comprehensive Government-wide
business-driven blueprint to enable the federal
government to identify opportunities to leverage enterprise Architecture
technology to:
If each division of an organization develops its
• Reduce redundancy; own business processes and IT infrastructure,
• Facilitate horizontal (cross-federal) and the end result may be lack of interoperability,
vertical (federal, state and local) informa- duplicated Enterprise architecture components,
tion sharing; functional gaps, and inability to share information.
• Establish a direct relationship between IT is required. (Tucker and Debrosse, 2003).
and mission/program performance to sup- To avoid these problems, the federal govern-
port citizen-centered, customer-focused ment now mandates the use of enterprise archi-
government; and tectures (EAs) by federal agencies seeking to
• Maximize IT investments to better achieve obtain funding for any significant IT investment.
mission outcomes. Enterprise architectures act as a kind of roadmap
for the design, development, and acquisition of
The FEA framework and its five support- complex, mission-oriented information systems.
ing reference models (Performance, Business, Creating enterprise architecture requires partici-
Service, Technical and Data) are now used by pation from many areas of the organization and
departments and agencies in developing their a great deal of communication to plan and imple-
budgets and setting strategic goals. Data forms ment each stage of the process. The result is a
the base of a transactional system and hence data roadmap that guides an organization through the
integrity is essential for inter-operability. The FEA modernization process and enables it to achieve
Data Reference Model (DRM) defined by the Of- its goals (Tucker and Debrosse, 2003).
fice of Management and Budget, US, provides a
“common language” for diverse agencies to use managing People and
while communicating with each other and with communication
state and local governments seeking to collaborate
on common solutions and sharing information for Managing the implementation requires a dedi-
improved services (Evans, 2005). cated team with adequate skills and a clear plan
The potential uses of the model have been and agenda. It needs to be managed as any large
summarized as: project with clear milestones and deliverables.
The migration to an enterprise resource planning
• Provides a FEA mechanism for identifying (ERP) system is often fraught with peril. Re-
what data the Federal government has and search and experience show that communication
how it can be shared in response to a busi- is a key mechanism for breaking down barriers
ness/mission requirement to change. Employees are better able to tolerate
• Defines a frame of reference to facilitate change if they understand why the change is
Communities of Interest (which will be important and if they feel the changes are being
121
handled with fairness and transparency. Good birth to a new field of knowledge called Citizen
communication throughout the enterprise builds Relationship Management. (Kannabiran, Xavier,
trust and understanding. People must see how the Anantharaaj, 2005). It is about making better use
changes will affect the organization, the citizens of the considerable amounts of information that
and themselves (Kauzlarich, 2003). The success government already collects (Smith, 2003). CzRM
of implementation of ICT projects depends on is about becoming “citizen-centric” (Nowlan,
attitudinal readiness of the beneficiaries for accept- 2001; Hunter & Shine, 2001).
ing change (Bowonder, Mastakar, Sharma, 2005). A citizen can be defined as a consumer of
public goods and services (Nowlan, 2001). In the
knowledge management emerging e-Governance scenario, citizens should
be treated as customers of business organizations,
Knowledge Management is a combination of where serving citizens is the sole purpose of
culture and technology. Culture drives knowl- governments. Citizen Relationship Management
edge management while technology enables it. (CzRM) is a division of customer relationship
The following characteristics of government based management that focuses specifically on how
on its structure and functions drive its knowledge governmental bodies relate to their constituents
management needs. (Xavier, 2002; Jha & Bokad, 2003).
Post-ERP trends are geared towards increasing
• Knowledge which is actionable informa- customer relationships and analysis of the market
tion (also known as knowledge assets) place for maximum profit. Data mining and Busi-
is a central resource of the government. ness Intelligence now are not only supporters of
Effective functioning of government rests growth but also act as initiators of growth. Today’s
on effective acquisition and dissemination organizations are looking for applications to in-
of knowledge. spire and lead the way. IT is no more an internal
• Similar knowledge requirements spread system of automation but is now an external means
across the states, districts, and other local of customer communication and market analysis
governments. (Tucker and Debrosse, 2003).
• Transfer of people across government de- The services strategy entails building an integra-
partments calls for a repository of knowl- tion layer that is separate and distinct from any of
edge which can be used wherever they the software applications, including ERP. Services
move. extract pieces of data and business logic from systems
• Proactive action is required if governments and databases and bundle them together into units that
want to transform themselves into “an- are expressed in ‘business’ terms. Implementing a
ticipatory governments” to meet the chal- service-oriented architecture can involve developing
lenges of the emerging E-governance era applications that use services, making applications
(Misra, Hariharan, Khaneja, 2003). available as services so that other applications can
use those services, or both. (Ort, 2005).
emerGInG trends Service-oriented architecture (SOA) is the
emerging trend in enterprise computing because
While ERP is concerned with the use of IT for it holds promise of IT becoming more agile in
efficient functioning of government depart- responding to changing business needs. Gartner
ments, attempts are on to morph the customer reports that “By 2008, SOA will be a prevailing
relationship management concepts for creating software engineering practice, ending the 40-year
effective service for the citizens which has given domination of monolithic software architecture”
122
A service-oriented architecture is an information Government agencies have the advantage of

technology approach or strategy in which applica- applying lessons learned from the commercial
tions make use of services available in a network business world, based on the analysis of both
such as the World Wide Web. failures and successes of ERP implementations.
The Internet aids good governance by increas- Governments today understand the need for a
ing transparency and customer-oriented service consistent and flexible information infrastructure
delivery. (Torres, Pina, Acerete, 2006) .By taking that can support organizational change and meet
advantage of Internet protocols and technolo- regulatory compliance. However, one cannot ex-
gies, one can minimize the need for ad hoc links pect to revolutionize the government operations
between your company and your service sup- with ERP as it affects mostly the existing “back
plier. Moreover, because the new service-based office” processes. This helps in optimizing the
applications run inside Web browsers, your staff way things are done internally which is essential
can connect to the service provider from various for building relationships with citizens, suppliers
locations. Another bonus of Web-based ERP ap- or partners.
plications is that the server location is transparent. If ERP is the focus of an effort to bring dramatic
(Apicella, 2000). improvements to the way government functions, it
will bring with it some post-ERP depression too.
The most common reason for the performance
conclusIon problems is that people see a change in the way
processes were executed earlier. When people
In the ultimate analysis, we find that the electronic can’t do their jobs in the familiar way and haven’t
governance wave has started worldwide. With the yet mastered the new way, they panic, and the
technologies to implement electronic governance operations go into spasms.
already available/understood by the government, A government that plans to implement ERP
managerial issues are of key importance. Change in should expect “to be very committed, because
the mindset of the people particularly at the top levels it’s a long process,” advised Jan Heckemeyer,
in the bureaucracy and policy making is important administrator of Statewide Advantage for Missouri
because it is they who provide the leadership. (SAM) II. It is especially important to involve
It is important to think beyond automation, every agency and resolve policy questions before
and towards redesign of the basic workflow configuring the system. “It drags out the process a
within the government. The next generation of bit, trying to build consensus and work out those
e- Governance will not be mere automation, but issues,” she said. “But it’s time very well spent.”
will require a reengineering of the government (Douglas, 2002).
structurally and functionally.
Enterprise modernization involves changes
to all dimensions of an organization. It affects. reFerences
(Kirwan, Sawyer, Sparrow, 2003).
Al-Mashari, M., & Zairi, M. (2000). Informa-
• Organizational structure tion and business process equality: The case of
• Policies, processes, and procedures SAP R/3 implementation. Electronic Journal on
• Business and technical architectures Information Systems in Developing Countries,
• Investment management practices 2 (http://www.unimas.my/fit/roger/EJISDC/
• Governance EJISDC.htm)
• Culture
123
Aladwani, A. (1998). Coping with users resistance Davenport, T. H. (1998). Putting the Enterprise
to new technology implementation: an interdisci- into the Enterprise System. Harvard Business
plinary perspective. Proceedings of the 9th IRMA Review, 76(4), 121–132.
Conference, Boston, MA, 17-20 May, pp. 54-9.
Dejpongsarn, N. (2005). ERP Framework with
Aladwani, A. (1999). Implications of some of the mySAP Solution. A presentation.
recent improvement philosophies for the manage-
Douglas, M. (2002). Planning for the Enterprise.
ment of the information systems organization.
New York, (April): 16.
Industrial Management & Data Systems, 99(1),
33–39. doi:10.1108/02635579910249594 Evans, K. S. (2005). Expanding E-Government:
Improved Service Delivery for the American
Aladwani,A. M. (2001). Change management strat-
People Using Information Technology. A Report
egies for successful ERP implementation. [Techni-
for the President’s Management Council Dec 2005
cal paper]. Business Process Management Journal,
7(3), 2001. doi:10.1108/14637150110392764 Flak, L. S., & Rose, J. (2005). Stakeholder
Governance: Adapting Stakeholder Theory To
Apicella, M. (2000). The hand that moves your
E-Government. Communications of AIS, 2005,
business - ERP software moves to the Web, pre-
16, 642-664, 23
senting both pitfalls and opportunities.(enterprise
resource planning) . InfoWorld, (June): 26. Gartner’s Four Phases of E-Government Model.
(2001) USA, Gartner Group. Europa (2001) “E-
Bartholomew, D. (2004). The ABC’s of ERP. CFO
government - Electronic access to public services.
IT, October 5, 2004. http://www.cfo.com/article.
(online)(cited on 20th December, 2003). Available
cfm/3171508/c_2984786?f=Technology_top-
from <URL: http://europa.eu.int/>
stories
Hunter, D. R., & Shine, S. (2001). Customer Re-
Boonstra, A. (2005). Information Systems
lationship Management- A Blueprint for Govern-
as Redistributors of Power Interpreting an
ment. White Paper, Australia, Accenture.
ERP implementation from a stakeholder per-
spective. som.eldoc.ub.rug.nl/FILES/reports/ Jha, B., & Bokad, P. (2003).Managing Multiplicity
themeA/2005/05A06/05A06.pdf of Citizens’ identity -A Taluka level case study.
International Conference on E-Governance,
Bowonder, B., Mastakar, N.,& Sharma, K. J.
1(5), 24-31
(2005). Innovatiive ICT platforms: The Indian
Experience. International Journal of Services Kannabiran, G. Xavier, M. J., & Anantharaaj, A.
Technology & Management, 6(3/4/5), 1-1. (2005). Enabling E-Governance Through Citizen
Relationship Management-Concept, Model And
Budhiraja, R. (2003). Electronic Governance – A
Applications. Journal of Services Research, 4(2)
key issue in the 21st century. A paper by Additional
(October 2004 - March 2005).
Director Electronic Governance Division Ministry
of Information Technology Govt. of India). Kauzlarich, V. (2003). Organizational Change
Management is Key to Program’s Success. Enter-
Carter, M. (2004). Key note address on E-gover-
prise Modernization Issue, fall 2003, 7(2).
nance – Transforming India. National summit–
India: The Knowledge Capital February 17, 2004
124
Khalil, M. A., Lanvin, B. D., & Chaudhry, V. Rao, T. P., & Rama, R. Venkata. V., Bhatnagar, S.
(2002). The E-government Handbook for De- C., & Satyanarayana, J. (2004). “E-Governance
veloping countries. infoDev Program The World Assessment Frameworks (EAF Version 2.0)” Re-
Bank Group. port for Department of Information Technology,
Government of India May 2004
Kirwan, K., Sawyer, D., & Sparrow, D. (2003).
Transforming Government Through Enterprise AMR Research, (2002). The Multibillion-Dollar
Modernization. Enterprise Modernization Issue, Enterprise Performance Planning Market,16
Fall 2003 7(2). August 2002.
Koch, C. (n/a). ABC: An Introduction to ERP Saxena, K. B. C. (2005). Towards excellence in
Getting started with Enterprise Resource Planning e-governance. International Journal of Public
(ERP). http://www.cio.com/article/40323/ABC_ Sector Management, 18(6), 498-513, 16
An_Introduction_to_ERP
Scott-Morton, M. (Ed.). (1991). The corporation
Markus, M. L., Axline, S., Petrie, D., & Tanis, C. of the 90s. Information technology and organiza-
(2000). Learning from adopters’ experiences with tional transformation. Oxford: Oxford University
ERP: Problems encountered and success achieved. Press.
Journal of Information Technology, 15, 245–265.
Smith, A. Opinion (2003). Citizen Relationship
doi:10.1080/02683960010008944
Management. (online) (cited June 12, 2003). Avail-
Microsoft. Enterprise Resource Planning- Manag- able from <URL:http://www.crm-forum.com>.
ing the Lifecycle of Government Business. lead
Stratman, J., & Roth, A. (1999), Enterprise
feature from the Tourism and Travel edition of
resource planning competence: A model, propo-
Microsoft in Government, Worldwide.
sitions and pre-test, design-stage scale develop-
Misra, D. C., Hariharan, R., & Khaneja, M. (2003, ment. 30th DSI Proceedings, 20-23 November,
March). E-Knowledge Management Framework pp. 1199-201.
For Government Organizations . Information
Torres, L., Pina, V., & Acerete, B. (2006, Apr).
Systems Management, 20(2), 38–48. doi:10.120
E-Governance Developments in European
1/1078/43204.20.2.20030301/41469.7
Union Cities: Reshaping Government’s Rela-
Mullin, R. (2000). ERP-2-ERP: “Forging a Pro- tionship with Citizens. Governance, 19(2), 277.
prietary Link” (Brief Article) . Chemical Week, doi:10.1111/j.1468-0491.2006.00315.x
(June): 28.
Tucker, R., & Debrosse, D. (2003). Enterprise
Nowlan, S. (2001). Citizen Relationship Man- Architecture: Roadmap for modernization” En-
agement E-CRM in the Public Sector,USA. terprise Modernization Issue, fall 2003, 7(2)
Pricewaterhouse-Coopers.
Xavier, M. J. (2002, April-June). Citizen Rela-
Ort, E. E. (2005). Service-Oriented Architecture tionship Management- Concepts, Tools and Ap-
and Web Services: Concepts, Technologies, and plications. South Asian Journal of Management,
Tools. Sun Developers Network.http://java.sun. 9(2), 23–31.
com/developer/technicalArticles/WebServices/
soa2/SOATerms.html#soaterms
125
key terms And deFInItIons of reform in the way Governments work, share
information, engage citizens and deliver services
Clean Data: For an organization to function to external and internal clients for the benefit of
effectively, data needs to be easily accessible both both government and the clients that they serve.
to customer sand internal users. Data is dispersed Enterprise Architecture: An organisation has
as governments like business organizations work assigned roles and responsibilities, and established
in silos. To make this data accessible to others data plans for developing products and services. The
conversion is a necessity as multiple sources and scope of Enterprise Architecture can be defined as
input formats, inconsistent styles and complex- encompassing the whole enterprise with confirmed
ity of data structures. For any new system to get institutional commitment to deliver products and
started it is necessary to convert the existing data services, both current and planned with clear tran-
from legacy systems or manual records to fit into sition plans. This in fact would define the organisa-
the new data structures. This is a critical factor in tion structure, functions and the relationships that
ERP implementation projects. By clean data it is would facilitate the organisation to meet its desired
meant that there are no duplicates definitions of goals. This is extremely essential for enterprise
data which cause inconsistency. modernisation. Development of the enterprise
Customization: In an attempt to deal with architecture will typically involve, analysing the
the potential problems presented by existing current architecture which will be a process of
information systems, there is a shift towards the description, documenting the architecture “as-is”
implementation of ERP packages. Generally it or “baseline” architecture, moving on to a defini-
is felt that ERP packages are most successfully tion of the architecture as it is planned to develop
implemented when the standard model is adopted. in the future - the architecture as it should be or
Yet, despite this, customisation activity still occurs “target” architecture which would align with the
reportedly due to misalignment of the functionality vision of the organisation.
of the package and the requirements of those in Enterprise Resource Planning (ERP): It is
the implementing organisation. In such a situation an integrated information system that integrates
the first thought that comes to a layman’s mind is all departments within an enterprise. Evolving out
to modify the software to provide the necessary of the manufacturing industry, ERP implies the
report or layout. Most ERP products are generic. use of packaged software rather than proprietary
Hence, some customisation is needed to suit the software. ERP modules may be able to interface
company’s needs. But optimal customisation in with an organization’s own software with varying
most cases is subjective with no definite rules as degrees of effort, and, depending on the software,
end-users are not always technically equipped to ERP modules may be alterable via the vendor’s
understand the far-reaching implications of the proprietary tools as well as proprietary or standard
changes that they are demanding. There is always programming languages. An ERP system can
a risk of destabilising the core application. Cus- include software for manufacturing, order entry,
tomisation can make or break the implementation accounts receivable and payable, general ledger,
of an ERP. It is therefore necessary to strike the purchasing, warehousing, transportation and hu-
right balance. man resources. The major ERP vendors are SAP
E-Government: Government’s foremost job and Oracle specialising in transaction processing
is to focus on safeguarding the nation / state and that integrates various departments. Information
providing services to society as custodian of the infrastructure, it is the technology infrastructure
nation’s / state’s assets. E-Government can there- required to manage information in an organisa-
fore be defined as a technology-mediated process tion. It consists of the computers, software, data
126
structures and communication lines underlying of public entities and their relationships with each
critical services that society has come to depend other and with the larger world. It addresses issues
on. It consists of information systems which cover such as how public sector organizations are orga-
critical aspects such as financial networks, the nized and managed, how public policy structures
power grid, transportation, emergency services the design of government programs that we rely
and government services. This is required to upon, how our states, cities, and towns work with
implement an ERP system. the federal government to realize their goals and
Inter-Operability: Various departments orga- plan for their futures, how our national govern-
nize work in silos and tend to work independently. ment creates and changes public policy programs
Data forms the basis of a transaction processing to respond to the needs and interests of our nation.
system like ERP which integrates these depart- Service-Oriented Architecture: Service-
ments so that basic data entered once can be used by oriented architecture (SOA) is the emerging trend
many. More efficient exchanging and processing in enterprise computing because it holds promise of
of data is required to provide a seamless execution IT becoming more agile in responding to changing
of a service. Inter-operability therefore means business needs. Implementing a service-oriented
the capability of different departments working architecture can involve developing applications
together to provide a service. that use services, making applications available
Public Administration: Every facet of our as services. A service-oriented architecture is an
daily lives is impacted in some way by the actions information technology approach or strategy in
of the federal, state, or local bureaucracies that which applications make use of services available
manage and organize the public life of the country in a network such as the World Wide Web.
and its citizens. Public administration is the study
127
128
Chapter 1.9
Enterprise Application
Integration (EAI)
Christoph Bussler
Merced Systems, Inc., USA
enterPrIse APPlIcAtIon their integration so that data can be exchanged or

InteGrAtIon (eAI) tecHnoloGy synchronized. The technology that supports the
integration of various application systems and
As long as businesses only have one enterprise their databases is called Enterprise Application
application or back end application system there Integration (EAI) technology. EAI technology is
is no need to share data with any other system in able to connect to back end application systems
the company. All data that has to be managed is in order to retrieve and to insert data. Once con-
contained within one back end application system nected, EAI technology supports the definition
and its database. However, as businesses grow, of how extracted data is propagated to back end
more back end application systems find their way application systems solving the general integra-
into their information technology infrastructure tion problem.
managing different specialized business data,
mainly introduced due to the growth. These back
end application systems are not independent of bAckGround
each other; in general they contain similar or
overlapping business data or are part of business Typical examples of back end application sys-
processes. Keeping the data in the various applica- tems that are deployed as part of a company’s
tion systems consistent with each other requires information technology (IT) infrastructure are an
Enterprise Resource Planning (ERP) system or a
Manufacturing Resource Planning (MRP) system.
DOI: 10.4018/978-1-60566-242-8.ch088
Enterprise Application Integration (EAI)
In the general case, different back end application in one system has to be applied to all the other
systems store potentially different data about the systems that store the same object. The objects are
same objects like customers or machine parts. replica of each other since all have to be updated in
For example, a part might be described in an ERP the same way so their content is exactly the same.
system as well as in a MRP system. The reason Only when all the objects are updated they are
for the part being described in two different back consistent again and the overall status across the
end application systems is that different aspects back end application systems is consistent again.
of the same part are described and managed. In In the replicated case it must not be possible that
fact, this means that the not necessarily equal the same object exposes different properties like
representation of the object exists twice, once in in the address example above.
every system. If there are more than two systems, A third use case is that applications participate
then it might be very well the case that the same in common business processes. For example, first
object is represented several times. Any changes a part is being purchased through the ERP system
to the object have to be applied to the representa- and upon delivery it is entered and marked as
tion of the object in all systems that contain the available in the MRP system. The business process
object. And, since this distributed update cannot behind this is consisting of several steps, namely
happen simultaneously (in the general case), dur- purchase a part, receive the part, make the part
ing the period of applying the change the same available, and so on. In this case the back end
object will be represented differently until the application systems do not share common data,
changes have been applied to all representations but their data state depends on the progress of a
in all back end application systems. It therefore business process and it has to update the back
can very well be the case that during an address end application systems accordingly. The data
update of a customer object the customer has two will change their state according to the progress
addresses. Some objects representing the customer of the business process. In this sense they share
have already the new address while others still a common business process, each managing the
have the old address. This situation exists until data involved in it.
the distributed update is complete. Furthermore, All these three use cases, while looking quite
in most cases there is no record of how many different from each other, have to be implemented
systems represent the same object. It might be by companies in order to keep their business
the case and actually often it is the case that a data consistent. EAI technology (Bussler 2003)
change is not applied to all objects because it is (Hohpe and Woolf 2003) allows to accomplish
not known which back end application system has this at it provides the necessary functionality as
a representation of the object in the first place. described next.
Only over time these cases will be detected and
rectified, mainly through the resolution of error
situations. enterPrIse APPlIcAtIon
In summary, the same object can be represented InteGrAtIon tecHnoloGy
in different back end application systems, the
updates to an object can cause delays and incon- Enterprise application integration technology
sistencies, and locations of object representations addresses the various scenarios that have been
can be unknown due to missing object registries. introduced above by providing the required func-
A second use case is that precisely the same tionality. In the following the different function-
object is replicated in different back end applica- alities will be introduced step by step. First, EAI
tion systems. In this case the update of the object technology provides a reliable communication
129
mechanism so that any communication of data back end application system that has interest in
between back end application systems is reliable. specific objects and changes to them subscribes to
This ensures that no communication is lost. This those. Every back end system publishes changes
is often achieved by sending messages through like object creation, update or deletion. If there
persistent and transactional queuing systems (Gray is a subscription that matches a publication, the
and Reuter 1993). Queuing systems store mes- same changes are applied to the subscribing back
sages persistently and transactionally achieving end application system.
an exactly-once message communication. While publish/subscribe technology is quite
EAI technology uses adapters (J2EE Connec- useful, it is not able to implement business
tor Architecture 2007) in order to connect to the process integration (Bussler 2003). In this case
proprietary interfaces of back end application several back end application systems have to be
systems. An adapter knows the proprietary inter- called in a specific order as defined by a business
face and provides a structured interface to the EAI process. Sequencing, conditional branching and
technology. This allows EAI technology to connect other control flow constructs define the invoca-
to any back end application system for which an tion order of the back end application systems
adapter is available. If no adapter is available for dynamically. Publish/subscribe cannot express this
a particular back end application system then a type of multi-step execution and more expressive
new one has to be built. Some EAI technologies technology is required.
provide an adapter development environment EAI technology incorporated workflow
allowing new adapters to be constructed. Adapt- management systems in order to enable more
ers can retrieve data from back end application complex and multi-step executions. Business
systems as well as store data within back end process based integration requirements are ad-
application systems. They therefore enable the dressed through workflows. Through a workflow
communication of the EAI technology with the specification the overall invocation order can be
back end application systems. established and complex processes like human
EAI technology by means of connecting to resource hiring or supply-chain management are
all required back end application systems with possible. Workflow steps interact with back end
adapters can propagate data changes to all of the application systems through adapters whereby
back end application systems. For example, this the workflow steps themselves are sequenced by
allows ensuring that all representations of an workflow definitions.
object in several back end application systems While business process based integration is
can be changed as required. EAI technology is a significant step in the development of EAI
required to process changes in an order preserving technology, it is not sufficient in heterogeneous
fashion to ensure object consistency. For example, environments. For example, different back end
an update of an object followed by a read of the application systems follow in general different
object value must be executed in the given order. data models. If this is the case the same type of
Not following the order would return an incor- data is represented in different ways. For example,
rect value (i.e., the value before instead of after one back end application system might imple-
the update). Any negative consequences of delay ment an address as a record of street name, street
and inconsistencies are avoided by ensuring that number, city and zip code while another back end
every request is processed in the order of arrival. application system might implement an address
In early implementations of EAI technology as two strings, “address line 1” and “address line
publish/subscribe technology was used to establish 2”. If data is transmitted from one back end ap-
the communication rules (Eugster et al. 2003). A plication system to another one the address has to
130
be mediated from one representation to the other support the EAI problem based on Web Service
representation for the data to be understood by the technology. Other vendors like IBM (IBM 2007),
receiving back end application system. Microsoft (Microsoft 2007) or Oracle (Oracle
2007) are also offering SOA based products. The
architecture underlying Web Service based ap-
current develoPments proaches is called Service-Oriented Architecture
And Future trends (SOA) and standardization work is ongoing, too
(OASIS SOA RM 2007).
The latest development in EAI technology takes Further out are Semantic Web technology based
data heterogeneity into consideration and provides approaches. They are in a research state today but
mediation technology. For example, Microsoft’s can be expected to be available in commercial
Biztalk Server (Microsoft 2007) has a mapping technology later on. A significant effort applying
tool that allows the graphical definition of mapping Semantic Web technology in order to solve the
rules that can transform a message in one format integration problems is the Web Service Modeling
into a message of a different format. Tools from Ontology (WSMO) (WSMO 2007). This ontology,
other vendors follow the same approach. The which is developed by a significant number of
research community is also working on the data organizations, uses Semantic Web Technology to
mediation problem and an overview is given in define a conceptual model to solve the integration
(Rahm and Bernstein 2001). problem that encompasses EAI integration. The
Also, this new generation realizes that back main conceptual elements proposed are ontolo-
end application systems expose not only data gies for defining data and messages, mediators for
but also public processes that define interaction defining transformation, goals for dynamically
sequences of data (Bussler 2003). A public pro- binding providers and web services for defin-
cess allows to determine which messages a back ing the interfaces of communicating services.
end application systems sends or receives and in In order to define a specific EAI integration, the
which order. This allows ensuring that all mes- formal Web Service Modeling Language (WSML)
sages are exchanged appropriately with the back (WSML 2007) is defined. This can be processed
end application system. by a WSML parser. In order to execute B2B
Web Services are a relatively new approach to integration the Web Service Modeling Ontology
communicate data and message over the Internet Execution environment (WSMX) (WSMX 2007)
(Alonso et al. 2004). Web Services are starting is developed. It serves as a runtime environment
to be used in EAI technologies in several places. for integrations defined through WSML. Another
One is for the communication between back end Semantic Web technology based approach is
application systems and the EAI technology itself. OWL-S (OWL-S 2007).
The other place is to define the interface of back Very recently a first standard in the Semantic
end application systems as Web Services. In this Web Service space was established as a W3C
case all back end application systems have a ho- recommendation. It is called SAWSDL and can
mogeneous interface technology they expose and be retrieved at (SAWSDL 2007). The main con-
an EAI technology does not have to deal with the cept behind this standard is the augmentation of
huge variety of interfaces of back end application the WSDL approach for describing Web Service
systems any more. interfaces.
Architectural approaches based on Web Service
technology are emerging. For example, BEA (BEA
2007) provides a whole suite of technologies that
131
Table 1. A Summary of Critical Issues
Autonomy Inconsistency
Back end applications are autonomous in their state changes and If a delay happens while changing different representations of the
EAI technology must be aware that since it cannot control the ap- same object inconsistencies can occur if these different representa-
plication’s state changes tions are accessed concurrently
Communication Mediation
EAI technology must provide reliable and secure communication Due to heterogeneity of back end application systems objects are
between back end application systems in order to ensure overall represented in different data models that have to be mediated if
consistency objects are sent between applications
Delay Process management
Changes of objects represented in several back end application Multi-step business processes across back end application systems
systems might not happen instantaneously due to the time required require process management to implement the particular invocation
to do the update and so a delay can occur between the update of the sequence logic
first and the last change Publish/subscribe
Distribution Back end application systems can declare interest in object changes
Back end application systems are distributed in the sense that each through subscriptions that are matched with publications of avail-
has its own separate storage or database management system with able object changes
each being separately controlled and managed Reliability
Duplication The integration of back end application systems must be reliable in
Objects that have to reside in several back end application systems order to neither loose data nor accidentally introduce data by retry
might require duplication in order to ensure back end application logic in case of failures
state consistency Replication
Heterogeneity Replication is a mechanism that insures that changes of an object
Back end application systems are implemented based on their own are automatically propagated to duplicates of this object. The vari-
data model and due to the particular management focus the data ous representations act as if they are one single representation
models differ in general not complying with a common standard Security
Communication of data between back end application systems
through EAI technology must ensure security to avoid improper
access
crItIcAl Issues oF eAI conclusIon

tecHnoloGIes
Enterprise Application Integration (EAI) technol-
The critical issues of Enterprise Application Inte- ogy is essential for enterprises with more than one
gration are listed in Table 1. Current generations back end application system. Current EAI technol-
of EAI technology have to address these issues in ogy is fairly expressive being able to handle most
order to be competitive and considered appropriate of the integration tasks. Newer developments like
implementations to the EAI problem. The same Web Services (Web Services 2004) and Semantic
applies to research. Recently research as well as Web Services (WSMO 2004) (OWL-S 2007) will
new products focuses on Web Services as a means significantly improve the situation by introducing
to integrate applications. However, Web Services semantic descriptions making integration more
are only a new technology and all the requirements reliable and dependable.
discussed above and issues listed below still apply.
Semantic Web Services (WSMO 2007) addresses
the requirements and issues taking all aspects into reFerences
consideration.
J2EE Connector Architecture (2004). J2EE Con-
nector Architecture. java.sun.com/j2ee/connector/
132
Alonso, G., Casati, F., Kuno, H., & Machiraju, SAWSDL. (2007). www.w3.org/2002/ws/sawsdl/.
V. (2004). Web Services-Concepts, Architectures
Web Services. (2007). Web Service Activity. www.
and Applications. Springer-Verlag.
w3.org/2002/ws/
BEA. (2007). BEA Systems, Inc. www.bea.com.
WSML. (2007). Web Service Modeling Language.
Bussler, C. (2003). B2B Integration. Springer- www.wsmo.org/wsml
Verlag.
WSMO. (2007). Web Service Modeling Ontology.
Eugster, P., Felber, P., Guerraoui, R., & Kermar- www.wsmo.org
rec, A.-M. (2003). The many faces of publish/
WSMX. (2007). Web Service Modeling Execution
subscribe. [CSUR]. ACM Computing Surveys,
Environment. www.wsmx.org
35(2). doi:10.1145/857076.857078
Gray, J., & Reuter, A. (1993). Transaction Process-
ing: Concepts and Techniques. Morgan Kaufmann.
key terms And deFInItIons
Hohpe, G., & Woolf, B. (2003). Enterprise
Integration Patterns: Designing, Building, and Adapters: Adapters are intermediate software
Deploying Messaging Solutions. Addison-Wesley that understand proprietary back end application
Professional. interfaces and provide easy access interfaces for
EAI technology integration.
IBM. (2007). International Business Machines, Back End Application Systems: Software
Corp. www.ibm.com. systems managing business data of corporations.
Microsoft (2007). Microsoft Corporation. www. EAI Technology: Software systems that pro-
microsoft.com. vide business-to-business integration functionality
by sending and receiving messages and retrieve
OASIS SOA RM. (2007) www.oasis-open.org/ and store them in back end application systems.
committees/tc_home.php?wg_abbrev=soa-rm Process Management: Process management
Oracle (2007). Oracle Corporation. www.oracle. allows defining specific invocation sequences of
com. back end application systems in context of EAI.
Publish/Subscribe: Publish subscribe is a
OWL-S. (2007). OWL-S. www.daml.org/services/ technology whereby interest can be declared and
owl-s/. is matched upon the publication of object changes.
Rahm, E., & Bernstein, P. (2001). A survey Workflow Technology.: Workflow technol-
of approaches to automatic schema matching. ogy is a software component that provides the
The VLDB Journal, 10, 334–350. doi:10.1007/ languages and interpreters to implement process
s007780100057 management.
This work was previously published in Handbook of Research on Innovations in Database Technologies and Applications: Cur-
rent and Future Trends, edited by Viviana E. Ferraggine, Jorge Horacio Doorn and Laura C. Rivero, pp. 837-843, copyright
2009 by Information Science Reference (an imprint of IGI Global).
133
134
Chapter 1.10
Enterprise Tomography:
An Efficient Approach for Semi-
Automatic Localization of
Integration Concepts in VLBAs
Jan Aalmink
Carl von Ossietzky University Oldenburg, Germany
Jorge Marx Gómez

Carl von Ossietzky University Oldenburg, Germany
AbstrAct changes within a specific domain. In this way root

cause analysis is supported.
Enterprise tomography is an interdisciplinary ap-
proach for an efficient application lifecycle manage-
ment of enterprise platforms and very large busi- IntroductIon
ness applications (VLBA). Enterprise tomography
semi-automatically identifies and localizes semantic This chapter covers an interdisciplinary approach
integration concepts and visualizes integration for an efficient Application Lifecycle Management
ontologies in semantic genres. Especially delta de- of Very Large Business Applications (VLBA) and
termination of integration concepts is performed in Enterprise Platforms.
dimension space and time. Enterprise tomography To be more precise, Enterprise Tomography is
supports software and data comprehension. SMEs, primarily seen as a new methodology for support-
large scaled development organizations and main- ing distributed software engineering teams in their
tenance organizations can benefit from this new incremental business development tasks and during
approach. This methodology is useful for tracking their enterprise software maintenance phases. We
database changes of business processes or coding regard enterprise software along with its meta-data,
business data and contextual data as an abstract
DOI: 10.4018/978-1-60566-856-7.ch012 information source. In this extended abstract data
Enterprise Tomography
universe our approach semi-automatically iden- containing in enterprise software conglomerates.

tifies semantic coherent entities of interest. We Based on the results of the Enterprise Tomograph
propose an algorithm for tracking the changes the operating teams are in the position to make
in this data universe in the dimension time and efficient decisions and have a reliable foundation
space. In contrast to Web 2.0 search engines, we for incremental steps of the development and
apply advanced indexing techniques. To meet maintenance life cycle.
developers and maintainers needs to the great- The Enterprise Tomograph represents a central
est extent possible, we take integration ontology ecosystem for sharing domain specific integra-
extraction algorithms into consideration, enable tion knowledge across the development teams.
controllable domain-specific indexing, apply Because of sharing and socializing of integration
delta analysis based on indices and visualizes knowledge across SCRUM teams, the Enterprise
search results of the Delta-Operator in semantic Tomography approach can be incorporated in the
categories. Furthermore, Enterprise Tomography Enterprise 2.0 initiative.
includes sharing of integration knowledge from In the research area VLBA (Very Large Busi-
individual integration experts across the enterprise ness Applications) located within business infor-
development and maintenance community. matics, Application Lifecycle Management is the
In Enterprise Software Industry development center of attention (Grabski, et. al, 2007). A real
and maintenance of VLBAs and Enterprise Plat- life VLBA, in dimension time, is in a permanent
forms is getting more and more complex. Large flux: Gradual development to meet business re-
and distributed teams are involved, teams are quirements, continuous improvements to exploit
changing and division of labor proceeds. Agile innovation, continuous maintenance to keep con-
development methods assume efficient develop- sistent business processing, horizontal connection
ment and maintenance means for software and of software systems to scale data processing and to
business data evaluation. extend business scope, recombination of loosely
Without knowing the semantic integration coupled services to exploit new business func-
of enterprise software it is inherently difficult to tionality, re-configuration and personalization,
control and orchestrate large scaled development data evolution resulting from service calls and
and maintenance processes. Domain-specific business transactions and background processing,
enterprise integration knowledge, coded in en- just to name a few.
terprise software, is normally not instantaneously VLBAs are not usually built from scratch and
available for development teams. Lack of precise deployed in an immutable stadium (Opferkuch,
knowledge of integration concepts in development 2004; Ludewig & Opferkuch, 2004). VLBAs are
and maintenance phases results in erroneous soft- not monomorphic. Some of the characteristics of
ware, is risky and might have negative business VLBAs are: Complexity, a long life-cycle, huge
impact for both the software manufacturer and continuous development and maintenance ef-
the consumer. forts, large user groups, inter- and intra-enterprise
In this paper we present a semi-automatic coupling.
environment for Application Lifecycle Manage- Today, in business reality, VLBAs are con-
ment of VLBAs and Enterprise Platforms based glomerates of inter-operating software systems.
on enhanced standard scientific algorithms. In Technical and semantic integration are the ‘DNA’
accordance with medicine diagnostics, we utilize of a VLBA. Integration can cross both system
the metaphor Enterprise Tomography for scan- and system boundaries. In this chapter we want
ning, indexing, identifying, visualization and delta to propose and outline a generic algorithm that
visualization of enterprise integration knowledge makes this VLBA integration visible and tangible
135
from different perspectives in different semantic able information to see the delta of this semantic
genres. Moreover, a delta operator is supposed to join between points of time t0 and t1. So one can
make the integration difference between points of track the evolution of the integration of a VLBA.
time t0 and t1 visible. Having in mind that VLBAs A comparison of VLBAs (e.g. modified code
consist of heterogeneous constituents, we need to clones) will be possible with our algorithm as
have an abstract holistic view on the normalized well. So it might be very interesting to evaluate
integration aspects. Beyond software, we also take the software Add-On between two releases or the
persistent data, meta data, system logs, business delta in the business data.
process activity journals, virtual and transient In accordance to tomography in medical
data, solution databases, model databases and diagnostics, we utilize a similar metaphor: The
so forth into consideration. So, we not only take Enterprise Tomograph. It is supposed to diagnose
the software of the Enterprise Platform itself into integration aspects of a complex VLBA, perform
consideration but the business data and meta data indexing of all scanned information, and provide
and contextual data as well. time efficient access to the assembled scanned data
Integration is a polymorphic topic. We regard from different semantic perspectives. Especially
integration on different levels of granularity. For the integration aspect delta is supposed to be
instance, on a low level of granularity, dynamic made available for large maintenance and service
method calls can be seen as an integration aspect. teams. Based on this information maintenance
On a medium level of granularity, cross component teams can locate real error symptoms in a VLBA
service consumption can be seen as an integration more easily and they are in a better position to
aspect or security as a cross cutting integration assess the consequences of any change in the
scattered in a VLBA. Registered integration VLBA software and therefore mitigate the risk
content (e.g. message routing rules of integration of inconsistent data processing.
hubs in enterprise service bus) is to be regarded In the scope of this chapter we assume that in
as high level integration. Logistical quantity flow, an enterprise software factory the construction
accounting value flow and financial data flow are procedure of a VLBA is very similar to that of
also prominent examples of integration on high the building industry. The first step is requirement
granularity level. Workflow, Webflow and Process specification, architecture modeling, review and
Integration can be regarded as integration on high acceptance, construction and finally continuous
granularity level as well. maintenance. The construction procedure in both
Developer teams and maintenance teams have industries is quite similar: If architecture models
different perspectives on VLBAs in their daily are in place, the construction procedure starts bot-
work: Developers are primarily assembly focused tom up: Beginning from the foundation and ending
(bottom up), whereas maintainers have the inverted in the construction of the roof, not vice versa.
top down view on a VLBA: Maintainers are Traditional top down model driven approaches
thinking rather in error symptoms and ontologies of Software Engineering do not necessarily meet
provoking the error symptoms (Abdullah, 2001). the needs of service and maintaining teams in
For example, they need to find out for a given the context of VLBAs adequately. Models are
(inconsistent) production order id the involved important, but usually those models are not linked
coding and meta data, the header and items in to the VLBA software. To bridge this technical
the persistency layer, material consumption, con- gap, the Enterprise Tomograph comes into place
firmation items, models, relevant documentation with multiple phase iteration: The Enterprise
etc. In this example we regard integration as a Tomograph knows three phases. The first phase
semantic join between those concepts. It is valu- in concerned with scanning of the VLBA, the
136
intermediate phase constructs and prepares indices Integration deltas are determined by modified
for efficient access and the third phase provides tree distance algorithms (Lu, et. al., 2001) and
access to integration data. displayed in a structured textual manner.
During the time-consuming scanning proce- Typically integration concepts are best known
dure (parallel crawling of VLBA), the Enterprise by individual developers and maintainers of the
Tomograph orchestrates concept mining algo- VLBA community. They may want to register addi-
rithms. The mining algorithms extract models, tional concept mining algorithms to the integration
software fragments, links of software fragments, ecosystem of the Enterprise Tomograph. In this
business objects, technical objects, meta data, way the integration knowledge of a few integration
solution databases and transforms those to stan- experts can be shared across the community. The
dard ontology representations that are grammar Enterprise Tomograph provides a domain-specific
compliant. The resulting ontology is the integra- search UI and displays the integration aspects. The
tion knowledge representation. We take a subset output can be iteratively refined and may serve
of the ontology representation standards into as input/workload for the Enterprise Engineer
consideration: The concept mining algorithms for further processing. For example, refactoring,
are supposed to map the integration ontology to navigation, semantic tagging, search & replace
rooted unordered labeled trees. The set of rooted within integration genres, multiple breakpoints,
unordered labeled trees are to be indexed and stored code & data compare, undo.
in PAT Arrays (PAT Array as a space efficient and The Enterprise Tomograph itself is dedicated
access optimized data structure of Patricia Tries for service enabling and can be hosted in a VLBA
and suffix trees; PAT Arrays are well known in side-by-side mode. It provides high efficient
genetic engineering and information retrieval). logarithmic access to integration concepts of a
One theme of the Enterprise Tomograph will be VLBA in the space time continuum.
the Enterprise Introspector. It generically calculates Enterprise Tomography is a proposed collab-
the footprint of a business transaction execution in orative index-snapshot based and domain-specific
the persistency layer, the footprint of a service call approach for efficient incremental development
or the footprint of a message choreography process and maintenance of VLBAs in the enterprise
instance between t0 and t1, or the delta between two software industry.
database schema in a specific domain area. Figure 1 outlines the Enterprise Tomography
The basis algorithm of the Enterprise Tomo- approach in the context of cross-organizatorial
graph is a modified DC3-algorithm (difference VLBAs. The Enterprie Tomograph crawler ex-
cover modulo 3 known in information retrieval tracts integration ontology forests from VLBAs,
(Dementiev, 2006; Burkhardt, 2006). It constructs Enterprise Platforms and databases. In-memory
PAT Arrays and stores textual suffixes in a com- indices and delta indices are created. Via search
pressed manner. The set of resulting PAT Arrays UI a full domain-specific search is possible. In
are organized in a genre-layered tenancy concept the Delta Monitor changes of integration concepts
allowing independent user groups working on can be traced.
the global index and updating it in a concurrent
polychronic way. There will be defined a quasi enterprise tomography - use
algebra on the set of PAT Arrays: Delta operator, cases and scenarios
plus operator, pseudo inverse operator.
For instance the plus operator merges 2 indi- First of all Enterprise Tomography supports
ces (PAT Arrays) resulting in one encompassing program comprehension in maintenance and
PAT Array. development phases of VLBAs and Enterprise
137
Figure 1. Enterprise tomography for VLBAs and enterprise platforms
Platforms. Views from different semantic per- entities and business data. Depending of the
spectives can be projected on a domain or area search criteria a launch pad can be generated.
of expertise. These views can incrementally be This launch pad points to the appropriate entitiy-
refined to a set of integration concepts of inter- associated operations. In contrast to Page-Rank
est. In this way a dynamic model is derived from prioritization in Web 2.0 search, the Enterprise
the enterprise software and its contextual data. Tomograph prioriztizes integrations concepts in
Maintainers and developers are in the position search result list based on reuse frequency. This
to make assessments for side effects of software means integration concepts with highest reuse
changes and enhancements. frequency appears first. Because of global impact
Furthermore, Enterprise Tomography supports in VLBA software, changing of those integration
refactoring of Enterprise Platforms and VLBAs in concepts requires highest attention.
software engineering development phases. Cross- Regarding real world VLBAs and Enterprise
system where-used lists of integration concepts Platforms there is normally no semantic linkage
can easily be generated. Textual search and replace between coding fragments, metadata, models,
functionality based on the where-used lists can be configuration data, business data, corrections,
provided. The search result list of the Enterprise consulting documents, documentation, process
Tomograph serves as workload for refactoring journals, system logs, training material, and
tasks. In-place navigation to software fragments contextual data. Taking those information sources
and contextual data enables software engineers into consideration, the Enterprise Tomograph
to drill down into the software and business data joins development and maintenance relevant
universe. information and provides ubiquituous search and
The Enterprise Tomograph can be seen as an delta tracking functionality. Integration ontologies
abstract navigation tool for software fragments, serves as an interlingua for the participating sys-
138
tems. Integration ontology extracting algorithms Enterprise software often is deployed in re-
generate normalized ontologies. With a generic leases. Codelines are the pendants in software
search UI software engineers can locate and vi- logistics. Development teams are focused on
sualize cross-cutting concerns. areas of expertise or domains respectively. Taking
In Test Driven Development (TDD) software functional upward-compatibility into consider-
engineers needs to reiterate business processes. A ation, functionality of lower releases is a subset of
generic inverse operator for business processes is functionality of higher releases. To put it in other
advantageous. Enterprise Introspection as a spezi- words: release n+1 = release n + delta. Typically
alization of Enterprise Tomography is focused on the codbase of delta is significant smaller than the
business data and configuration data of VLBAs and codbase of release n. Release n+1 and release n
Enterprise Platforms respectively. The Enterprise are similar code clones and nearly identical. The
Introspector visualizes the data delta (database Enterprise Tomograph serves as a engineering
footprint) of a business process, of a business trans- environment for controlling the code clones: Incre-
action or of a service call sequence in a domain. mental deltas can be tracked with the delta monitor.
Based on the delta index the Enterprise In- The delta can be visualized in semantic genres.
trospector enables restoring the business process E.g. new service calls, new code fragments, new
data to its former condition. The business process reusables, new business objects, new UI Objects,
chain can be executed again. new metadata or new configuration data between
Traditional IDEs support debugging function- code clone A and code clone B. Corrections to be
ality. Breakpoints can be set directly in coding. made in lower releases have normally to be up-
Conditional breakpoints and watchpoints can be ported to the latest release. This is a prominent
defined. Typically, in enterprise software industry example when code clones comes into play: the
software maintainer are not the authors of the delta ist the correction of release 1..n. . The same
coding they are responsible for. This means it is applies for downports of funktionality to lower
very difficult for software maintainer to identify releases.The Enterprise Tomograph supports to
appropriate breakpoints, not to mention to define add the delta to each individual release.
complex conditional breakpoints. With refinement Sharing integration knowledge amongst
technique, the Enterprise Tomograph supports SCRUM teams and within SCRUM teams is an
software engineers to find appropriate software essential use case. The Enterprise Tomograph
fragments. The contextual data for finding appro- serves as a SCRUM environment. It provides an
priate locations is taken into consideration. infrastructure for sharing integration knowledge
Along the lifecycle of VLBA and Enterprise to be categorized in semantic genres. A full search
platforms, quality management engineers need in a genre and a search on delta in a genre allows
indicators for quality assessment purposes. efficient location of involved integration concepts.
Software metrics and architecure metrics are ap- This supports comprehension of Integration in
propriate means for that purpose. The Enterprise Enterprise Platforms and VLBAs.
Tomograph attach metric figures to integration Generic operations and navigation based on
ontology concepts or generates virtual metric the located integration concepts allow further
entities in tree based formats. Delta tracking of development of the enterprise software. This in-
the metric figures on a domain is facilitated by cludes mass processing on the worklist provided
the Enterprise Tomograph. by search result of the Enterprise Tomograph.
Integrated exception handling and alerting are Changes and enhancements made by the
efficient functions for setting up regulator circuits SCRUM teams can be visualized instantaneously.
in software engineering. In the delta monitor the results can be verified.
139
enterprise tomography - changes model changes amongst others. The En-

Placement in Agile development terprise Tomograph abstracts the technical data
and Agile maintenance sources of the enterprise data universe and makes
integration knowledge accessible.
In Enterprise Business Engineering the evolution The Enterprise Tomograph can be feeded with
of VLBA integrity is to be regarded in a holistic integration ontology mining algorithms. In real
way. In Application Life Cycle Management the world scenarios those algorithms are provided by
VLBA software itself cannot be taken isolated few integration experts. In this way the integra-
into consideration; there are multiple correlations tion knowledge is shared by the SCRUM teams
between the VLBA software, the business data, the using the Enterprise Tomograph. The Enterprise
contextual data like documentation, solution data- Tomograph functions as a generic domain-specific
bases, e-training databases and model databases, search engine based on registered ontology min-
just to name a few. In our approach we extract ing algorithms.
Enterprise Integration Ontologies out of this ho- To make our approach applicable for real world
listic data universe, map the retrieved ontologies VLBAs and Enterprise Platforms we assume the
to a hierarchical textual representation, perform existence of a dendrogram and an enumeration of
parallel indexing on the textual representation, the dendrogram nodes. According Figure 3 a den-
organize the indices on abstract domain levels drogram categorizes the software and contextual
and make the integration knowledge available for data fragments in domains (sub-trees in the dendro-
development and maintenance teams via a search gram). A dendrogram sub-tree bundles semantic
engine. The main contribution of our approach is related entities of a VLBA. The dendrogram must
an efficient algorithm for delta determination of cover the complete VLBA software. A visitor enu-
Enterprise Integration Ontologies. merates the dendrogram nodes. Ontology mining
Figure 2 illustrates the embedding of the algorithms operate on dendrogram nodes. Indices
Enterprise Tomograph in the SCRUM develop- are created and organized in multidimensional
ment process. SCRUM teams are concerned grids spanned by axis of integration genres and
with refactoring, error detection and functional dendrogram nodes.
enhancement of VLBA software. For the sake of Moreover we assume that the dendrogram can
consistent change of the software, SCRUM teams be processed randomly and independently. This
need during their operations permanently an actual implies that the nodes are the basis granularity for
view on the integration concepts in a domain of parallel processing of ontology mining.
a VLBA software. SCRUM development is in- For instance in SAP Enterprise Platforms the
herently an incremental approach. The software package hierarchy or application component hi-
engineers needs to track the progress of their erarchy are representatives of a dendrogram.
teamwork between subsequent point of times and In a typical scenario software maintainers are
verify their enhancements and modifications. The starting with error symptoms of a VLBA and are
Enterprise Tomograph provides extended domain- interested in the involved integration concepts
based search capabilities to identify the integration and the related entities. For instance a production
concepts of interest. For instance when executing order 4711 has inconsistent confirmations on con-
a business transaction on the Enterprise Platform sumed components. The maintainer would start
the Enterprise Tomograph visualizes the footprint with a node of the manufacturing domain. The
(data delta) on the database between two points Enterprise Tomograph in this case would explode
of time. Beyond that the Enterprise Tomograph the associated sub-tree and merge the attached
visualizes software changes and documentation indices to a domain index. The resulting index
140
Figure 2. Application lifecycle management with enterprise tomograph
can be queried with the production order id. As a enumerates the domains of the dendrogram path
result he gets the data structure of the production beginning from root node to leaf node.
order, the production order instance, persistency, The difference of 2 vectors is calculated by
documentation and the production order software the Enterprise Tomograph Delta Operator. In our
fragments of job floor planning. In the next step example we see the difference of SCM (coding +
he can refine his query for the related components. data at different points of time) on a fixed location.
Now he may want to create new confirmations on Another use case is the comparison of a VLBA at
a production order component. In a subsequent different locations on the y-axis. For example the
step he can track the evolution of the identified difference of a VLBA in a development system and
ontology concepts (in our example the production in a production system can be determined. This
order component confirmations) with the delta means we compare code clones. In this regard
operator of the Enterprise Tomograph. Based on the difference of configuration data in different
the delta results the software engineer has the software locations is valuable information as well.
context of the error symptom as a foundation for On the first sight our approach is similar to
his diagnosis. traditional Web 2.0 search engine techniques.
For deeper understanding Figure 4 displays The VLBA data universe is crawled, indexed
a VLBA in an abstract space-time continuum. A with semantic data organization plus querying
vector in this space represents a dendrogram node via search UI by developer and maintainers. To
index at a point of time . The x-axis represents meet developers and maintainers requirements, we
the instances of the VLBA software and its busi- need to make optimizations in different aspects
ness data persistency at different points of time. as shown in Figure 5. In addition to traditional
The y-axis enumerates the locations of the VLBA approaches, in the Enterprise Engineering context
software within the software landscape. The z-axis we need to focus on domain specific search refine-
141
Figure 3. Dendrogram of a VLBA
Figure 4. VLBA data universe
142
ments and a flexible delta calculation based on It supplies rooted unordered labeled trees. Those
domains. Generic operators on generic in memory trees are sequenced. The skewed algorithm DC3
data representations are the main focus of this (difference cover modulo 3) is performed on the
scientific paper. sequenced data. The resulting index and the origi-
Enterprise Tomography is focused on inte- nal sequenced data is compressed and in-memory
gration ontologies and their evolution. It can be organized in an 2 dimensional node-genre grid.
regarded as a special case of ontology integration Each element of this grid points to a compressed
(Abels, et. al., 2005; Nicklas, 2005). PAT Array and its associated sequenced data.
Based on a VLBA dendrogram this procedure
enterprise tomograph - building can be repeated for different data sources, i.e. for
blocks and data Flow different VLBA locations or for different point of
times. Later on the data sources can be compared
The Enterprise Tomograph is divided in building with the Enterprise Tomograph Delta Operator.
blocks for time consuming tasks (data extraction When the developer needs to perform a search
and data organization) and building blocks for in a domain, he selects a dendrogram node. The
time efficient tasks like querying, index merging, node sub-tree is exploded and all node-associated
and visualization. Figure 6 illustrates the anatomy PAT Arrays together with sequenced data are
and high level architecture of the Enterprise To- decoded. The PAT Arrays are merged. The results
mograph. First of all the VLBA dendrogram is are published in the Domain Index Layer. A Query
determined. Based on a dendrogram the Enterprise Engine executes queries based on the domain index.
Tomograph starts the VLBA crawling for contain- Monitor 1 visualizes the search results. Now the
ing nodes. For each node and for each semantic developer may want to refine the search results.
genre an Ontology Mining Algorithm is executed. The search results are sequenced and re-indexed
Figure 5. Design principles of the enterprise tomograph
143
with DC3 Algorithm. A new PAT Array comes into in Figure 7. Basically for a given dendrogram
being, is coded and placed into a data source which node and a semantic genre a forest of rooted un-
is published to the Domain Index Layer again. ordered labeled trees is determined. The VLBA
The Query Engine searches for the new search data universe serves as the input. Data extraction
pattern and the refined results are displayed again happens with subsequent stemming algorithms
in Monitor 1. This search-refinement round-trip is for filtering significant information relevant to
depicted in Figure 6 with the circular arrow. The the integration context. According to rule sets,
phases of the refinement round-trip can be started ontologies are calculated and normalized. The
asynchronously: While the user is evaluating search resulting trees are annotated with tree hash values
results on Monitor 1, the time consuming PAT Ar- and with tree node hash values as labels.
ray construction via DC3 can be started in parallel The view projects tree nodes to the node set
as preparation for the next refinement query. In of interest. With parameter the behavior of the
this way the user is not aware of interruptions of ontology mining algorithm is influenced. Ontol-
subsequent time consuming index construction. ogy Mining Algorithms reside in a framework
In delta mode of the Enterprise Tomograph, and are orchestrated according to the Inversion
delta calculation is performed on two data sources. of Control Design Pattern. Integration ontology
As explained later on in more detail, domain indi- mining algorithms can be registered by integra-
ces of source m and source n are merged together tion experts.
with watermark data. Delta trees of both tree sets Rooted unordered labeled trees are used for
are determined. Monitor 1 serves for displaying representing integration concepts. An exemplary
of delta trees of source m in full version whereas integration ontology instance is displayed in Figure
Monitor 2 displays delta trees of source n in full 08. Here the business object instance relations,
version. The edit scripts of the delta trees are coding fragments, data persistency, APIs amongst
visualized in Delta Monitor. others are highlighted. Trees containing the inte-
The main part of the VLBA crawling mecha- gration concepts are sequenced to a textual data,
nism is the ontology mining algorithm as outlined that is indexed with DC3 algorithm.
Figure 6. Anatomy and high level architecture of enterprise tomograph
144
Figure 7. Integration ontology mining
enterprise tomograph integration ontologies. Tree distance algorithms

- delta operator generated considerable interest in the research
community within the last years. Assuming there
In this chapter we want to examine the delta de- are 2 ontology trees originating from different
termination and delta visualization of enterprise data sources as depicted in Figure 9. Each node is
Figure 8. Example for integration ontology
145
annotated with a label, i.e. with a hash value calcu- trees into consideration. We assume that delta de-
lated by the ontology mining algorithm mentioned termination for unordered non-unique labeled trees
before. Now the question arises which minimal is a relaxative or similar problem in comparison
set of operations need to be performed on tree F to delta determination for unique labeled trees. A
to retrieve tree G. This minimal set of operations proof for this can be made with Simplex Algorithm
is called edit script. The minimal number of edit known in Dynamic Programming (Bille, 2005).
operations is called edit distance. The edit script This issue is not detailed in this paper.
serves as a basis for visual mapping of tree F into In the next section we want to enlighten the
tree G in the Enterprise Tomograph Delta Monitor. skewed DC3 algorithm for PATArray construction.
It is worth noting that as of today delta de- As explained in Figure 10, PAT Array construction
termination for rooted unordered labeled trees is means an indirect lexicographical sort of semi-
considered as NP-complete. Delta determination finite suffixes of a given string. In (Burkhardt,
of ordered labeled trees is much more efficient. 2006) is is proven that this construction consumes
Because of paging in the Delta Monitor, it is not linear operations. The DC3 Algorithm is a skewed
mandatory to determine all tree deltas at a time. divide & conquer algorithm. Basically the steps
Although extremely time consuming procedure mentioned in Figure 10 are executed. A detailed
for large trees, delta determination for reasonable pseudo algorithm and a concrete implementation
sized unordered labeled trees can be performed of DC3 is given in (Burkhardt, 2006).
efficiently on demand at visualization time. PAT Arrays are space efficient data structures
In contrast to the original definition of labeled containing indirect hierarchical lexicographic sort
trees (Shasha, 1992), we take non-unique labeled order according to Patricia Tries. Extended PAT
Figure 9. Tree distance visualization with edit script
146
Figure 10. DC3 algorithm for PAT array construction
Arrays accelerates the search procedure. Longest The Delta Operator of the Enterprise Tomo-
Common Prefix LCP can be skipped during search graph is concerned with detecting Delta Trees
(Abouelhoda, et. al., 2005). A drawback of the of two forests originating from different data
extension is the additional space consumption sources. As shown in Figure 12 the intersection
in PAT Array. represents integration ontologies existing in both
PAT Arrays in the context of Enterprise To- data sources A and B. A subset of this intersection
mograph are advantageous because PAT Arrays are delta integration ontologies and needs to be
can be merged in linear time. identified and visualized according to Figure 09.
A PAT Array is the basis for logarithmic in- The delta identification algorithm is explained in
direct search. Assuming all semi-finite suffixes the next section in more detail.
with prefix = search string are to be identified. The basic idea for identifying delta trees is to
A’ points to semi-finite suffixes. A’ is divided in integrate watermarks into the sequenced forest
2 intervals: [left bound...medium] and [medium... text content. As outlined in Figure 12 a watermark
right bound]. Medium serves as a basis for indirect is assembled of a 4 fixed-lenght tuple with hash
comparison: if search string is lower than semi- value of TreeID, hash value of tree, the location
finite suffix of medium then the new right border:= of the tree and the offset of the sequenced tree.
medium otherwise new left border:= medium. The Watermark integration is done for all trees in both
new medium is determined in the middle of the forest A and B. PAT Array construction is done for
new interval. This procedure is re-iterated until the textual content of A and B each. The resulting
border convergence. PAT Arrays are merged. On this constructed PAT
All lexicographical neighbors can be found in Array a binary indirect search is performed for
a row direct behind the location identified previ- the tag <WATERMARK>. Watermark neighbors
ously via indirect logarithmic search. can be found in a sequence. Iterating over this
147
Figure 11. PAT array search
Figure 12. Delta trees
sequence we can easily determine trees located With this approach we can reuse PAT Array
in A only, in B only, trees in A and B, and delta algorithms for negative doublet recognition of trees,
trees in A and B. Delta trees have a common i.e. delta tree detection. Of course, the administra-
prefix ‘<WATERMARK>Hash Value TreeID’ tive data integrated in the textual content must not
with different hash value tree in consecutive PAT be part of any visualized search result. Appropriate
Array positions. measures can suppress these occurrences.
148
Figure 13. Delta tree determination
enterprise tomograph - This collaborative index update policy ensures

Index organization a consistent index without the need for re-indexing
the whole VLBA data universe.
In large-scaled VLBA development and mainte-
nance projects numerous organizational entities on Generic operators in
different granularity levels are involved. Entities maintenance networks
are potentially performing changes on the VLBA
data universe. The aim is to keep the index up- Enterprise software manufacturers normally pro-
to-date without complete re-indexing after each vides maintenance services for their customers.
change made. They are granted access to their individual cus-
We assume that the project organization is hier- tomers for detailed root cause analysis. In fact, the
archical. Figure 14 reflects this situation. The root code bases of the individual deployed enterprise
node represents the global index. Each development software may be modified or enhanced. Such
location may redefine subsets of the global index. A individual changes may result in error symptoms.
user for example works in a user-specific workspace. To make modifications and enhancements visible,
He may want to redefine a subset of the index grid the Inter-Delta Operator of the Enterprise Tomo-
of his development team at a different point of time. graph comes into place. E.g. enterprise software
With the Enterprise Tomograph he can track the manufacturer SAP provides an encoded index of a
progress of his work (delta determination) in com- domain of its Enterprise Platform (reference index)
parison to the snapshot on team level. As soon as he into a P2P network. The customer can calculate
accepts the correctness of his changes, he activates the delta of its changed enterprise software and the
user-specific index. In this case his index subset is decoded reference index. In semantic genres the
updated along the path up to the root node. After this modifications and enhancements are listed. Now
all predecessor entities can access the user specific the maintainer semi-automatically can assess and
changes. A lock mechanism during redefinition estimate, if the error symptom relates to modifica-
phase circumvents conflict resolution. tions and enhancements made by the customer.
149
Figure 14. Enterprise tomograph index organization
Another meaningful operator in maintenance With Reed Solomon Codes, retrieval of arbi-
networks is the align operator. Assume the enter- trary n of m fragments suffices for reconstruction
prise software manufacturer wants to deploy a best of the original domain index (Grolimund, 2007).
practice. This best practice (delta) can be added The basics behind is the fundamental theorem
to customers configuration. In this case, adding of algebra (Korner, 2006). If a fragment is dam-
a delta is equivalent to aligning configuration in aged, or not transmitted consistently, the retrieval
a domain. of an additional encoded fragment may be trig-
The align operator can also be applied for code gered (n different consistent encoded fragments
bases. In this way updates of enterprise software are necessary for reconstruction of the original
can be provided. domain index).
The locate operator performs concept location The independent fragment retrieval approach
in integration ontologies. This operator can be ensures an efficient upload and download of large
iteratively be applied for refinement purposes. Enterprise Tomograph domain indices because n
Based on the result other orthogonal operators (n of m) arbitrary encoded fragments can be re-
can be applied e.g. multiple break-point opera- trieved in parallel for reconstructing the original
tor, launch operator, modify operator. The locate domain index (Grolimund 2007).
operator is the basis for refactoring purposes. In addition, with RS-Codes (Reed Solomon
In the P2P network the sender of the encoded Codes) damaged or not consistently transmitted
domain index is the owner of the key. Enterprises encoded fragments can be reconstructed with high
knowing this key are able to decrypt and use the reliability. Encoded fragments contains attached
domain index only. The domain index is divided redundant information (polynom). With polynom
into n fragments. These fragments are redundantly division the correctness can be evaluated. With
encoded and the resulting amount of encoded help of RS erasure codes damaged fragments can
fragments m = n + k is distributed to the peers of be repaired (Forward Error Correction).
the network. The retrieval of the encoded index Figure 15 outlines the interplay of the generic
happens according to an implementation of a operators in the P2P maintenance network.
distributed hash table (DHT).
150
Figure 15. Generic operators in maintenance network
relAted Works conclusIon
Enterprise Tomograph with its delta operator Enterprise Tomography is an efficient approach for
has similarities to ontology versioning in ontol- Application Lifecycle Management of VLBAs and
ogy management frameworks (Noy & Musen, Enterprise Platforms. It supports development and
2004). Integration concept location is related to maintenance teams to track incremental changes of
identifying cross-cutting concerns in enterprise integration concepts in VLBA software, business
software. Ontology mining relates to static and data and contextual data. Full domain-specific
dynamic aspect mining in a more abstract sense search of integration concepts, refinement search
(Krinke, 2006). Techniques used in (Hollingsworth and delta search and its visualization is realized
& Williams, 2005) can be seen as a special case with enhanced standard scientific algorithms.
of Enterprise Tomography. Interchangeable enhanced PAT Arrays, PAT ar-
Semantic Integration in VLBAs is covered in ray based delta tree recognition, tree distance
(Brehm & Haak, 2008). In Enterprise Tomography algorithms and tree mappings based on dynamic
integration ontologies are derived from VLBAs. programming are the main algorithms used in
Integration ontologies as an intermediate dynamic Enterprise Tomography. Complexity calculation
model are used for ontology integration. for Enterprise Tomography can easily be derived
Ontology concepts are also used as model from complexity of the individual involved en-
for program comprehension in combination with hanced standard algorithms.
technical unrelated artifacts (Panchenko, 2007). According to VLBA definition, the Enterprise
Tomograph with its generic operators in P2P main-
tenance networks can be regarded as a VLBA. The
business process behind is maintenance service
provisioning for deployed enterprise software.
151
reFerences Grolimund, D. (2007). Wuala - a distributed file

system. Caleido AG, ETH Zuerich. Online Publica-
Abdullah, R., Tiun, S., & Kong, T. E. (2001). Au- tion in Google Research, Google TechTalks.
tomatic topic identification using ontology hierar-
chy. In Proceedings, Computational Linguistic and Hollingsworth, J. K., & Williams, C. C. (2005).
Intelligent Text Processing, Second International Automatic mining of source code repositories
Conference CICLing, Mexico City, Mexico (pp. to improve bug finding techniques. IEEE Soft-
444-453). Berlin, Germany: Springer. ware Engineering, 31(6), 466–480. doi:10.1109/
TSE.2005.63
Abels, S., Haak, L., & Hahn, A. (2005). Identi-
fication of common methods used for ontology Korner, T. E. (2006). On the fundamental theorem
integration tasks. Interoperability of heteroge- of algebra. Journal Storage, 113(4), 347–348.
neous information systems. In Proceedings of the Krinke, J. (2006). Mining control flow graph
first international workshop on Interoperability from crosscutting concerns. In Proceedings of the
of heterogeneous information systems, Bremen, 13th Working Conference on Reverse Engineer-
Germany (pp. 75-78). New York: ACM. ing (WCRE): IEEE International Astrenet Aspect
Abouelhoda, M. I., & Kurtz, S. (2005). Replacing Analysis (AAA) Workshop, Benevento, Italy (pp.
suffix trees with enhanced suffix arrays. Journal 334-342).
of Discrete Algorithms, 2, 53–86. doi:10.1016/ Lu, C. L., Su, Z.-Y., & Tang, C.-Y. (2001). A new
S1570-8667(03)00065-0 measure of edit distance between labeled trees. In
Bille, P. (2005). A survey on tree edit distance Proceedings of the Computing and Combinatorics,
and related problems. Theoretical Computer 7th Annual International Conference, Cocoon
Science, 337(1-3), 217–239. doi:10.1016/j. 2001, Guilin, China (pp. 338-348).
tcs.2004.12.030 Ludewig, J., & Opferkuch, St. (2004). Softwarew-
Brehm, N., & Haak, L. (2008). Ontologies support- artung - eine taxonomie. Softwaretechnik-Trends,
ing VLBAs: Semantic integration in the context Band 24 Heft 2, Gesellschaft für Informatik.
of FERP. In Proceedings of the 3rd International Nicklas, D. (2005). Ein umfassendes umge-
Conference on Information and Communication bungsmodell als integrationsstrategie für orts-
Technologies: From Theory To Applications, bezogene daten und dienste. Unpublished doctoral
ICTTA 2008, (pp. 1-5). dissertation, University Stuttgart, Online Publica-
Burkhardt, S., Kärkkäinen, J., & Sanders, P. tion, Stuttgart.
(2006). Linear work suffix array construction. Noy, N. F., & Musen, M. A. (2004). Ontology
[JACM]. Journal of the ACM, 53(6), 918–936. versioning in an ontology management frame-
doi:10.1145/1217856.1217858 work. IEEE Intelligent Systems, 19(4), 6–13.
Dementiev, R. (2006). Algorithm engineering for doi:10.1109/MIS.2004.33
large data sets. Unpublished doctoral dissertation, Opferkuch, St. (2004). Software-wartungsproz-
University of Saarland, Saarbrücken. esse - ein einblick in die industrie. Fachbericht
Grabski, B. Günther, S., Herden, S., Krüger, L., Informatik, Nr. 11/2004, Universität Koblenz-
Rautenstrauch, C., & Zwanziger, A. (2007). Very Landau.
large business applications. Berlin, Germany:
Springer.
152
Panchenko, O. (2007). Concept location and pro- Shasha, D., Statman, R., & Zhang, K. (1992).
gram comprehension in service-oriented software. On the editing distance between unordered la-
In Proceedings of the IEEE 23rd International beled trees. Information Processing Letters, 42,
Conference on Software Maintenance: Doctoral 133–139. doi:10.1016/0020-0190(92)90136-J
Symposium, ICSM, Paris, France (pp. 513–514).
IGI Global).
153
154
Chapter 1.11
Enterprise Information
System Security:
A Life-Cycle Approach
Chandan Mazumdar
Jadavpur University, India
Mridul Sankar Barik

Anirban Sengupta
AbstrAct consider enterprise information system security

as a management issue and detail the information
There has been an unprecedented thrust in em- security parameters. Finally, the proposed security
ploying Computers and Communication technolo- engineering life-cycle is described in detail, which
gies in all walks of life. The systems enabled by includes, Security Requirement Analysis, Security
Information Technology are becoming more and Policy Formulation, Security Infrastructure Advi-
more complex resulting in various threats and sory Generation, Security Testing and Validation,
vulnerabilities. The security properties, like confi- and Review and Monitoring phases.
dentiality, integrity, and availability, are becoming
more and more difficult to protect. In this chapter,
a life-cycle approach to achieve and maintain IntroductIon
security of enterprises has been proposed. First,
enterprise information systems are looked at in There has been unprecedented thrust in employing
detail. Then, the need for enterprise information Computers and Communication technologies in
system security and problems associated with se- all walks of life including business, education and
curity implementation are discussed. The authors governance in almost all the countries. This is a
one-way trend in the sense that there is no going
DOI: 10.4018/978-1-60566-132-2.ch007 back. While, this means lower cost, operational
Enterprise Information System Security
efficiency and client satisfaction, on the flip side, bAckGround

the systems enabled by Information Technology
are becoming more and more complex resulting enterprise and its Functionality
in various vulnerabilities. Also, there are innumer-
able threats which exploit those vulnerabilities. The Compact Oxford English Dictionary (Weiner,
More importantly, due to global connectivity 1991) defines an “enterprise” as “a project or
through the Internet, threats are not confined to undertaking, especially a bold one”; “bold re-
a particular area or region; they are omnipresent. sourcefulness”; or, “a business or company”.
These threats pose problems to the economic and Webopedia states that an enterprise is “a business
administrative independence. organization. In the computer industry, the term
In the context of Information Security, as our is often used to describe any large organization
information infrastructures are becoming more that utilizes computers”. Combining these, we
and more complex, and connected, the proper- define an enterprise as an organization (Industry/
ties like confidentiality, integrity and availability Govt./Academic) created for business or service
are becoming more and more difficult to protect/ ventures. From the Information Security point of
achieve. The adoption of Information Technol- view, an enterprise is characterized by its business
ogy Act in different countries provides the legal goals, business activities, organizational structure,
framework for acceptance of electronic documents and assets and infrastructure.
in business and governance, as well as to deter the The Compact Oxford English Dictionary
wrong-doers. Also, the international community (Weiner, 1991) defines “information” as “facts or
is adopting Standards such as ISO 27001 (Snare, knowledge provided or learned; what is conveyed
2005) and ISO 17799 (Plate, 2005) for best prac- or represented by a particular sequence of symbols,
tices in security management. All these standards impulses, etc”. The Wikipedia entry for informa-
have evolved from the knowledge, experience tion is “the result of processing, manipulating and
and expertise of international experts. It has been organizing data in a way that adds to the knowl-
recognized that the security of enterprises has to be edge of the receiver”. Thus, information can be
tackled from the point of view of a management viewed as data that is organized and accessible in
structure than from purely technological angle. a coherent and meaningful manner. The generation
The rest of this chapter is organized as follows. and use of information has some commonalities
Section 2 presents some background information in different types of enterprises. For example, all
and defines an enterprise and its functionality. It of them rely on user and operator interactions,
also details enterprise information systems and reliable storage and retrieval, correct processing,
discusses the need for enterprise information as well as timely and good quality dissemination
system security. Section 3 lists the issues, con- of information. More and more enterprises are
troversies and problems associated with security becoming dependent on the efficiency and qual-
implementation. It describes enterprise informa- ity of generation and processing of information.
tion system security as a management issue and Information has become the prime mover in the
details the information security parameters. Sec- growth and sustenance of all kinds of enterprises.
tion 4 describes our proposed security engineering Information and the technology supporting the
life-cycle. Section 5 lists few future research areas creation, and management of information act as
in enterprise information system security. Finally, important assets of any enterprise. Thus there is
our conclusions are in section 6. a specific need to protect these assets.
155
enterprise Information system need for enterprise Information

system security
Various information resources which are building
blocks of any enterprise information system are To get a feeling of the urgent need for information
typically classified into five categories (not neces- system security in the present day scenario, we
sarily disjoint): containers, transporters, sensors, have to look back at the age of purely paper-based
recorders, and processors (Denning, 1999). enterprises. In those days, security of valuable
Containers are physical media that hold in- information was primarily provided by means
formation. Examples include human memories, of physical and administrative means. But the
computer memories, print media, tapes, disks, and widespread use of sophisticated data processing
containers for the above. Objects and communica- equipments in day-to-day enterprise activities ne-
tion systems that carry information from one loca- cessitated a face change for enterprise information
tion to another are Transporters. Examples include security. This also became evident with the intro-
human messengers, who carry the information duction of distributed transaction systems such
whenever they move from one place to another; as railway reservation systems, ATM networks,
physical transportation systems, i.e. trucks, planes, enterprise WAN etc. Here the main concern is the
and postal systems; point-to-point telecommuni- security of data during transmission over private
cations systems; broadcast media; and computer or public network (Hassler, 2001).
networks. Sensors extract information from other Documents play a key role in most of the hu-
objects and the environment. Examples are human man activities ranging from commerce, foreign
sensors, cameras, microphones, scanners, and policy, military action, and personal interactions
radar. Devices like printers, tape recorders, and etc. and success in most of the cases depend on
disk writers that place information in containers are all the parties involved, having confidence over
called Recorders. Finally, objects that manipulate the integrity of the documents involved. Docu-
information are information processors. Examples ments typically carry signatures and dates which
include humans, computer hardware, and software. require protection from disclosure, tampering, or
All these resources in an information system destruction. With information systems becoming
work in unison enabling information flow from ever more pervasive and essential to our day-to-
one container to another and over a variety of day activities, electronic information is taking over
transporters. Sensors extract information from many of the roles traditionally performed by paper
the physical environment, which is then com- documents. These call for efficient technologies
puterized, processed, broadcast over radio and for electronic documents to replace functions tra-
television, transmitted over telecommunications ditionally performed with paper-based documents.
systems, computer networks, and fed into devices To survive in the present day competitive
that control processes in the environment such business scenario, enterprises are often driven to
as heating and cooling. This interconnectivity of provide their services online. But security of online
resources also paves the way for hackers to initiate transactions becomes a major issue for enterprises
operations that affect resources other than those which need to take such a decision. Also the secu-
explicitly hit. rity requirements of individual transactions vary
widely from one type to another. For example, a
million-dollar commercial construction loan being
approved by all the parties involved has a much
156
stronger security requirement than basic consumer business with an enterprise easier, know-
transactions such as purchasing a CD or a customer ing that best security practices are adopted
service request. Additionally, confidential docu- by the enterprise.
ments such as financial or health records when • To comply with regulatory requirements
put online, requires enhanced security to ensure and fiduciary responsibilities: Information
personal privacy. Security officers of every enterprise have
In the light of the above discussion, we are a responsibility to ensure the safety and
now in a position to answer the question: “Why soundness of the enterprise, which also
is information security important?”, as it is crucial includes the continuing operation of the
for enterprises to define why they want to achieve enterprise. As a result, enterprises that rely
information security, in order to determine how on computing elements for their business
they will achieve it. Information security is im- must develop policies and procedures that
portant for the following reasons: address not only the protection of informa-
tion assets but also protection of the enter-
• To protect enterprise assets: This is one of prise from liability. For example financial
our primary goals. By “assets” we mean the organizations must protect shareholders’
hardware, software, and information assets investments and maximize return. In ad-
of an enterprise. Hardware assets (Casar, dition, organizations are often subject to
2004) consist of computing systems, com- government regulations, which generally
munication links, and IT sites (or, rooms) stipulate safety and security requirements
in which computing systems are installed. of an organization. Severe penalty is im-
Software assets are the automated and posed on those enterprises which fail to
manual applications that process informa- comply with regulations. In some cases,
tion. Information asset pertains to input, corporate officers who have not properly
processed, output and stored data. performed their regulatory and fiduciary
• To gain a competitive advantage: responsibilities are personally liable for
Developing and maintaining effective se- any losses incurred by the financial institu-
curity measures can provide an enterprise tion that employ them.
with a competitive advantage over its
peers. Information security is particularly Issues, controversIes And
important in the area of e-commerce and fi- Problems oF enterPrIse
nancial services provided over the Internet. InFormAtIon system securIty
It can mean the difference between wide
acceptance of a service and a mediocre Present scenario
customer response. For example, how
many people do you know who would use Information infrastructure consists of technologies
a bank’s Internet banking system if they for gathering, handling, and sharing information.
knew that the system has been success- As the dependence on information infrastructure
fully hacked in the past? Not many. They is significantly increasing there are also rising
would rather go to the competitors for their cases of vulnerability breaches, threats, cyber
Internet banking services. Enterprise secu- attacks and cascading or pervasive failures. They
rity practices instill confidence in custom- are caused due to lack of inherent security in new
ers and employees. This, in turn, makes technologies and protocols, flaws in commonly
potential customers’ decision of doing used products, failures to address security con-
157
cerns in system design and use, lack of manage- approach to security is no longer sufficient. The
ment commitment and approach, and lack of user reasons are:
awareness about security.
Organizations such as banking, finance, tele- • The network perimeter is no longer static;
communication, transportation and energy have remote users, contractors, partners, suppli-
infrastructures that are highly interdependent. ers, even web services-based applications
Attacks on one infrastructure can damage other are all allowed network access regardless
infrastructure as well. Concentrated infrastructure of physical location.
attacks could thus have a significant effect on our • Automated attacks often enter the network
national security and economy. through legal TCP ports before creating
A look at the information security requirements havoc on internal systems.
of organizations reveal that the amount of IT assets, • Many attacks come from disgruntled em-
employees and clients using those assets, nature of ployees, not outside hackers. According
their use and geographic distribution of all these to the CSI/FBI 2003 survey (Richardson,
are huge. Even for a medium sized enterprise, 2003), 77% of companies claim that em-
the task of security management is too complex. ployees are the most likely to commit se-
Access to critical assets must be controlled after curity crimes.
considering their specific security needs (like
confidentiality, integrity, and availability needs, As business requirements blur the perimeter
as discussed later in this chapter). Other factors between internal and external users, network avail-
like the dynamic nature of the security require- ability is more important than ever. Thus worm
ment and business context, combined with the and virus-driven interruptions must be minimized
technological infrastructure, patch management, and contained. To do so, security managers are
and security testing, complicate matters. contemplating to apply new technologies to seg-
For years, network security was based upon ment and protect internal networks including
three primary products: firewalls, Vitual Private end-point security, network behavior modeling,
Networks (VPNs) and anti-virus software, but Intrusion Detection/Prevention Systems (IDS/
this security triad has reached its limit. Automated IPS), and internal firewalls.
Internet worms, viruses, and Distributed Denial Today’s reactive network security strategy
of Service (DDOS) attacks are more prevalent brings heartburn to security staff while leaving
and virulent than ever before causing billions enterprises vulnerable to attacks. Fortunately,
of dollars in worldwide damage and impacting this situation faces extinction. Business execu-
companies like Bank of America, Continental tives now realize that security is an important and
Airlines, eBay, and Yahoo. Also, new technologies necessary component of their overall business
like IP telephony, WLANs, and Instant Messaging infrastructure and must be comprehensive and
are gaining rapid acceptance opening up another sound. Over the next few years, companies will
potential avenue for attacks. invest heavily in network security technology
while they formalize policies and improve pro-
the Future cedures. At the same time, security features will
aggregate and integrate into network hardware
For the past 10 years, security managers have fo- and management software. These efforts will
cused their efforts on the network perimeter with help companies establish an enterprise security
the belief that all external users are ‘untrusted’ view while lowering operating costs – a major
while internal users are trusted. This kind of bottleneck of security technology today.
158
enterprise Information system of occurring and then protect against them. For
security is a management Issue example, it is possible that an organization could
be subjected to a van Eck monitoring or a high-
Enterprise information system security is not energy radio frequency (HERF) attack, but the
absolute. All security measures are relative. probability is low. It is also important to balance
Information security should be thought of as a business needs with the need for security, assessing
spectrum that runs from very insecure to very the operational impact of implementing security
secure. The level of security for an information measures. Security measures and procedures that
system is dependent on where it lands along that interfere with the operation of an organization are
spectrum relative to that point. There is no such of little value. Those types of measures are usually
thing as an absolutely secure information system. ignored or circumvented by company personnel,
Information security is a balancing act that re- so they tend to create, rather than plug, security
quires the deployment of “proportionate defense”. holes. Whenever possible, security measures
The defenses that are deployed or implemented should complement the operational and business
should be proportionate to the threat. Enterprises needs of an enterprise.
determine what is appropriate in several ways like In other words, enterprise information system
balancing the cost of security against the value security is a management issue where, a fine bal-
of the assets they are protecting; balancing the ance is to be drawn between system efficiency,
probable against the possible; balancing business security expenses and information protection.
needs against security needs.
Enterprises must determine how much it would enterprise Information system
cost to have each system or network compro- security Parameters
mised – in other words, how much it would lose
in monetary terms owing to un-authorized access The goal of enterprise information system security
to the system or information theft. By assigning a is to “enable an organization to meet all of its mis-
monetary value to the cost of having a system or sion/business objectives by implementing systems
network compromised, enterprises can determine with due care consideration of IT-related risks to
the upper limit they should be willing to pay to the organization, its partners and customers”. The
protect their systems. For many enterprises this security goal can be met through the following
exercise is not necessary, because information security parameters (Stoneburner, 2001).
systems are lifeblood of the business. Without
them there is no business. Enterprises also need Availability (of Systems and
to balance the cost of security against the cost of Data for Intended Use Only)
a security breach. Generally, as the investment
in security increases, the expected losses should Availability (Yu, 1990) is a requirement intended
decrease. Companies should invest no more in to assure that systems work promptly and service
security than the value of the assets they are is not denied to authorized users. This objective
protecting. This is where cost benefit analysis protects against:
comes into play.
Moreover, enterprises must balance possible • Intentional or accidental attempts to either
threats against probable threats, as it is not possible perform unauthorized deletion of data or
to defend against every possible type of attack. otherwise cause a denial of service or data.
Therefore it is necessary to determine what types • Attempts to use system or data for unau-
of threats or attacks have the greatest probability thorized purposes.
159
Availability is frequently an organization’s Assurance (That the Other

foremost security parameter. Four Parameters Have
Been Adequately Met)
Integrity (of System and Data)
Assurance is the basis for confidence that the
Integrity (Biba, 1975; Clark, 1989; Goguen, 1982 ; security measures, both technical and operational,
Mayfield, 1991) has two facets: work as intended to protect the system and the
information it processes. The other four security
• Data integrity (the property that data has parameters (integrity, availability, confidentiality,
not been altered in an unauthorized man- and accountability) have been adequately met by
ner while in storage, during processing, or a specific implementation when:
while in transit) or
• System integrity (the quality that a system • Required functionality is present and cor-
has when performing the intended function rectly implemented,
in an unimpaired manner, free from unau- • There is sufficient protection against unin-
thorized manipulation). tentional errors (by users or software), and
• There is sufficient resistance to intentional
Integrity is commonly an organization’s most penetration or by-pass.
important security parameter after availability.
Assurance is essential; without it the other
Confidentiality (of Data and parameters are not met. However, assurance is a
System Information) continuum; the amount of assurance needed varies
between systems.
Confidentiality (Bell, 1976; Lampson, 1974) is the The five security parameters are interdependent
requirement that private or confidential informa- (Stoneburner, 2001). Achieving one parameter
tion not be disclosed to unauthorized individuals. without consideration of the others is seldom pos-
Confidentiality protection applies to data in stor- sible. This is depicted in Figure 1. Confidentiality
age, during processing, and while in transit. For is dependent on Integrity, in that if the integrity
many organizations, confidentiality is frequently of the system is lost, then there is no longer a
behind availability and integrity in terms of im- reasonable expectation that the confidentiality
portance. Yet for some systems and for specific mechanisms are still valid. Integrity is dependent
types of data in most systems (e.g., authenticators), on Confidentiality, in that if the confidentiality
confidentiality is extremely important. of certain information is lost (e.g., the super-
user password), then the integrity mechanisms
Accountability (to the Individual Level) are likely to be by-passed. Availability and Ac-
countability are dependent on Confidentiality and
Accountability is the requirement that actions of Integrity, in that:
an entity may be traced uniquely to that entity.
Accountability is often an organizational policy re- • If confidentiality is lost for certain infor-
quirement and directly supports non-repudiation, mation (e.g., super-user password), the
deterrence, fault isolation, intrusion detection mechanisms implementing these parame-
and prevention, and after-action recovery and ters are easily by-passable; and
legal action.
160
Figure 1. Security parameter dependencies
• If system integrity is lost, then confidence supporting services

in the validity of the mechanisms imple-
menting these parameters is also lost. Supporting services (Stoneburner, 2001) are, by
their very nature, pervasive and inter-related with
All of these parameters are interdependent with many other services. The supporting services are:
Assurance. When designing a system, an architect
or engineer establishes an assurance level as a • Identification (and naming): This ser-
target. This target is achieved by both defining vice helps in uniquely identifying users,
and meeting the functionality requirements in processes, and information resources.
each of the other four parameters and doing so Typically this service in turn is used by
with sufficient “quality”. Assurance highlights other services to identify subjects and
the fact that for a system to be secure, it must not objects.
only provide the intended functionality, but also • Cryptographic key management: This
ensure that undesired actions do not occur. service enables secure key management
which plays a key role in implementing
security services other services where cryptographic func-
tions are used.
Security services (Stoneburner, 2001) used in • Security administration: This service en-
implementing an information technology security ables efficient administration of security
capability can be classified into following three features of an information system to meet
broad categories. the security requirements of a specific in-
stallation and also to account for constant-
• Support: Comprises mainly of generic ly evolving operational environment.
services which underlie most information • System protections: This service is based
technology security capabilities. on the notion of confidence in the techni-
• Prevent: Comprises of services which pre- cal implementation of various security
vents security breaches from occurring functional capabilities. The quality of the
• Recover: Comprises of services which fo- implementation is measured from both the
cus on the detection and recovery from a perspective of the design processes used
security breach. and the specific manner in which the im-
plementation was accomplished. Examples
of system protections are: residual infor-
161
mation protection, least privilege, process transacting with information systems. This
separation, modularity, layering, and mini- has become an essential requirement for
mization of what needs to be trusted. enterprises both in government and private
sectors.
Prevention services
detection and recovery services
These services (Stoneburner, 2001) aim at prevent-
ing the security breaches from ever happening. Because no system can be made absolute secure,
it is necessary to both detect security breaches and
• Protected communications: In a distrib- to minimize their impact (Stoneburner, 2001).
uted system, the ability to accomplish se-
curity objectives is highly dependent on • Audit: This service enables auditing of se-
secure communications which helps in curity relevant events for intrusion foren-
achieving integrity, availability, and confi- sics and recovery from security breaches.
dentiality of information while in transit. • Intrusion detection and containment:
• Authentication: This service helps in en- This service provides ability to timely de-
suring that a claimed identity is valid. tect signs of intrusions into information
• Authorization: This service enables spec- systems and also to take counter measures
ification and subsequent management of to confine it and minimize the effect.
the allowed actions of a given information • Proof of Wholeness: This service pro-
system. vides the ability to determine that integ-
• Access control enforcement: This service rity of an information system has been
enables enforcement of defined security compromised.
policy of the enterprise, when any subject • Restore ‘secure’ state: This service pro-
requesting access to particular object has vides a way to restore the system state to a
been validated. The level of security ob- known secure state in the event of a secu-
tained in any information system depends rity breach.
not only on the correctness of the access
control decision, but also the strength of solutIons And
the access control enforcement. A common recommendAtIons
access control enforcement mechanism
is to check identity and requested access enterprise Information security
against access control lists. File encryption engineering lifecycle
is another example of an access control en-
forcement mechanism. Information Security (IS) of Enterprises depends
• Non-repudiation: This service enables on several factors. The major determinant is the
the ability to ensure that senders cannot Business Goal; besides, the operation context,
deny sending information and that receiv- technology used, organizational structures and
ers cannot deny receiving it. This service is connectivity also play important roles in determin-
treated as a prevention service because it ing the approach towards IS. The other important
prevents the ability to successfully repudi- concern is that the IS need of an Enterprise is not
ate an action. static, but depends on the dynamics of operation,
• Transaction privacy: This service helps changing business goals, technological upgrades,
in maintaining the privacy of individuals etc. The process of developing and deploying a
162
proper IS infrastructure for an Enterprise is not a This leads to the concept of Information Secu-
one-time affair; rather it is a continuous process rity Engineering Life-Cycle, or simply “Security
of analysis, design, monitoring and adaptation to Engg. Life-Cycle” (Anderson, 2001; Sengupta,
changing needs. In many enterprises, the changes 2005). Following phases are considered to be
encountered are frequent. This calls for structured of prime importance for this Life-cycle concept
representation of the different specification docu- (refer Figure 2).
ments and automatic analysis and generation of
them with interoperable representation.
Figure 2. Security Engineering life cycle
163
security requirement Analysis • A structured Security specification format

• The Risk Analyses Report
The very first step in building up a (somewhat)
secure system is to understand and correctly security Policy Formulation Phase
specify the requirement. In this phase, information
regarding the business modalities and IT assets The goal of the security policy is to translate,
of an enterprise are extracted. clarify and communicate management’s posi-
The categories of information extracted from tion on security as defined in high-level security
the enterprise are: principles. A security policy manual (Peltier,
1999) is a set of high level statements describing
• The modes of activity of the business. the objectives, beliefs and goals of the organiza-
• The assets including the information asset, tion as far as security is concerned, but not how
software asset, hardware asset and service these solutions are engineered and implemented.
asset and their security requirements vis- Security policies define the overall security
à-vis confidentiality, integrity, availability, and risk control objectives that an organization
authentication, non-repudiation, legal and endorses. The policy should be implement-able
contractual obligation and the perceived and enforceable; they should assign responsibility
security threats to the assets. and accountability. They must be documented,
• The connectivity details. distributed, and communicated. They should be
• The access control requirements on the ba- technology-independent as much as possible and
sis of “need-to-know” and “need-to-use”. should not change often.
In this phase, the Security requirement Specifi-
Based the on the above information provided cation document is parsed to determine the security
by the enterprise, an analysis is performed to needs of an enterprise. Based on these security
identify the required security controls (this in- needs, a model policy manual is generated which,
cludes technical as well as managerial controls, should conform to a particular information security
as suggested by security standards such as ISO standard like ISO 17799 (Plate, 2005). Guidelines
17799 (Plate, 2005)). An overall Risk Analysis and procedures, required to implement the policies,
(Peltier, 2005) is also performed to categorize the are also generated. Procedures consist of detailed
assets into high, medium and low risk zones for steps to implement the policies, while guidelines
implementation of the security controls. During are optional guidance provided to an enterprise
risk analysis, an attempt is made to identify the to help in the implementation of policies and
threats to which an IT system is exposed due to procedures. The model policy document should
existing security weaknesses. The probability of be analyzed by the management in conjunction
each of these threats occurring is then estimated with the non-security enterprise policies in place
and combined with the protection requirements to finalize the Security Policy document.
to rate the existing risks. The reports generated in this phase are:
For any risks which are unacceptable a set of IT
security measures is then selected so as to reduce • The policy manual
the probability of occurrence and/or the extent • The guideline manual
of the potential damage. The OCTAVE Method • The procedures manual
(Alberts, 2001) and the TEN STEP Process are
two well-known risk analysis methodologies.
The reports generated in this phase are:
164
security Infrastructure security testing and validation

Advisory Phase
The main objective of security testing is to ascer-
Security infrastructure advisory is a specified set tain that the proposed security infrastructure is in
of entities, both physical as well as software, in place and working. Other than this, the system
order to implement the set of identified security can be tested for faulty combination of software,
controls. It tells an enterprise the details regard- known security holes and potentially dangerous
ing the software necessary, the access rights of applications that can be compromised to breach
individuals, the exact location of security tools, the security infrastructure. An integrated security
etc., required to mitigate the security risks of testing tool may be visualized to automate the
the enterprise. This phase will parse the security objective of security testing as described above.
requirement specification document to identify The security testing phase can be further di-
the security needs of an enterprise and generate vided into several sub-phases as described below.
a security infrastructure advisory. The advisory
will list the security tools required by the orga- • Test Case Generation - A test case can be
nization in order to protect its assets. The exact viewed as a collection of information based
installation location and configuration advice are on which the security test is conducted.
also included in the advisory. The report generated The preparation of a test case requires in-
in this phase is put from a requirement specification, valid
security infrastructure selection, database
• The security infrastructure advisory (vulnerability database) containing infor-
mation on known security holes and inputs
Some information security standards suggest from the security analyst. The test plan
probable security tools which are to be installed would be generated in a domain-specific
in an enterprise in order to protect its assets. For format. Test cases are generated from the
instance, ISO 17799 (Plate, 2005) consists of test plan templates.
security controls which detail the types of assets • Compliance Test - In this sub-phase the
and corresponding security parameters (confi- security test tool determines whether the
dentiality, integrity, etc.) to be protected, and the specified security infrastructure is in place
security tools needed for the same. with proper configuration. The security
After getting the security infrastructure speci- analyst would have the provision on de-
fication, a cost-benefit and detailed risk analysis ciding whether to proceed with the later
is performed by the management of the enterprise sub-phases of the security test if the secu-
concerned. Based on this, the enterprise decides rity infrastructure is found to be not im-
on the particular infrastructure that it would like to plemented, partially implemented or fully
implement. This is stated in the form of a security implemented.
infrastructure selection document. This selection • Common Vulnerability Test - After testing
will take into account all kinds of Policies of the the existence of the security infrastructure
enterprise. Finally, the enterprise implements the in a corresponding domain, several tests
security infrastructure as decided by it. would be conducted to find out the pres-
ence of known vulnerabilities that may be
used to compromise the system even in the
presence of a security infrastructure.
165
• Validation Report Generation - It is the last Future trends

sub-phase in the security test mechanism.
It is used to generate a validation report The future trend in Enterprise Information Security
for the security requirement in the form of approach is to overlay the Plan-Do-Check-Act
compliance and gaps in the implemented (PDCA) Cycle on the Security Engineering Life-
system. This is carried out by applying the cycle. In fact the various Information Security
Risk Analysis methodology on the secu- Management Standards are trying to spell out the
rity infrastructure in place, to see how far activities to be taken up in different phases of the
the risks are mitigated, thus, justifying the PDCA Cycle and their outcome. A major paradigm
ROI. The reports generated in this phase shift is happening from the qualitative approach
are: to the monitoring with quantitative measures. The
• Compliance test report quantitative measures are being proposed to judge
• Vulnerability test report the quality of security management solutions in
• Validation report a more objective manner. This would help also
in justifying the ROI. As already mentioned, the
review and monitoring technology trend is to integrate security func-
tionalities with the operational functionalities of
The different phases of activities may have to components and subsystems – both hardware and
be monitored and the reports may have to be software. This would cater for allowing secure
reviewed. The Monitoring process is continuous transactions only in order to secure rather open
and may need certain tools and procedures to enterprises, where the boundaries are becoming
generate monitoring reports. The Review process fuzzier, because of the globalization of economy
involves the review of the requirement, policy, in the cyber space.
infrastructure and testing methodologies, so that
the different entities do not become outdated and,
consequently, create holes in the security frame- conclusIon
work. The review may be triggered by changes
in the organizational structure, business goals and This chapter looks at different aspects of Enterprise
activity, technology, legal obligations, etc. Even in Information System Security. We began by defin-
the absence of such triggers, review may be done ing an enterprise and its information systems and
periodically, the period being fixed by a policy. discussed the need for securing the same. Then
The other important thing triggering a review is the problems associated with the implementation
the report of a security incident, which may affect, of enterprise information security were discussed.
or have already affected, the enterprise security. The various security parameters, and their inter-
The outcome of the Review and Monitoring dependencies, were stated. Finally, a management
may re-initiate the activities of the Requirement approach to securing an enterprise was proposed
analysis, Policy Formulation, Risk Analysis, and a security engineering life-cycle methodology
Infrastructure updation and Testing & Validation to achieve the same was discussed.
phases. This is how the Security Engg. Life-cycle
is expected to work.
166
AcknoWledGment Denning, D. E. (1999). Information Warfare and

Security. Singapore: Addison Wesley.
The authors acknowledge the role of over a
Goguen, J. A., & Meseguer, J. (1982). Security
dozen students and researchers in the Centre for
Policies and Security Models, Symposium on
Distributed Computing, Jadavpur University, for
Security and Privacy, pp.11-20. Menlo Park,
their contributions over the years in shaping and
CA: IEEE.
clarifying the concepts presented in this chapter.
Most of the work related to the concepts presented Hassler, V. (2001). Security Fundamentals for
in this chapter has evolved during the execution E-Commerce. Norwood, MA: Artech House Inc.
of a couple of R&D Projects sponsored by the
Lampson, B. W. (1974). Protection. Proc. Fifth
Dept. of IT, Ministry of Communications and
Princeton Symposium on Information Sciences
IT, Govt. of India, and a number of consultancy
and Systems. Princeton University, reprinted in
projects taken up by the Centre.
Operating Systems Review, 8(1), 18-24. Palo
Alto, CA.
reFerences Mayfield, T., Roskos, J. E., Welke, S. R., & Boone,
J. M. (1991). Integrity in Automated Information
Alberts, C., & Dorofee, A. (2001). An Introduction Systems. Prepared for National Computer Secu-
to the OCTAVE Method, Software Engineering rity Center (NCSC). Alexandria, Virginia: NCSC.
Institute, Carnegie Mellon University. Retrieved
October 11, 2001, from http://www.cert.org/ Peltier, T. R. (1999). Information Security Policies
octave/methodintro.html. and Procedures. Boca Raton, Florida: Auerbach
Publications.
Anderson, R. (2001). Security Engineering: A
Guide to Building Dependable Systems. New Peltier, T. R. (2005). Information Security Risk
York, NY: John Wiley & Sons, Inc. Analysis. Boca Raton, Florida: Auerbach Pub-
lications.
Bell, D. E., & LaPadula, L. J. (1976). Secure
Computer Systems: Unified Exposition and Mul- Plate, A., & Weissmann, O. (Eds.). (2005). Infor-
tics Interpretation, ESD-TR-75-306, MTR 2997 mation Technology – Code of practice for informa-
Rev. 1. Bedford, Massachusetts: The MITRE tion security management, ISO/IEC 17799:2005.
Corporation. Berlin, Germany: ISO.
Biba, K. J. (1975). Integrity Considerations for Richardson, R. (2003). CSI/FBI Computer Crime
Secure Computer Systems, MTR-3153. Bedford, and Security Survey. Hayward, CA: Computer
MA: MITRE Corporation. Security Institute.
Casar, E., Scheer-Gumm, G., & Simons-Felwor, Sengupta, A., Mukhopadhyay, A., Ray, K., Roy,
P. (2004). IT Baseline Protection Manual. Ger- A. G., Aich, D., Barik, M. S., & Mazumdar, C.
many: BSI. (2005). A Web-Enabled Enterprise Security Man-
agement Framework Based on a Unified Model
Clark, D., & Wilson, D. (1989). Evolution of a of Enterprise Information System Security: (An
Model for Computer Integrity, in Report of the Ongoing Project Report), First International Con-
Invitational Workshop on Data Integrity, Gaithers- ference on Information Systems Security, ICISS
burg, Maryland, A.2-1A.2-13. Gaithersburg, MD, 2005. Kolkata, India, Pages 328 – 331: LNCS.
National Institute of Standards and Technology,
NIST Special Publication 500-168: NIST.
167
Snare, J., & Kuiper, E. (Eds.). (2005). Information fax machines), magnetic media (tapes and disks),
technology - Security techniques – Information other equipment, cabinets, safes.
security management systems - Requirements, BS Risk Analysis: This term defines the process
ISO/IEC 27001:2005 BS 7799-2:2005. Geneva, of analyzing a target environment and the relation-
Switzerland: ISO. ships of its risk related attributes. This analysis
will identify threat-vulnerabilities, associate those
Stoneburner, G. (2001). Underlying Technical
vulnerabilities with affected assets, identify the
Models for Information Technology Security.
potential for and nature of an undesirable result
NIST Special Publication 800-33. Gaithersburg,
and specify risk mitigating controls.
MD: NIST.
Risk Management: This includes the process
Weiner, E. S. C., & Simpson, J. A. (Eds.). (1991). of assigning priority to, budgeting, implementing
The Compact Edition of The Oxford English Dic- and maintaining the appropriate risk-reducing
tionary. USA: Oxford University Press. measures.
Safeguard: This term represents a risk re-
Yu, C.-F., & Gligor, V. D. (1990). A Specifica-
ducing measure that acts to detect, prevent or
tion and Verification Method for Preventing
minimize loss associated with the occurrence of
Denial of Service. [IEEE.]. IEEE Transactions
a specified threat.
on Software Engineering, 16(6), 581–592.
Security Concern: Security concern of an
doi:10.1109/32.55087
asset is a function of threat and vulnerability of
that asset.
Severity: Level of exploitation of vulner-
key terms And deFInItIons ability on a qualitative scale is defined by the
severity value.
Asset: Asset means anything that has value Software Asset: Application software, System
to an organization. With respect to security, as- software, development tools, and utilities.
set may imply physical resources, or information Threat: Threats are any unwanted activities
contained within the organization. or events that under certain conditions could
Information Asset: Databases, data files, jeopardize either the integrity, confidentiality or
system documentation, user manuals, training availability of information and other assets.
material, operational and support procedures, Vulnerability: This is an inherent weakness
intellectual property, continuity plans, fallback associated with an enterprise asset. It is a condition
arrangements, archived information. that has the potential to allow a threat to occur
Hardware Asset: Computer equipment with greater impact and greater frequency or both.
(processors, monitors, laptops, modems), com-
munication equipment (routers, hubs, PABXs,
This work was previously published in Handbook of Research on Social and Organizational Liabilities in Information Security,
edited by Manish Gupta and Raj Sharman, pp. 118-132, copyright 2009 by Information Science Reference (an imprint of IGI
Global).
168
169
Chapter 1.12
From ERP to Enterprise Service-
Oriented Architecture
Valentin Nicolescu
Technische Universität München, Germany
Holger Wittges
Helmut Krcmar
AbstrAct landscapes and, on the other hand, to show the

comparison of the presented concepts with respect
This chapter provides an overview of past and to application and technology view.
present development in technical platforms of
ERP systems and its use in enterprises. Taking
into consideration the two layers of application IntroductIon
and technology, we present the classical scenario
of an ERP system as a monolithic application With the emergence of the SOA concept, the
block. As the demands of modern enterprise classical architecture of ERP system has started
software cannot be met by this concept, the shift to change and is in a constant flux towards new
to a more flexible architecture like the service- structures. We want to show these changes, starting
oriented architecture (SOA) is the current status with the architecture of ERP systems and describ-
quo of modern companies. Keeping in mind the ing the different parts of this concept. To exemplify
administrative complexity of such structures, we it, we will present the most important aspects of
will discuss the new idea of business Webs. The concrete implementations of these principles. As
purpose of our chapter is, on the one hand, to one of the most important ERP systems, we will
show the historical development of ERP system focus on the structure of the SAP ERP system and
will describe the changes of this platform.
DOI: 10.4018/978-1-59904-859-8.ch023
From ERP to Enterprise Service-Oriented Architecture
Our analysis will comprise of an application- focus on data integration and also support process
centered and a technical view, considering changes integration within one company.
in business paradigms and new technologies that The technical basis of an ERP system today is
enabled new kinds of business and process man- usually a Client/Server architecture, where often
agement. Starting at classical ERP systems and more than one application server is connected to
their implementation in SAP R/3, we will move the central database server. The user of a classical
on to the current concept of SOA and Enterprise ERP system most time works with basic business
SOA. This goes along with a change in technical transactions like “create order”, “update customer
architectures as well. The SAP NetWeaver plat- contact data”, “print invoice”, “execute report xy”
form will be presented as an example of complete etc. Changes within such a system due to business
Enterprise SOA platforms. Its most important transactions are usually propagated in “business”
functions will be pointed out and utilizing this realtime meaning a few seconds or minutes.
example, components that are necessary to realize The following figure from Davenport visual-
Enterprise SOA are identified. The light in which izes the architecture of a classical ERP system.
SAP NetWeaver is seen has changed in the last In addition to the facts mentioned before, there
years as not the technical components are in the is usually a wide range of reporting functionality
spotlight anymore but the applications that are for management and stakeholders, based on the
made possible by such a platform. Finally, we will central ERP data.
show the future concept of business webs which There are a lot of advantages that unfold with
will base on Enterprise SOA and conclude our the use of ERP Systems (Vogel and Kimbell 2005;
chapter with an outlook to the further development N.N. 2007) In the following a few of them are
in this area of topic. The structure of our chapter spotlighted:
is illustrated in Figure 1.
• Standardisation
• Integrated “best practice” business
clAssIcAl erP systems knowledge
• Data quality
Classical ERP system can be described as com- • Data- and process integration
mercial software products that are adaptable to • Central authorisation, authentication
company-specific demands. Their typical func-
tional modules include: purchasing, manufac- However, there are also a lot of potential risks
turing, sales, finances, human resources, service related with the use of ERP-Systems.
and in general reporting. Classical ERP systems
Figure 1. Structure of this chapter
170
• Possible single point of failure sAP r/3 bAsIs

• Problems integrating different
ERP-Systems For the discussion of the technology of classical
• Tracking of complex processes – even if ERP systems, we will take a closer look at SAP
they are handled by only one ERP-System R/3 and the underlying technological platform
• operated by personnel with inadequate ed- SAP Basis. Since SAP’s application was the first
ucation in ERP in general enterprise software fulfilling the characteristics
of an ERP system and still represents the leading
Additional, considerable risks that need to be edge in ERP technology, analyzing this product
especially emphasized arise from deficiencies in provides a complete outline of the past and present
the ERP introduction project. Example thereof development in ERP system technology.
could be: SAP R/2 was the first enterprise application in-
tegrating functions of different enterprise divisions
• mistakes in the selection of the ERP in one piece of software and thus representing the
System, first ERP system. It provided functions for finance,
• misunderstandings/mistakes in the system accounting, human resources, sales, procurement,
customizing and manufacturing. The system ran on mainframe
• failures in the business / IT alignment architectures, while its successor, R/3, relies on
a client server architecture and also includes a
Despite the possible problems that served as whole set of new functions. (Vogel and Kimbell
expamples, ERP systems can be seen as something 2005). Because SAP R/3 for a long time was the
that Carr calls ubiquity information systems ERP system, it shall serve as our starting point
power (Carr 2003). Nowadays it is not the ques- for the detailed analysis of the technology layer
tion of using an ERP System in your company, Moving from a mainframe structure to the
but how. There is also research available describ- client server architecture the technical part of
ing the value of ERP Systems more concretely. the SAP application was divided in a 3-tier land-
See for example (Martin, Mauterer et al., 2002). scape consisting of the front end tier typically
running on the user’s desktop, the application
Figure 2. Anatomy of an enterprise system (Davenport, 1998)
171
tier and the database tier. The front end tier is a new products were based on SAP Basis. This is
fat client that displays all kinds of screens based why a highly integrated system landscape could
on small information packages exchanged with be implemented despite the different systems.
the application tier. The application tier resides By using the same technical layer, all systems
on high performance servers and represents the can communicate and exchange data very easily
application logic and enterprise functionality as utilizing the same functions. These functions are
well as the connection to the database layer. The arranged in a hierarchical structure defined by
databases used for R/3 have been provided by business areas and this way facilitate changes and
different manufacturers and, with their database enhancements of business processes. Furthermore,
product SAPDB, also by SAP itself. By dividing most functions are encapsulated in so called func-
the technical architecture into these three tiers, it tion modules which can be called remotely by
became scalable and could be enhanced by adding another application like another SAP system or
more desktop computers or servers to the tier that a third party product. Therefore communication
lacks computational power. with applications of third party vendors can be
The technical layer of the application tier and established within limitations. This packaging
thus of SAP R/3 systems is called SAP Basis. of functions is also an important step towards
This layer’s characteristics of capital importance a service oriented architecture as will be shown
indclude platform and database independence with later (Buck-Emden 1998).
respect to the interpretation of application source Although the SAP Basis of SAP R/3, as shortly
code. The code is written in SAP’s programming presented above, was already quite flexible and
language called Advanced Business Application enabled the communication to other applications
Programming (ABAP). All of SAP’s out of the box to and from SAP or third party vendors, the degree
functions can be viewed in plain text, edited (with of integration was limited. There was no complete
some limitations) and hence also be enhanced. The integration of functions and data of different
integrated development environment called ABAP applications to create completely new business
workbench can be used for these tasks. Providing processes across system boundaries. Addition-
these possibilities, the platform can be adopted in ally, changing or enhancing crucial functions or
a very flexible way to individual needs. In practice processes was very difficult as the dependencies
however, only few modifications are made to the between systems and function modules were not
source code delivered by SAP because the algo- clearly documented. A fundamental problem of
rithms represent the know-how of many business this technical layer was also its integration into
experts. Framed in other words, the implemented the application layer: SAP Basis could not be used
algorithms are procedures cast into best practices without SAP R/3 or other products based on this
such as how to support a specific process at best. very platform. These issues forced a fundamental
When using SAP R/3 for crucial functions, change to provide a much more flexible technical
enterprises have a data centralization spanning the layer for future business applications.
whole company. Nonetheless SAP R/3 turned out
to be insufficient for the integration of suppliers soA and enterprise soA
and customers as well as for performing in-depth
analysis of business data. Thus new applications We will rely on the following definition of a SOA:
like SAP Customer Relationship Management “SOA is an architectural style whose goal is to
(CRM), Supplier Relationship Management achieve loose coupling among interacting software
(SCM) or the Business Warehouse (BW) were agents. A service is a unit of work done by a ser-
created to meet the growing demands. All these vice provider to achieve desired end results for a
172
service consumer. Both provider and consumer are Key features of SOA as described by Erl (Erl
roles played by software agents on behalf of their 2005) are:
owners.“ (He 2003) In addition to this definition,
SOA is characterized by intensive use of standards • Self-describing: The service can be fully
(like SOAP, UDDI, etc.) to implement services. described on a formal level; that is, what
But the concept of SOA itself is not connected to a operations are available and which data
technical or business domain. It is a concept of how types can be exchanged. Formal self-de-
functions can be implemented on distributed enti- scription is the basis for automatically gen-
ties, just like the client-server concept or the idea erating proxies. Note that in this context
of object orientation (OO). Both of the mentioned the word “description” does not refer to the
actually can be related to SOA. Services within semantic description.
a SOA could for example be implemented using • Locatable: Potential consumers can locate
an OO programming language. The result could and contact the service. There is a registry
be run technically in an environment that follows for this purpose that operates as a “Yellow
the client server concept. The central components Pages” of services and allows users to
of a SOA are shown in Figure 3. search for services.
The central element within a SOA is the service. • Roughly structured: Services are roughly
Whenever an application wants to use a service, it structured when they return comprehen-
communicates with the service repository (for sive data (for example, a sales order) in re-
example via UDDI) using the (enterprise) service sponse to a service call instead of deliver-
bus as technical backbone. Examples for service ing individual attributes (such as an order
busses include IBM WebSphere ESB, Microsoft quantity).
biztalk, Oracle ESB and SAP XI.. The application • Loosely linked: Services are said to be
front end is used to access the network of services. loosely linked if potential functional
A service itself consists of an implementation changes within the services have no effect
describing the business logic and data. Also it is on the availability of operations. Services
formalized by a service contract (i.e. IDL, WSDL) are considered to be independent if they
and a service interface (i.e. SOAP) (Booth, Haas have no dependencies on other services
et al., 2004). i.e. if the availability of a service is not af-
Figure 3. The key abstractions of a SOA (Krafzig, Blanke et al., 2004)
173
fected by a potential unavailability or im- NetWeaver is comparable to the SAP R/3 Basis,
proper functioning of any other service. however doess not itself provide any business
• Reusable: Reusability is one of the most functions but is the technical basis for enterprise
important and longest-standing require- applications like SAP Enterprise Resource Plan-
ments in software technology. It aims at ning (ERP) (Woods and Word 2004). Since the
ensuring that components are reused as of- components of this layer are already a few years
ten as possible without modification to the old, the view on this platform has changed. First
functional core or interfaces. we will present the classical component based
view covering the specific technical systems that
Enterprise SOA extends the features of the are part of this platform. Afterwards we will show
SOA concept with a business view on services. the shift from the component based to an applica-
Therefore services additionally have the follow- tion based view that does not care anymore about
ing key features: technical systems but focuses on the application
of a set of systems for business purposes.
• Services represent business services in-
stead of technical services component based view
• Support of inter-organizational collabora- on netWeaver
tion on a business level
The component based view on SAP NetWeaver is
The technical basis of an Enterprise SOA to- also called the “NetWeaver Fridge” because you
day is usually a client-server architecture, where can use specific components of the NetWeaver
many application server and many database serv- platform just as you can take ingredients from
ers are part of the - often regionally distributed your fridge for lunch. This technical fridge con-
- IT-landscape. In contrast to a list of basic ERP sists of four layers representing different aspects
Transactions, Enterprise SOA offers access to high of an enterprise system landscape as well as
level business processes like “hire employee”, two cross-functional areas that are used for all
“produce good”, “distribute good” etc. Seen in other products. We can distinguish the layers
this light one might define Enterprise SOA as the for application platform, process integration,
business glue connecting isolated business tasks, information integration and people integration.
such as the previously mentioned examples, to The general areas are life cycle management and
comprehensive business processes. Working that SAP’s Composite Application Framework (CAF).
way, Enterprise SOA opens the door to realtime The specific layers combine different software
business processes instead of realtime transactions. products of the SAP portfolio. Thus this view on
NetWeaver is a rather technical one (Karch, Heilig
et al., 2005). In Figure 4 you find an overview of
dIFFerent vIeWs on all layers and their elements at which we want to
sAP netWeAver have a closer look.
The application platform consists of three
The presented idea of service oriented architec- elements whereas the DB and OS abstraction is
tures and its use in an enterprise domain forces an integral part of the two other elements. These
the need for a technical platform that can enable two elements are the ABAP and the Java core of
the interaction of the different services. SAP all SAP applications. They are called Web Ap-
developed the currently most complete technical plication Server (Web AS) ABAP and Web AS
layer for Enterprise SOA called SAP NetWeaver. Java. Depending on the application that is to be
174
Figure 4. Component based view SAP NetWeaver
executed on the Web AS, either one or both cores ment automated processes by triggering and
are used as a runtime environment. The SAP coordinating data exchange between all kinds
NetWeaver Portal for example needs only the Java of enterprise systems which can be accessed by
core while SAP Process Integration uses both Java the integration broker. A business process can
and ABAP. The Web AS can also be used for therefore automatically manage the selection of
technical purposes like monitoring without enter- the supplier with the best price after receiving all
prise applications building upon it (Heinemann quotations and may subsequently place an order.
and Rau 2003). NetWeaver ‘s process integration layer and the
The process integration layer represents the product of the same name can be regarded as the
elements of an integration broker and of business communication backbone of a modern enterprise
process management which are combined in SAP’s system landscape. It provides centralized functions
product called Process Integration (PI, formerly for data exchange within the landscape and with
Exchange Infrastructure). The integration broker’s systems in supplier or customer domains.
task within NetWeaver is the exchange of data The information integration layer combines
using different formats and protocols. It can for three tasks which are arranged in three different
example convert messages sent via FTP in an applications. Master Data Management (MDM)
XML format to a message for an SAP function and Business Intelligence (BI) can be found in the
module based on one of SAP’s internal protocols. products of the same name while Knowledge Man-
Besides this technical conversion, also changes to agement (KM) is part of the SAP NetWeaver Portal
the content can be processed such as converting which is part of the subsequent layer. MDM helps
the values of the gender (female, male) to values to consolidate data within an enterprise across
for a suitable salutation (Ms, Mr.). Business Pro- system boundaries. The same data is often saved
cess Management (BPM), also known as cross- in different places because of technical reasons.
component BPM (ccBPM), provides functions In many cases this redundancy can be changed so
for workflow management across boundaries of that the data is saved only in one place or at least
single enterprise systems. It can help to imple- synchronization between different databases can
175
be set up. Thus the same data is available in all and maintenance, upgrade and migration tasks.
connected systems. Using this correct data proper The SAP product supporting these tasks is the
propped business decisions can be made. As the SAP Solution Manager which provides part of
amount of data on which decisions are based on these functions as a first step towards the imple-
can be vast, SAP’s Business Intelligence helps to mentation of ITIL.
arrange and visualize data from different points As a SOA consists of many different small
of view. By using SAP’s Strategic Enterprise services, an integration layer for coordinating the
Management (SEM), balanced scorecards can be steps within a business process is needed. This
used as an enterprise dash board or management layer is the Composite Application Framework
cockpit. While MDM and BI are used for structured (CAF) which allows the development of appli-
data, KM components help to manage unstruc- cations based on the underlying services - web
tured data like files containing different kinds services or SAP function modules provided by
of information. As KM is part of the NetWeaver SAP applications. The implementation of a persis-
Portal it can easily be integrated in the front end tence layer as well as the design of user interfaces
and therefore in daily work. are as well part of this development. In order to
In the layer of people integration, three dif- access backend systems, persistence and user
ferent tasks are combined in 2 applications. On interfaces, the appropriate Java source code can
the one hand one can identify the multi channel be generated. This code is running within the SAP
access provided by SAP’s Mobile Infrastructure NetWeaver Portal providing the connectivity of a
(MI). On the other hand there is SAP NetWeaver Web AS Java and the user interface of the portal
Portal including the tasks of portal, collaboration framework. CAF can be used to have professional
and knowledge management which form the developers generate Java-based applications or
information integration layer. This layer hence to quickly implement an adhoc-workflow called
contains applications for end user communication. guided procedure. By relying on other NetWeaver
The Mobile Infrastructure provides a client server products, a new CAF application or a guided pro-
architecture that enables the synchronization of cedure can take advantage of existing functions
data on mobile devices like PDAs or Notebooks like the connection to third party systems or user
with the enterprise system landscape. This way interface functions.
traveling salesmen have all the data they need to
help customers place an order. The portal itself Application based view
provides a web based framework for arranging and on netWeaver
displaying screens and information. Furthermore
it integrates business process screens providing a As we already pointed out, the component based
single point of entry to front ends in the enterprise view on the NetWeaver fridge is dominated by
system landscape. It can be used to implement a technical aspect. For most scenarios more than
flexible wizards working across system bound- one of these products is necessary to attain the
aries. These basic functions can be enhanced by goal. Thus the NetWeaver fridge has been cut
synchronous or asynchronous collaboration and into slices to show that the NetWeaver platform
knowledge management. does not exist for its own sake, but to enable and
The cross function area of life cycle man- support different IT practices (see Figure 5).
agement covers all activities for running an ap- The result of cutting the NetWeaver fridge into
plication in an enterprise system landscape. This slices are 10 IT practices which shall be discussed
includes the installation preparation of specific now. Each of the practices itself is divided into
applications, customizing activities, monitoring IT scenarios describing concrete activities to
176
Figure 5. Shift from component based view to application based view on NetWeaver
achieve the target of a scenario, for example 4. Unlike the business information manage-
implementing single sign-on for an application ment that provides functions for proactive
or in a whole system landscape (Nicolescu, Funk work with business data, the business event
et al., 2006). management works by a Push technology.
Whenever important events happen in the
1. The first IT practice called user productiv- company the relevant people will be in-
ity enablement aims at improving the daily formed and can react. Reacting on events is
work of users by providing tools to exchange abstracted to the necessary data and entry to
information between them, any by means of appropriate business transactions for a given
personalization for an easier access to data situation. Main components of this practice
and business processes. This IT practice are again SAP Business Intelligence and SAP
mainly builds upon the SAP NetWeaver NetWeaver Portal.
Portal. 5. End-to-end process integration allows the
2. The Practice of data unification tries to implementation and monitoring of business
consolidate and harmonize data that is stored processes across system boundaries within a
in different databases all over the enterprise. company’s own system landscape and with
As this task is mainly connected in identify- systems of business partners. This way, cru-
ing redundancy and keeping data in different cial processes can be monitored and adjusted
places synchronous, the vital application of centrally. The component providing these
this task is the Master Data Management features is SAP Process Integration.
(MDM). 6. Development and adaptation of enter-
3. Business information management is an prise applications represents the sixth
important support task for decision makers IT practice which includes the creation of
in a company. This practice tries to provide completely new software as well as changing
structured and unstructured information to and enhancing existing functions. Because
make the right decision based on the given development in an enterprise environment
data. As this combines elements of Business poses many challenges, it is supported by
Intelligence and Knowledge Management, components like the Composite Application
BI and portal are the main components of Framework.
this practice.
177
7. The IT Infrastructure Library (ITIL) plays 10. The last IT practice enterprise service-ori-
an important role in today’s IT management, ented architecture design and deployment
which is why the IT practice of unified covers the areas of planning, developing,
life cycle management covers most of its reusing und maintaining SOA applications.
concepts. An important building block for By separating the formerly monolithic ap-
attaining this goal is the use of the SAP plications into smaller services the adminis-
Solution Manager. trative effort to keep track of these modules
8. The topic of application governance and and to wisely reuse them is a major issue for
security management is covered by the development.
eighth IT practice. It deals with the holistic
management of applications within a system busIness Webs
landscape, addressing issues like commu-
nication and security of the systems. The “Low-Cost Information and communication tech-
Concept of implementing single sign-on nologies, global markets, and global competition
is just one example. As this task affects all have forced enterprises to rethink their traditional,
systems in an enterprise, there is no specific vertically integrated structure. As the cost of col-
component to point out. laboration plummets, it increasingly makes sense
9. While the presented IT practice of data for companies to focus on core competencies
unification aims at the consolidation of data, and partners for all other functions. As a result,
the IT practice of consolidation deals with corporations are transforming into flatter and
all approaches to simplify the whole system more specialized enterprises, linking with sup-
landscape. Examples of these approaches pliers and other businesses to form a larger, more
are service oriented architecture or server open, value-creation entity” (Tapscott 2007) – the
virtualization. business web.
Figure 6. Line 56 – ecosystem (56 o.D.)
178
The innovation potential and practical rel- nally value is generated by your community if
evance of the business web is a recent topic for they use your enterprises services within the busi-
ERP Software providers. SAP (Karch, Heilig ness web to enable their own business processes.
et al., 2005; Tapscott 2007) and salesforce.com
(Gilbert 2006) described it as an important shift
in the future similar to the migration from the reFerences
classical world wide web to Web 2.0.
The business web can only be realized if an Booth, D., H. Haas, et al., (2004). Web Services
adequate ecosystem has been set up. The following Architecture. W3C.
figure illustrates such an ecosystem. Buck-Emden, R. (1998). Die Technologie des
In addition to the well know eCommerce or SAP-Systems R/3. München, Addison-Wesley.
supply chain models this ecosystem not only fo-
cuses on special partners but instead on a whole Carr, N. G. (2003). IT doesn‘t matter. Harvard
community. The idea of integrating virtually Business Review, 81(5), 41–49.
everything is not a new one, but it often failed
Davenport, T. H. (1998). Putting the Enterprise
due to high complexity. Enterprise SOA now of-
into the Enterprise System. Harvard Business
fers a concept and appendant tools that allow to
Review, 76(4), 121–131.
manage complexity by partitioning and linking
the emerging parts (partner processes) together. Erl, T. (2005). Service-Oriented Architecture -
Concepts, Technology, and Design. Upper Saddle
River NJ: Prentice Hall.
conclusIon Gilbert, A. (2006). Salesforce CEO’s vision for
‘business Web’. www.news.com. DOI
Companies try to save investments in old appli-
cations by wrapping old code into web services. He, H. (2003). What Is Service-Oriented Archi-
This can be described as a three step approach: tecture.www.xml.org Retrieved 14.06.2007, 2007,
Legacy to SOA is the first step to go. The idea from http://www.xml.com/lpt/a/1292.
is to make your business transactions accessible,
Heinemann, F., & Rau, C. (2003). SAP Web Apli-
independent of the platform they run on and to
cation Server. Bonn: Galileo Press.
abstract from the programming language they
were implemented in. The second step is to map Karch, S., Heilig, L., et al. (2005). SAP NetWeaver.
these services onto different business processes Bonn: Galileo Press.
by focusing on your internal processes and your
Krafzig, D., Blanke, K., et al. (2004). Enterprise
immediate business partners, i.e. customers and
SOA - Service-Oriented Architecture Best Practise.
suppliers. The consequential third step is to extend
Upper Saddle River, NJ: Prentice Hall.
the reach of your services – and along with it the
reach of your business – by integrating them into 56L. (n/a.). SOA Ecosystem. from http://www2.
a business web. sims.berkeley.edu/academics/courses/is290-4/
This demonstrated the power of implementing s02/readings/line56ecosystem.gif.
(Enterprise) SOA. As a first benefit value is gen-
Martin, R., & Mauterer, H. (2002). Systema-
erated by reducing the effort for integrating your
tisierung des Nutzens von ERP-Systemen in der
internal IT-Systems. Then additional value is
Fertigungsindustrie. Wirtschaftsinformatik, 44(2),
generated by offering your direct partners access
109–116.
to your (Enterprise) SOA implementation. Fi-
179
N. N. (2007). Enterprise resource planning. Re- IT Practice: The application based view on
trieved 8.06.2007, 2007, from http://en.wikipedia. SAP NetWeaver identifies different main use cases
org/wiki/Enterprise_resource_planning. within an enterprise system landscape which are
called IT Practices. In different abstraction levels,
Nicolescu, V., Funk, B., et al. (2006). SAP Ex-
they describe the activities necessary to achieve
change Infrastructure for Developers. Bonn:
the implementation of a specific technical task
Galileo Press.
in a company.
Tapscott, D. (2007) Rethinking Enterprise SAP R/3: SAP R/3 has been released 1992 as
Boundaries: Business Webs in the IT Industry. the first ERP system for very large businesses based
NEWPARADIGM, DOI on a client-server architecture. It was divided in
different functional modules and was the central
Vogel, A., & Kimbell, I. (2005). mySAP ERP For
point of enterprise system landscapes.
Dummies. Indianapolis: Wiley Publishing, Inc.
SAP NetWeaver: Moving from SAP R/3
Woods, D., & Word, J. (2004). SAP NetWeaver to SAP ERP, the technical foundation of SAP
For Dummies. Wiley Publishing, Inc. R/3 SAP Basis was separated from the business
functions and enhanced by many other technical
features. This new technical basis that enables
SOA is called NetWeaver.
SOA: SOA is an architectural style whose
goal is to achieve loose coupling among interact-
Business Web: Corporations are transforming
ing software agents. A service is a unit of work
into flatter and more specialized enterprises, link-
done by a service provider to achieve desired end
ing with suppliers and other businesses to form
results for a service consumer. Both, provider and
a larger, more open, value-creation entity – the
consumer are roles played by software agents on
business web.
behalf of their owners.
180
181
Chapter 1.13
Data Reengineering of
Legacy Systems
Richard C. Millham
Catholic University of Ghana, Ghana
IntroductIon second step where the data are converted to the

new or converted system. Usually, the second step
Legacy systems, from a data-centric view, could involves substantial changes not only to the data
be defined as old, business-critical, and stand- structures but to the data values of the legacy data
alone systems that have been built around legacy themselves (Aebi & Largo, 1994).
databases, such as IMS or CODASYL, or legacy Borstlap (2006), among others, has identified
database management systems, such as ISAM potential problems in retargeting legacy ISAM
(Brodie & Stonebraker, 1995). Because of the data files to a relational database. Aebi (1997), in
huge scope of legacy systems in the business world addition to data transformation logic (converting
(it is estimated that there are 100 billion lines of sequential file data entities into their relational
COBOL code alone for legacy business systems; database equivalents), looks into, as well, data
Bianchi, 2000), data reengineering, along with its quality problems (such as duplicate data and in-
related step of program reengineering, of legacy correct data) that is often found with legacy data.
systems and their data constitute a significant part Due to the fact that the database and the pro-
of the software reengineering market. gram manipulating the data in the database are
Data reengineering of legacy systems focuses so closely coupled, any data reengineering must
on two parts. The first step involves recognizing address the modifications to the program’s data ac-
the data structures and semantics followed by the cess logic that the database reengineering involves
(Hainaut, Chandelon, Tonneau, & Joris, 1993).
DOI: 10.4018/978-1-60566-242-8.ch005
Data Reengineering of Legacy Systems
In this chapter, we will discuss some of the schema, of the source database through various
recent research into data reengineering, in par- reverse engineering techniques, and then the
ticular the transformation of data, usually legacy development of the target database, using this
data from a sequential file system, to a different conceptual schema, using standard database de-
type of database system, a relational database. This velopment techniques. This strategy produces a
article outlines the various methods used in data higher quality database with full documentation
reengineering to transform a legacy database (both as to the semantic meaning of the legacy data, but
its structure and data values), usually stored as this approach is more expensive in terms of time
sequential files, into a relational database structure. and effort that it entails (Hainaut et al., 1993a).
In addition, methods are outlined to transform the Hainaut et al.’s approach first uses the physical
program logic that accesses this database to access conversion strategy to convert data and then
it in a relational way using WSL (wide spectrum uses a trace of the program, which accesses the
language, a formal language notation for software) legacy data, in order to determine how the data
as the program’s intermediate representation. are used and managed. In this way, additional
structures and constraints are identified through
the procedural code. Through an analysis of the
relAted Work application’s variable dependency graph and of the
record and file definitions, data fields are refined,
In this section, we briefly describe the various ap- foreign keys are determined, and constraints on
proaches that various researchers have proposed multivalued fields are discovered. During the
and undertaken in the reengineering of legacy database conceptualization phase, the applica-
data. Tilley and Smith (1995) discuss the reverse tion’s physical constructs of indexes and files are
engineering of legacy systems from various ap- removed and the program’s objects of arrays, data
proaches: software, system, managerial, evolution, types, fields, and foreign keys are transformed into
and maintenance. their database equivalents (Hainaut et al., 1993a).
Because any data reengineering should address Initially, database reengineering focused on
the subsequent modifications to the program that recognizing the legacy database structure and
the program’s data access’ logic entails, Hainaut transforming these structures into a new model
et al. (1993) have proposed a method to transform (Aiken & Muntz, 1993; Joris, 1992; Pomerlani &
this data access logic, in the form of COBOL Blaha, 1993; Sabanis & Stevenson, 1992). The
read statements, into their corresponding SQL values of legacy data were used solely to identify
relational database equivalents. the legacy system’s dependencies in terms of keys
Hainaut et al. (1993) identify two forms of between records (Pomerlani & Blaha).
database conversion strategies. One strategy Aebi and Largo (1994), in the transformation of
(physical conversion) is the physical conversion database structures, recognize that the transforma-
of the database where each construct of the source tion of structural schemas involves many issues.
database is translated into the closest correspond- The first issue is that the different attributes and
ing construct of the target database without any entities of the old system must be mapped to the
consideration of the semantic meaning of the new schema of the transformed database. Con-
data being translated. One of the problems with straints, during the migration from the old to the
this strategy is that the resulting target database new system, may be added, dropped, or changed.
produced is of very low quality. The second Entity sets in the new system may be identified by
strategy (conceptual conversion) is the recovery new attributes or by old attributes with changed
of precise semantic information, the conceptual domains or data types.
182
Wu et al. (1997), with their “butterfly” ap- dynamic traces along with data flow and depen-
proach, assume that the legacy data are the most dency analyses are some successful techniques
important part of the legacy system and it is the used in data reverse engineering that have been
schema rather than the values of this legacy data obtained from the program understanding phase
that are the most crucial. This legacy data are of program reverse engineering (Cleve, Henrard,
modified in successive iterations with the legacy & Hainaut, 2006; Millham, 2005).
data being frozen and used for reading purposes Aebi and Largo (1994) also recognize problems
only. The “Chicken Little” strategy allows the with the transformation of database values. Some
legacy system to interact with the target system problems include duplicate data being used in the
during migration, using a gateway to serve as a same context, changes in the primary keys of tables
mediator. This gateway is used to translate and often entailing changes in foreign keys of the same
redirect calls from the legacy system to the target table, values of the attribute possibly exceeding its
database system, and then the gateway translates given range, different encoding schemes for data
the results of the target database for use by the values may be used, recording errors incorporated
legacy system and by the legacy database. Al- into the existing data, and no existing distinction
though the legacy system is allowed to interact between unknown and null values.
with the target database during migration, each Hainaut et al. (1993b) outline some practical
data access involves two database accesses: one schema-relational transformations, such as proj-
to the target database and another to the legacy ect-join, extension transformation, and identifier
database (Bisbal, Lawless, Wu, & Grimson, 1999). substitution, that are dependent on the file manager
Bianchi, Caivano, and Visaggio (2000) pro- such as CODASYL, relational, TOTAL/IMAGE,
posed a different method of data reengineering and IMS DBMS in order to function. For other
where the data structures are reengineered rather databases, such as COBOL file structure databases,
than simply migrated. This reengineering involves such transformations are impractical. In the case
the use of several steps. The first step is analyzing of COBOL file structure databases, multivalued
the legacy data through monitoring of calls to the attributes can be represented by list attributes
legacy database from the legacy system (dynamic only. Referential constraints upon files can be
trace). This dynamic trace is used to identify which detected through careful analysis of the COBOL
data could be classified as conceptual (data spe- procedural code, file contents, and secondary keys.
cific to the application domain and that describe Identifying one-to-many relationships in files is
specific application concepts), control (data that accomplished through depicting multivalued,
are used for program decisions or to record an compound attributes B of A as a many-to-one
event), structural (data that are used to organize relationship of B to A. Multirecord types within
and support the data structures of the system), or a sequential file may be portrayed as a many-to-
calculated (data that are calculated by the applica- one relationship. Multivalued attributes, such as
tion). The second step involves redesigning the those used in foreign keys, are represented in the
legacy database after identifying the dependen- relational target model as rows in a separate table
cies among the data. This dependency diagram is with a link to the main referencing table. If the
then converted to a target database schema. After source database is of a network type, a recursive
the legacy data are migrated to the new target relational type may be represented by an entity
schema, the legacy system is then altered such type and two one-to-many or one-to-one relational
that data accesses to the target database reflect types (Hainaut et al., 1993b).
the new database schema (Bianchi et al.). Using
183
mAIn Focus: dAtA another database format. However, in Millham

reenGIneerInG WItHIn (2005), the sequential records and their accesses
ProGrAm trAnsFormAtIon by the application have been specified in terms
of sets and set operations. Codd, in his relational
Because the semantic understanding of the data to database model, has specified this model in terms
be reengineered depends to a large degree upon of sets and relational calculus; WSL, in turn, may
the program accessing this data and because the be specified in terms of sequences, sets, and set
program accessing the target database needs to operations (Ward, 1992). Consequently, a method
be modified in order to account for reengineered to integrate the relational database model of Codd,
data, one of the areas of focus must be the analysis in its formal notation involving database speci-
of the underlying program and its usage of the fication and operation, and WSL, in its formal
data (Bianchi et al., 2000; Hainaut et al., 1993b). notation involving program transformations and
One of the problems with analyzing a program sequential file structures, could be accomplished.
for data access usage of legacy data is determin- Other formal notations, similar to WSL, with the
ing exactly when and where this legacy data are same representational capacity, could be utilized
accessed. Design pattern analysis of code has in the same manner. In this article, a method to
been proposed as a method of determining what transform a WSL-represented hierarchical data-
sections of code might be used for accessing this base, with its underlying program logic, into its
legacy data (Jarazabek & Hitz, 1998). relational database equivalent is provided. This
Another method is to convert the programming- method, a combination of the data reverse engi-
language-specific code into a programming- neering techniques of Hainaut et al. (1993) and
language-independent intermediate representa- Bianchi et al. (2000), occurs in conjunction with
tion. This intermediate representation can be in the program understanding phase of the WSL
a formal notation. Formal notations have been program reverse engineering process. During
used to specify the transformation of both pro- this program understanding phase, a static and
grams and their associated data mathematically. dynamic analysis of the program code is under-
One advantage of these transformations are that taken. This analysis produces a dependency graph
these transformations are generic (programming- of variables and identifies which variables serve
language independent) such that a transformation as control, structural, conceptual, and calculated
for a construct accessing a legacy database to a field variables. While WSL, in its representation,
construct accessing the target database will be does not distinguish between programs and file
the same regardless of the types of legacy and record types, the source code to WSL translation
target databases. For example, the formal notation process keeps track of which records are of type
WSL (Ward, 1992), with a proven set of program file record. Similarly, WSL is type-less but, dur-
transformations, has been used to transform a ing the source code to WSL representation, the
procedurally structured and driven COBOL legacy original data types of the source code are recorded
system using a WSL intermediate program rep- for later use during the data reengineering process
resentation into an object-oriented, event-driven (these processes are outlined in Figure 1).
system (Millham, 2005). In Figure 2, the data types of each record field,
Although WSL has been extended to represent along with their default values, are recorded in a
data structures, such as records (Kwiatkowski & database for future use during the data reengineer-
Puchalski, 1998), little attention has been paid to ing process. Hence, it is possible to derive the
transforming the data structures as represented program’s file and record types, along with the
in WSL as their original sequential file form into foreign and primary keys, from the information
184
Figure 1. Flow of reengineering processes and data store
obtained during these translation and analysis logical database structure is derived along with
processes. If one, A, has an array field that forms the necessary subsequent program data access
a dependency in another record B, this depen- transformations. These transformations are based
dency is depicted in a one-to-many relationship on a formal, platform-independent representation
between records A and B. Anti-aliasing of records (WSL) such that the source system, whether
and their fields reduces the number of records COBOL or assembly language, is not important
that refer to the same file but use different names. and the advantages of a formal notation, such as
Calculated data may appear as calculated fields preciseness, are achieved.
in the database schema. Structural data are used WSL equivalentWong and Sun (2006) have
to represent the target database schema. Record been working on detecting data dependencies in
fields that are used as control data may indicate programs using a hybrid UML (unified modeling
a relationship, involving the control data, between language) collaboration and activity diagram that
the record that defines this control data and any is expressed in a platform-independent XML
record(s) enclosed within the control block that (extensible markup language) markup graph.
is governed by that control data (Millham, 2005).
Constraints on data may be determined through
an analysis of their declaration within a program Future trends
or through the use of Hainault et al.’s (1993b)
methods. Because this data reengineering process Because up to 50% of a legacy system’s main-
occurs in conjunction with program reengineering, tenance costs, in terms of program and database
any changes to the underlying database structure changes, can be attributed to changing business
can easily be propagated to the program reengi- rules, there is a strong need to adopt new techniques
neering phase where the program data access in reengineering legacy data and their associated
logic will be altered to reflect these database applications. Jarazabek and Hitz (1998) propose
changes. Through this data reengineering, a more the use of domain analysis and the use of generic
Figure 2. A sample COBOL record and its WSL equivalent
185
architectural design techniques in reengineer- Because the underlying application(s) that ac-
ing as a method of substantially reducing these cess the legacy database and the legacy database
maintenance costs. Atkinson, Bailes, Chapman, are so intrinsically linked during the data reengi-
Chilvers, and Peake’s (1998) preferred approach neering phase, there is a need to be able to analyze
is to develop interfaces to general persistent data the applications, in a programming-language-
repositories in the direction of generic reengineer- independent way, for data usage and then transform
ing environment design. From these interfaces, these applications’ data accesses and the legacy
binding between the data stores, represented in database, using a set of generic transformations
the target databases, and the underlying applica- and tools, to the target database. Formal notations
tion, which will access this target database, can be (with their programming-language independence,
made. Another method is to use formal-language sets of transformations, and basis in relational
intermediate representations that can represent database theory and software reengineering) have
both the database and program along with a set been proposed as a means to generically analyze
of corresponding transformations to transform the application(s), once in the formal notation’s
the legacy to the target database and transform intermediate representation, for data usage and
the program accessing the legacy database to then transform this analysis into an accurate and
a program capable of accessing the new target meaningful target database schema for database
database without errors. Through the use of reengineering.
formal-language intermediate representations,
a set of generic reengineering tools for both the
legacy database and its underlying application(s) reFerences
could be designed.
Aebi, D. (1997). Data engineering: A case study.
In C. J. Risjbergen (Ed.), Proceedings in advances
conclusIon in databases and information systems. Berlin,
Germany: Springer Verlag.
Data reengineering has gone beyond a simple Aebi, D., & Largo, R. (1994). Methods and tools
physical conversion of a legacy database to its for data value re-engineering. In Lecture notes in
target database. Both Hainaut et al. (1993a) and computer science: Vol. 819. International Con-
Wu et al. (1997) utilize an analysis of the underly- ference on Applications of Databases (pp. 1-9).
ing application(s) accessing the legacy database Berlin, Germany: Springer-Verlag.
in order to determine the database’s underlying
data types and foreign keys. Bianchi et al. (2000) Aiken, P., & Muntz, A. (1993). A framework for
go further in using an analysis of the underlying reverse engineering DoD legacy information
application(s) accessing the legacy database in systems. WCRE.
order to identify the categories of the data be-
Atkinson, S., Bailes, P. A., Chapman, M., Chilvers,
ing used in the application such as whether the
M., & Peake, I. (1998). A re-engineering evalu-
data is used as conceptual, structural, control, or
ation of software refinery: Architecture, process
calculated data and in order to identify the de-
and technology.
pendencies among them. In this manner, a better
determination of the data usage can be made and
a better target database schema can be derived
from analysis of this usage.
186
Behm, A., Geppert, A., & Diettrich, K. R. (1997). Jarazabek, S., & Hitz, M. (1998). Business-
On the migration of relational schemas and data oriented component-based software development
to object-oriented database systems. Proceed- and evolution. DEXXA Workshop.
ings of Re-Technologies in Information Systems,
Jeusfeld, M. A., & Johnen, U. A. (1994). An
Klagenfurt, Austria.
executable meta model for reengineering of da-
Bianchi, A., Caivano, D., & Visaggio, G. (2000). tabase schemas. Proceedings of Conference on
Method and process for iterative reengineering of the Entity-Relationship Approach, Manchester,
data in a legacy system. WCRE. Washington, DC. England.
Bisbal, J., Lawless, D., Wu, B., & Grimson, Joris, M. (1992). Phenix: Methods and tools for
J. (1999). Legacy information systems: Issues database reverse engineering. Proceedings 5th
and directions. IEEE Software, 16(5), 103–111. International Conference on Software Engineer-
doi:10.1109/52.795108 ing and Applications.
Bohm, C., & Jacopini, G. (1966). Flow diagrams, Kwiatkowski, J., & Puchalski, I. (1998). Pre-
Turing machines, and languages with only two processing COBOL programs for reverse engi-
formation rules. CACM, 9(5), 266. neering in a software maintenance tool. COTSR.
Borstlap, G. (2006). Understanding the tech- Mehoudj, K., & Ou-Halima, M. (1995). Migrating
nical barriers of retargeting ISAM to RD- data-oriented applications to a relational database
BMS. Retrieved from http://www.anubex.com/ management system. Proceedings of the Third
anugenio!technicalbarriers1.asp International Workshop on Advances in Databases
and Object-Oriented Databases (pp. 102-108).
Brodie, M. L., & Stonebraker, M. (1995). Migrat-
ing legacy systems: Gateways, interfaces, and the Millham, R. (2005). Evolution of batch-oriented
incremental approach. Morgan Kaufmann. COBOL systems into object-oriented systems
through the unified modelling language. Un-
Cleve, A., Henrard, J., & Hainaut, J.-L. (2006).
published doctoral dissertation, De Montfort
Data reverse engineering using system depen-
University, Leicester, England.
dency graphs. WCRE.
Pomerlani, W. J., & Blaha, M. R. (1993). An
Hainaut, J.-L., Chandelon, M., Tonneau, C., &
approach for reverse engineering of relational
Joris, M. (1993a). Contribution to a theory of data-
databases. WCRE.
base reverse engineering. WCRE. Baltimore, MD.
Rob, P., & Coronel, C. (2002). Database systems:
Hainaut, J.-L., Chandelon, M., Tonneau, C., &
Design, implementation, and management. Bos-
Joris, M. (1993b). Transformation-based data-
ton: Thomas Learning.
base reverse engineering. Proceedings of the 12th
International Conference on Entity-Relationship Sabanis, N., & Stevenson, N. (1992). Tools and
Approach (pp. 1-12). techniques for data remodeling COBOL applica-
tions. Proceedings 5th International Conference
Janke, J.-H., & Wadsack, J. P. (1999). Varlet:
on Software Engineering and Applications.
Human-centered tool for database reengineer-
ing. WCRE. Tilley, S. R., & Smith, D. B. (1995). Perspectives
on legacy system reengineering (Tech. Rep.).
Carnegie Mellon University, Software Engineer-
ing Institute.
187
Ward, M. (1992). The syntax and semantics of the databases during the data reengineering phase
wide spectrum language (Tech. Rep.). England: through the use of a gateway that translates data
Durham University. access requests from the legacy system for use
by the target database system and then translates
Weiderhold, G. (1995). Modelling and system
the result(s) from the target database for use by
maintenance. Proceedings of the International
the legacy system.
Conference on Object-Orientation and Entity-
Conceptual Conversion Strategy: A strategy
Relationship Modelling.
that focuses first on the recovery of the precise
Wong, K., & Sun, D. (2006). On evaluating the semantic meaning of data in the source database
layout of UML diagrams for program comprehen- and then the development of the target database
sion. Software Quality Journal, 14(3), 233–259. using the conceptual schema derived from the
doi:10.1007/s11219-006-9218-2 recovered semantic meaning of data through
standard database development techniques.
Wu, B., Lawless, D., Bisbal, J., Richardson, R.,
Domain Analysis: A technique that identifies
Grimson, J., Wade, V., et al. (1997). The butterfly
commonalties and differences across programs and
methodology: A gateway-free approach for mi-
data. Domain analysis is used to identify design
grating legacy information system. In ICECOS
patterns in software and data.
(pp. 200-205). Los Alamos, CA: IEEE Computer
Legacy Data: Historical data that are used by a
Society Press.
legacy system that could be defined as a long-term
Zhou, Y., & Kontogiannis, K. (2003). Incremental mission-critical system that performs important
transformation of procedural systems to object- business functions and contains comprehensive
oriented platform. Proceedings of COMPSAC, business knowledge
Dallas, TX. Multivalued Attribute: When an attribute, or
field, of a table or file may have multiple values.
For example, in a COBOL sequential file, its
corresponding record may have a field, A, with
several allowable values (Y, N, D). Translating
this multivalued attribute to its relational data-
Butterfly Approach: An iterative data reengi-
base equivalent model is difficult; hence, lists or
neering approach where the legacy data are frozen
linked tables containing the possible values of
for read-only access until the data transformation
this attribute are used in order to represent it in
process to the target database is complete. This
the relational model.
approach assumes that the legacy data are the
Physical Conversion Strategy: A strategy
most important part of the reengineering process
that does not consider the semantic meaning of
and focuses on the legacy data structure, rather
the data but simply converts the existing legacy
than its values, during its migration.
constructs of the source database to the closest
Chicken Little Approach: An approach that
corresponding construct of the target database.
allows the coexistence of the legacy and target
This work was previously published in Handbook of Research on Innovations in Database Technologies and Applications:
Current and Future Trends, edited by Viviana E. Ferraggine, Jorge Horacio Doorn and Laura C. Rivero, pp. 37-44 , copyright
188
189
Chapter 1.14
Semantically Modeled
Databases in Integrated
Enterprise Information Systems
Cheryl L. Dunn
Grand Valley State University, USA
Gregory J. Gerard
Florida State University, USA
Severin V. Grabski
Michigan State University, USA
IntroductIon Geerts and McCarthy, 2002), and they allow the

database to evolve as the enterprise does through
Semantically modeled databases require their time (Abrial, 1974).
component objects to correspond closely to real Organizations have implemented enterprise
world phenomena and preclude the use of artifacts resource planning (ERP) systems in an attempt
as system primitives (Dunn and McCarthy, 1997). to improve information integration. Much of the
Enterprise information systems (also known as value of these ERP systems is in the integrated
enterprise resource planning systems) based on database and associated data warehouse that is
semantically modeled databases allow for full implemented. Unfortunately, a significant por-
integration of all system components and facilitate tion of the value is lost if the database is not a
the flexible use of information by decision-makers. semantic representation of the organization. This
Researchers have advocated semantically de- value is lost because the semantic expressiveness
signed information systems because they provide is insufficient -- relevant information needed to
benefits to individual decision-makers (Dunn and reflect the underlying reality of the organization’s
Grabski, 1998, 2000), they facilitate organizational activities is either not stored in the system at all,
productivity and inter-organizational communi- or it is stored in such a way that the underlying
cation (Cherrington et al., 1996; David, 1995; reality is hidden or disguised and therefore cannot
be interpreted.
DOI: 10.4018/978-1-60566-242-8.ch026
Semantically Modeled Databases in Integrated Enterprise Information Systems
Partly as a result of systems lacking expres- underlying semantic model, many of the current
sive semantics, researchers have been develop- problems can be eliminated.
ing ontologies. Gruber (2008) provides a useful This chapter first presents a normative semantic
definition of ontology: model for enterprise information systems that has
its roots in transaction processing information
“In the context of database systems, ontology systems. We use this model because the majority
can be viewed as a level of abstraction of data of information processed and tracked by informa-
models, analogous to hierarchical and relational tion systems is transactional in nature. We review
models, but intended for modeling knowledge empirical research on semantically modeled in-
about individuals, their attributes, and their formation systems and then provide an example
relationships to other individuals. Ontologies company’s semantic model as a proof of concept.
are typically specified in languages that allow We next discuss how this model can be applied to
abstraction away from data structures and imple- ERP systems and to inter-organizational systems
mentation strategies; in practice, the languages and present future trends and research directions,
of ontologies are closer in expressive power to and provide concluding comments.
first-order logic than languages used to model
databases. For this reason, ontologies are said semantic model development
to be at the “semantic” level, whereas database
schema are models of data at the “logical” or In this chapter, we adopt a definition of an enter-
“physical” level. Due to their independence from prise information system that is based on David et
lower level data models, ontologies are used for al.’s (1999) definition of an accounting informa-
integrating heterogeneous databases, enabling tion system: an enterprise information system that
interoperability among disparate systems, and captures, stores, manipulates, and presents data
specifying interfaces to independent, knowledge- about an organization’s value-adding activities
based services.” to aid decision-makers in planning, monitoring,
and controlling the organization. This definition is
We base our discussion in this paper on the also consistent with much of the research on ERP
Resources-Events-Agents (REA) ontology (Mc- systems. We recommend that the REA ontology
Carthy, 1982; Geerts and McCarthy 1999; 2000; (REA semantic model) (McCarthy, 1982) be used
2004; 2001; 2002; Haugen and McCarthy, 2000) as the core foundation of enterprise information
which is considered an enterprise ontology or a systems due to the model’s robust and general
business domain ontology. Ontologically-based nature. The semantics of the REA model are de-
information systems with common semantics are signed to capture the essential features of value
regarded as a necessity to facilitate inter-organiza- added activities – activities that correspond to
tional information systems (Geerts and McCarthy, exchanges of resources (e.g., giving inventory and
2002). Presently, most inter-organizational data is receiving cash) and transformations of resources
sent via EDI (which requires very strict specifica- (converting raw materials into finished goods). The
tions as to how the data are sequenced and requires basic REA model is presented in Figure 1 using
some investment by adopting organizations). entity-relationship notation (Batini et al., 1992),
The same requirement holds true for web-based however, it has also been implemented in NIAM
systems. There is no or very limited knowledge (Geerts and McCarthy, 1991) and in object notation
inherent in those systems. Alternatively, if trading (Nakamura and Johnson, 1998). The general REA
partners implement systems based on the same model for any particular transaction cycle consists
190
Figure 1. The original REA model (McCarthy, 1982)
of the following components (these components • Stock-flow relationships between the

are presented in list form for the sake of brevity). events and the associated resources, repre-
Readers are encouraged to read McCarthy (1982) senting the inflows or outflows of the re-
and Dunn et al. (2005) for more detail. sources resulting from the events.
• Responsibility relationships between the
• Two economic events that represent alter- events and the internal agents.
native sides of an economic exchange (one • Participation relationships between the
increment event and one decrement event). events and the external agents.
• Two resources that represent what is re-
ceived and given up in the economic The general model outlined above can be modi-
exchange. fied to accommodate specific needs. The original
• Two internal agents that represent the com- model was described at only the operational level,
pany’s personnel responsible for the eco- which is sufficient for historical accounting pur-
nomic events (usually one agent type for poses. The REA model has now been extended to
each event). an inter-organizational database systems ontology
• One external agent that represents the that incorporates both the operational (what has
person or company with whom the com- happened) and knowledge levels (what should
pany is engaging “at arms’ length” in the happen as a policy or expectation) and includes
exchange. the following additional features and compo-
• Duality relationship between the increment nents, again presented in list form (Geerts and
and decrement economic events. McCarthy, 2002).
191
• Integration of process (cycle) level models • Scenario history relationships between ab-
into a value-chain level model at a higher stract occurrents and other abstractions,
level of abstraction. e.g. event type-resource type.
• Expansion of process (cycle) level models • Business Process as a description of the
into workflow or task level models at lower interaction between resources, agents, and
levels of abstraction. dual events.
• Separation of components into continuants • Partnering as the purpose of the inter-
(enduring objects with stable attributes that action between resources, agents, and
allow them to be recognized on different commitments.
occasions throughout a period of time) and • Segmentation as a description of the group-
occurrents (processes or events that are in ing of physical categories into abstract
a state of flux). continuant categories.
• Type images that represent category level • Policy or Standard as a description of
abstractions of similar components. the expression of knowledge level rules
• Commitment images that represent agree- between abstract types (e.g. scripts and
ments to engage in future economic events. scenarios).
• Assignment relationships between agents • Plan as a description of the application of a
that represent the designation of an agent script to physical occurrents.
category to work with another agent • Strategy as a description of rules for
category (e.g. salesperson assigned to the execution of a Business Process or
customer). Partnering.
• Custody relationships between agents and
resources that represent the agents that are The semantics contained within the expanded
accountable for various resources. REA model facilitate the information exchange
• CommitsTo relationships between com- between trading partners and likely provide a
mitment images and the resulting econom- needed core foundation for ERP systems. For
ic events. example, research reported by Andros et al. (1992)
• Partner relationships between commitment and Cherrington et al. (1996) found that IBM was
images and the participating agents. able to obtain significant benefits from a semanti-
• Reserved relationships between commit- cally modeled system based on the REA model.
ment images and the resources that are the Reported benefits of the new system included
proposed subject of the future exchange. a significant reduction in the time to process
• Typification description relationships be- employee reimbursements, significant cost re-
tween continuant components and the cate- ductions, and a generally high level of employee
gories to which they belong, e.g. resource- satisfaction. These studies demonstrate how the
resource type relationships. semantic models were used as the basis for systems
• Characterization description relationships design, and then how the resultant systems were
between continuant type images, e.g. agent perceived by the end-users, thereby completing
type-agent type relationships and agent the research loop from design to end-user.
type-resource type relationships. Weber (1986) empirically evaluated the REA
• Typification history relationships between semantic model. His objective was to determine
physical occurrents and their types, indi- whether software practitioners had both identified
cating that the occurrents share the same and solved the same problems as identified by aca-
script, e.g. event-event type. demicians. He found that the REA model fulfilled
192
its objective as a generalized model and that it entation, based in part on accounting artifacts. Part
was a good predictor of the high-level semantics of the compromises are likely due to the fact that
found in all twelve packages. He believed that the software vendors’ products are constantly evolv-
REA model excluded certain types of events such ing in an incremental fashion, and vendors never
as contracts and suggested extensions (many of do a software redesign starting with a clean slate.
which have been subsequently incorporated into Research has demonstrated that organizations
the model). Weber reported that important differ- are able to obtain significant benefits from a
ences in the low-level semantics existed across all semantically modeled system based on the REA
packages, and that these seem to reflect the relative framework (Andros et al., 1992; Cherrington et
complexity of the different packages. Nonetheless, al., 1996). However, even semantically modeled
the clustering of the low-level semantics of the systems are not sufficient to ensure success.
packages according to the REA model allowed for Rather, success is dependent upon a variety of
the identification of similarities and differences factors including top management support and
among the packages, and the likely strengths and corporate philosophy in addition to the benefits
limitations of the package become apparent. On inherent in the semantically modeled system
a datalogical level, Weber found few violations (Satoshi, 1999). Research has also demonstrated
of normal form given the number of fields in the that the REA model is sufficient as a generalized
packages. He concluded that theory was a good model (Weber, 1986; O’Leary, 2004), and the more
predictor of design practice and that the empirical REA-like a system is, the more it is associated
evidence supports normalization theory. with productivity and administrative efficiencies
David (1995) developed a metric to classify (David, 1995).
organizations’ accounting systems characteristics We next present an REA model of a prototypi-
(ASC) along a continuum between traditional cal retail firm. This example will serve as further
general ledger based accounting systems and proof of concept and allow the reader to verify
REA systems. The ASC metric was developed that the semantic data model accurately represents
based upon characteristics that were identified the reality of the described organization, and that
in theoretical research as critical characteristics there is an application of the REA model with
for REA systems. David visited companies in the consistent patterns. First, we will provide a nar-
pulp and paper industry and conducted structured rative for the firm and describe in detail the sales
interviews and used the ASC metric to determine cycle. Then we will describe the components of a
each system’s position on the continuum. Data was REA model (in entity-relationship notation using
also gathered as to the companies’ productivity, guidelines set forth by Batini et al., 1992) that
efficiency, and the company executives’ percep- captures the reality. Finally, we will discuss the
tions of competitive advantage. REA-like systems benefits of the REA model that occur independent
were found to be associated with productivity and of the entity-relationship notation. Our example
administrative efficiencies. is for a company called Robert Scott Woodwinds
O’Leary (2004) compared the REA seman- (RSW), a retail organization that sells, leases
tic model with an enterprise resource software and repairs woodwind musical instruments, and
package. He compared information about SAP sells music supplies. This organizational setting
from various sources and determined that SAP encompasses many of the activities performed by
is consistent with the REA model in its database, most business organizations. While the semantics
semantic and structure orientations. However, of the model are specific to RSW’s environment,
SAP was also found to contain implementation the model provides a generic template for a retail
compromises in the structuring and semantic ori- organization’s semantic model, it also provides
193
a template for leasing organizations, and it has school (usually this means that the children will
significant commonalities with manufacturing rent from RSW, not that the school will purchase
firms in the repair aspects of the business. the instruments from RSW). Following is informa-
tion about RSW’s revenue (including sales, sales
REA Semantic Model Example returns, repairs, and leases) cycle.
For the scenario presented below, the resources, Revenue Cycle

events, and agents are captured in REA diagrams
(in entity-relationship form) in Figures 2 through 5. RSW generates revenue in three ways. The first is
RSW sells, repairs, and leases musical woodwind through sales of instruments; the second is through
instruments from a single store. The business ca- rental of instruments; and the third is through
ters to professional musicians, casual musicians, the repair of instruments. In the retail instrument
and children who are involved in school music business, renting instruments so that customers
programs. While RSW clearly needs to have a can “try before they buy” is a great way of gener-
web presence, it anticipates only a minor porating sales. This is particularly true in the school
tion of its revenues from web sales (primarily marketplace. Thus, RSW sends salespeople (who
for consumables such as reeds and sheet music). also have music teacher training) into the schools
Professional musicians want to play and get a to host “music talent exploration” sessions for
feel for the particular instrument that they are the children. When a child has been identified as
interested in purchasing and will always come having talent for (and some interest in) a specific
into the store. Further, these musicians are very instrument, the parents will be offered the oppor-
particular about who will repair their instrument. tunity to rent that instrument for 3 months. At the
For repairs, the instrument must be brought or sent end of the 3-month trial period, the parents may
to the store. School sales are a result of sales staff purchase that instrument (or a different instru-
calling on the band directors at the various schools ment) and the sale price they pay will reflect a
and gaining permission to be the supplier for that discount equal to the amount of rent paid. When
Figure 2. RSW sales revenue expanded REA data model
194
Figure 3. RSW rental revenue cycle expanded REA data model
Figure 4. RSW sales returns expanded REA data model
an instrument is rented out, it is transferred from delivered to the customer at the school during the
“Inventory for Sale” to “Inventory for Rent”. If salesperson’s next sales call, or the customer may
it is later sold, the “Inventory for Rent” category pick up the instruments at the store. Cash receipts
will be decreased. are processed by cashiers and are deposited into
RSW’s salespeople complete sales invoices the company’s main bank account each day. The
and rental invoices for customers. The customer semantic data models for sales and for leases are
may purchase (rent) multiple instruments on a presented in Figures 2 and 3 respectively.
single sales (rental) invoice. Instruments are either
195
Figure 5. RSW repair service revenue expanded REA data model
As can be seen in the sales revenue data model relationship established between salesperson and
(Figure 2), the original REA template has been customer. Sales personnel can be assigned to
applied. The resource (inventory) is associated many different customers, and a customer has, at
with an economic decrement event (sale) via a most, one sales person assigned to them. The REA
stockflow relationship. The sale event is associ- ontology allows for the inclusion of additional
ated with the internal agent (salesperson) and the entities that managers need information about for
external agent (customer) via control relation- planning, controlling and evaluating individual
ships. There is a duality relationship between and organizational performance. These additions
the economic decrement event (sales) and the do not change the underlying REA template; they
economic increment event (cash receipt). Cash simply add to it. For example, the sales call event
receipt is associated with the resource (cash) via that leads to the sales order event is modeled, and
a stockflow relationship and to the internal agent is linked to the associated resource and agents.
(cashier) and the external agent (customer) via Further, the external agent set has been expanded
control relationships. Note that the cardinalities to include school and district. A customer does not
between sale and cash receipt disclose that a sale need to be associated with a school, and a school
does not require the immediate receipt of cash. may be listed even if no customers from that
Rather, a sale may be paid for at one later point school have begun to trade with RSW. A school
in time, in installments over a period of time, or does not need to belong to a school district (i.e.,
not paid for at all (because of a sale return or bad it is a private school), however, if it belongs to a
debt). Newer aspects of the REA ontology have school district, it can only belong to one school
also been included in this model. The sales order district. A school district has at least one school.
event is a commitment image and is related to the The information presented in the semantic
resulting economic event (sales) via a CommitsTo model is a representation of the reality of the world
relationship. It is also related to salesperson via a of RSW. The cardinalities contain the “business
Control relationship, to customer via a Partner- rules” that are followed by the entities in their
ing relationship, and to inventory via a Reserved relationships. A similar analysis can be performed
relationship. In addition, there is an Assignment for the lease revenue data model (Figure 3), and
196
in fact, the only differences between the two are etc.). This serves as the bill of materials for the
the events of rental order and rental contract in service and allows storage of information about
place of sales order and sale. The observation of the “categories” to which the repair services be-
such commonalities allows for the development of long. Another internal agent is also included, that
common business practices for all revenue activi- of repair person. This allows RSW to track the
ties and facilitates reengineering at the task level. person was responsible for a given repair. This is
In a similar manner to the sales and leases, a similar to any type of job shop or organization that
semantic model of sales returns and lease returns must track employee time (e.g., consulting firms).
can also be created. The sale return data model For simplicity and in the interest of brevity, we
(Figure 4) has the economic decrement sale event do not present the acquisition, payroll or financ-
associated with the economic increment event ing cycles. The acquisition model is similar to the
sale return, and the sale return is associated with revenue cycle, with cash disbursement being the
the subsequent economic decrement event cash decrement event, purchase the increment event
disbursement (the sale return might not result in a and vendor is the external agent. The payroll
cash disbursement if no cash was received at the cycle would be similar to the acquisition cycle
time of the sale; this is represented by the 0 min except that services are acquired from employees
cardinality on the sale return). The rental returns (hence employees act as external agents) instead
data model is analogous (the difference is in the of from vendors or suppliers. The financing
economic increment event of rental return which model is also similar to the acquisition process
is associated with the rental contract, and this except that the resource acquired and consumed
model is not presented in the interest of brevity) is cash. Sometimes the flow of resources may not
The third form of revenue generated by RSW be measurable, for example, in the payroll cycle
is through instrument repair services. RSW offers the resource acquired is employee labor. This is
a variety of repair service types. “Complete over- an intangible resource that may be difficult, if
hauls” are available for many instruments. Prices not impossible to measure in terms of stockflow.
of these complete overhauls vary by instrument The REA ontology requires that such things be
type. “Repad overhauls” are also available for considered and only left out of the model if they
each instrument type. These are less expensive are impossible to implement.
than complete overhauls and price varies by instru- The REA ontology can be implemented using
ment type. Customers may have individual keys alternative notations, including object notation,
replated at a fixed cost per key. Crack repairs will UML, and others. We have chosen to represent
be judged as minor or major and will be priced it with entity-relationship notation because of its
accordingly. Other repairs are also available. The simplicity and widespread use. The benefit of the
semantic model for repair revenue is presented REA ontology is not found in the notation itself,
in Figure 5. but in the repeated application of the pattern to all
In the repair service REA model, more of the of the various business cycles. Consistent use of
features of the newer REA ontology are introduced. the template gives users an ability to understand
Typification is used, with an event to event-type the reality being represented, and it also provides
relationship between repair service and service system reusability and extendibility. Dunn and Mc-
type and also a commitment-commitment type Carthy (2000) describe four benefits of the REA
relationship between repair service order and model. First, the standardized use and definition
service type. The service type entity represents of information structures across organizational
the details of the various repair service types boundaries facilitates electronic commerce, by
RSW can provide (e.g. complete overhauls, repad, enabling the retrieval, sending, and integration of
197
data among business partners or research sources. critical (Bodart et al., 2001), perhaps because of
Second, intellectual complexity is made manage- the semantics inherent in the modeled system.
able by the use of natural primitives that abstract System designers distinguish between entities
to generalized descriptions of structures which in and relationships, with entities being primary
turn cover many thousands of cases with as few (Weber, 1996). Data and process methodologies
exceptions as possible. Third, consistent use of the are easier for novices than the object methodology,
REA pattern can enable system-based reasoning and resulted in less unresolved difficulties during
and learning. Fourth, use of system primitives that problem-solving processes (Vessey and Conger,
closely mirror the underlying phenomena enables 1994), and there is a more pronounced effect
system adaptability. for process-oriented tasks (Agarwal, Sinha, and
It is important to note that the REA ontology Tanniru, 1996a). Experience in process modeling
and other semantic modeling approaches have matters, regardless of whether the modeling tool
been studied at the individual decision-making (process versus object-oriented) is consistent
level, both for system designers and users. Dunn with the experience (Agarwal, Sinha, and Tan-
and Grabski (2002) provided an extensive litera- niru, 1996b).
ture review of individual-level semantic models
research. Overall conclusions of their review semantic models, erP systems,
were as follows. Accounting systems based on the and Inter-organizational systems
REA model are perceived as more semantically
expressive by end users than are accounting sys- The American Production and Inventory Control
tems based on the traditional debit-credit-account Society (APICS, 1998) defined an ERP system
model. Also, accounting systems perceived as as “an accounting-oriented information system
semantically expressive result in greater accuracy for identifying and planning the enterprise-wide
and satisfaction by end-users than do non-seman- resources needed to take, make, ship, and account
tically expressive accounting systems (Dunn and for orders.” David et al. (1999) proposed using
Grabski, 2000). Conceptual modeling formalisms REA as a basis for comparison among systems
are superior to logical modeling formalisms for and ERP packages. This was based, in part, on
design accuracy (Sinha and Vessey, 1999; Kim and Weber (1986) who reported that a comparison at
March, 1995). The ability to disembed the essen- the level of symbol sets made semantic similarities
tial objects and relationships between the objects and differences apparent. Additionally, O’Leary
in complex surroundings depends on a cognitive (2004) compared SAP to the REA model and
personality trait, field independence and leads to determined that SAP is REA-compliant; however,
more accurate conceptual model design (at least SAP has significant implementation compromises
for undergraduate students) (Dunn and Grabski, based on accounting artifacts. Consistent with
1998). The focus on increment and decrement David et al. and based upon the APICS definition
resources and events along with the associated of ERP systems, we believe that REA is a robust
agents to those events is consistent with data- candidate to which ERP systems may be compared
base designers’ thought processes. Additionally, because of its strong semantic, microeconomic
knowledge structures consistent with the REA and accounting heritage. More importantly, we
template’s structuring orientation are associated believe that semantic models must be used as a
with more accurate conceptual accounting data- basis for the information system because of the
base design (controlling for knowledge content, information contained within the semantics.
ability, and experience level) (Gerard, 2005). The Watson and Schneider (1999) also emphasized
lack of mandatory properties with entities is not the importance of ERP systems in providing the
198
process-centered modeling perspective necessary and what resources flow between them. In RSW,
for successful organizations. They emphasized financing is obtained and as a result, cash is dis-
understanding the underlying process model tributed, as needed to the various acquisition
inherent in ERP systems. That is, the underlying processes. Some of that cash is used to acquire
semantics of the system. Further, the SAP R/3 labor, which is then in turn distributed to each of
business blueprint provides four enterprise views; the other business processes. Some of the cash is
process, information, function, and organization. used to acquire instruments and supplies (both
It is in the organization view that the semantic supplies for sale and also supplies used in the
relationships among organizational units are repre- repair services). Some of the cash is used to acquire
sented. As most ERP implementations will require G&A services. The instruments, supplies, G&A
some modifications to meet the specific business services, and labor are received by the three rev-
needs, the analysts, designers, and users need enue processes and are combined in various ways
to understand the changes and their subsequent to generate revenue. The resulting cash is then
planned and unplanned impact. A semantically distributed to the financing process where it is
based model will facilitate this understanding. used to repay financing (including distribution of
The REA ontology provides a high-level defini- earnings to stockholders if this were a corporation)
tion and categorization of business concepts and or to re-distribute cash to the other business pro-
rules, enterprise logic, and accounting conventions cesses. The value chain level demonstrates inte-
of independent and related organizations (Geerts gration opportunities for the enterprise-wide in-
and McCarthy, 2002). The REA ontology includes formation system. Resources that flow from one
three levels: the value chain level, the process level, business process to another can be modeled in the
and the task level. The value chain level models database one time and accessed by the appropri-
an enterprise’s “script” for doing business. That ate business processes. Then as the separate busi-
is, it identifies the high-level business processes ness processes have their events linked to those
or cycles (e.g. revenue, acquisition, conversion, resources, the entire system is integrated and can
financing, etc.) in the enterprise’s value chain be examined at a level of abstraction higher than
and the resource flows between those processes. the individual business process level.
The process level model represents the semantic For inter-organizational system modeling, a
components of each business process. The example fourth level could be added to the REA ontol-
in section 2 depicted the process level of the REA ogy – the value system level. This enables an
ontology for RSW. The task (or workflow) level integration that spans the conceptual models of
of the REA ontology is the most detailed level, each organization to recognize that many busi-
and includes a breakdown of all steps necessary ness activities and phenomena such as electronic
for the enterprise to accomplish the business commerce and supply chain management that
events that were included at the process level. The involve multiple organizations. A company’s
task level can vary from company to company value system consists of its relationships with
without affecting the integration of processes or all other entities with which the company makes
inter-organizational systems; therefore we do not exchanges. For example, a company engages in
present an elaboration of this level (it is also the exchanges with its employees, its suppliers, its
level at which most reengineering occurs). An customers, and its creditors. An enterprise system
illustration of the value chain level for RSW is model at this level would represent the resource
presented in Figure 6. flows between the various types of organizations
The value chain level describes how the busi- in the enterprise’s value system. For example,
ness processes within the company fit together for RSW, the value system model is depicted in
199
Figure 6. RSW value chain level model
Figure 7. Partnering organizations that build their agents. Automated intensional reasoning systems
enterprise databases on the REA ontology will make inferences based on database table inten-
be better able to integrate their databases. This sions, and require completely consistent use of
will probably be best-accomplished using object any underlying pattern such as REA.
technology and artificial intelligence concepts Although REA is being used by some consul-
such as automated intensional reasoning (Geerts tants and software developers, as indicated in the
and McCarthy, 1999) and automated intelligent literature review in section 2 (also see e.g., REA
Figure 7. RSW value system level model
200
Technology at http://reatechnology.com/ and ages conformance to a core pattern at the business

Workday at http://www.workday.com), there are process level (or at least to a core set of pattern
various other approaches to integrating systems components), but allows considerable variation at
within and across organizations. Approaches to the task level. We suggest that the integration of
integrating systems within organizations have organizations’ information systems does not need
primarily been ERP software-driven and have to occur at the task level; it can occur at the value
also focused on document exchange. Wakayama system and process levels where the core pattern
et al. (1998) state that “rethinking the role of components can easily be standardized (however,
documents is central to (re)engineering enter- as noted above, the systems need to include task
prises in the context of information and process level specifications of homonyms and synonyms to
integration” because documents “are a common facilitate integration). As Swagerman, Dogger, and
thread linking integration issues.” Electronic data Maatman (2000) note, standardization of patterns
interchange is certainly document-oriented, with of behavior ensure the semantic context is clear
standards developed for issues such as which field to every user. We believe the standardization of
on an electronic purchase order transmitted be- patterns and pattern components along with the
tween companies represents the quantity ordered. development of appropriate artificial intelligence
More recent advances for web-based transmission tools will allow system integration without impos-
of documents, such as various markup languages, ing formal structures on the social domain.
have also focused on identifying fields that exist Limited research has been conducted as to the
on these documents. similarities of the semantic models underlying the
We believe it is not the documents themselves current ERP packages. Nonetheless, these models
that provide the common thread among company’s do exist and many organizations reengineer them-
business processes. Rather it is the similar nature selves to become consistent with the best practices
of the underlying transactions and information embodied within these models. Unfortunately,
needs associated with managing these transac- this is often at the task level and the benefits of
tions that are depicted within the organization’s the underlying semantics are lost. This is very
semantic model. Although organizations can have apparent when organizations seek to extend their
very different “scripts” for doing business, the value chains up and down their supply chain. If
core of the scripts is often the same. Therefore, the underlying semantics were preserved, along
rather than taking a document-oriented approach with the standardization of semantic patterns,
to inter-organizational system integration, we then automated intensional reasoning and other
believe it is more important to apply an enterprise knowledge-based tools would be able to facilitate
ontology approach such as REA. the inter-organizational trading. Semantically
Vendors of enterprise resource planning sys- modeled enterprise information systems will pro-
tems recognize the need for inter-organization vide many benefits, from the individual decision
integration and supply chain management. This maker level to the inter-organizational level. The
integration issue has not been easily resolved. critical issue is to ensure that the semantics are
“Bolt-on” applications are typically used. It is not lost upon the implementation of the system
possible that some of the difficulties in integrating and obscured by the task level mechanics. When
systems across organizations with ERP systems are this occurs, all subsequent benefits are lost and
due to the requirement of many ERP packages that we are faced with the task of integrating disparate
organizations conform to the software (in the form systems that are conceptually identical.
of “best practices”) at the task level. REA encour-
201
Future trends In semAntIcAlly they face. Again, object-based systems seem to

modeled systems provide the most benefit, but additional research
is needed.
Electronic commerce is commonplace; however Regardless of what the research demonstrates,
the majority of transactions in business to business organizations will need to be convinced that they
electronic commerce rely on the transmission of should change the ERP systems that they have
data based on rigidly structured electronic data acquired. These systems required an immense
interchange formats. We believe that the use of investment, and in some cases, these systems are
an ontological modeling approach such as the still not functioning in an acceptable manner. It is
REA ontology discussed in the previous sections most likely that change will need to be driven by
has the potential to enhance business to business the ERP vendors themselves. They would have
electronic commerce. a vested interest in selling upgrades to their sys-
The issues related to supply chain management tems as long as they can demonstrate some type
are similar to e-commerce. In fact, when examining of advantage for the consumer. This has occurred
inter-organizational systems from a supply chain with a great deal of regularity in the PC market.
perspective, the same set of issues apply as found in A significant number of organizations would
business-to-business e-commerce. Consequently, need to make the change in order for the benefits
our recommendations and expectations are the discussed in this chapter to occur. The first fax
same, and these have been presented in the prior machine sold did not provide as much value as
sections. The use of an ontological system like the one millionth fax machine sold; the more fax
the REA ontology is a necessary but not sufficient machines are sold, the greater are the opportunities
condition to facilitate effective and efficient inter- for sending and receiving information. Similarly,
organizational systems. Again, intelligent agents the value of the first REA based system for inter-
and automated intensional reasoning is also re- organizational use will be limited (although it will
quired for this to occur. Further, the semantics of facilitate intra-organization needs), but the more
the systems must not be obscured by subsequent companies that realize the value of these systems
implementation artifacts. and build them, the more value will accrue to the
There are many issues that still need to be first such system.
resolved, and as such these present many research
opportunities. One issue focuses on the scalability
of the systems based on the REA ontology. The conclusIon
system presented in this chapter is for a relatively
small organization. How this translates into a In this chapter we presented a normative seman-
system for a large multinational firm needs to be tic model for designing integrated databases for
explored. Also, while research has been conducted enterprise information systems. This model was
on automated intensional reasoning (Rockwell and developed by McCarthy (1982) and its expanded
McCarthy, 1999), much more is needed. Further, form has been proposed as an enterprise ontology
this needs to be extended to the use of intelligent by Geerts and McCarthy (2002). This ontology is
agents and to object-based environments. intended to serve as a foundation for integrated
Another issue is that of preserving the seman- enterprise-wide and inter-organizational systems.
tics at an operational level, beyond the level of To take full advantage of the semantically-rich on-
the database itself. This would allow decision- tological patterns and templates, the REA ontology
makers additional insight into the problems and must implemented with current advances in arti-
the information available to address the issues that ficial intelligence technology and object-oriented
202
database technology. Many of the current problems Batini, C., Ceri, S., & Navathe, S. B. (1992).
faced by companies who attempt to install ERP Conceptual Database Design: An Entity Approach.
systems and integration tools such as EDI can be Redwood City, CA: Benjamin Cummings.
minimized by use of common semantic patterns
Bodart, F., Patel, A., Sim, M., & Weber, R. (2001).
that can be reasoned about by intelligent systems.
Should optional properties be used in conceptual
Companies may integrate their systems without
modeling? A theory and three empirical tests.
using identical business practices.
Information Systems Research, 12(4), 384–405.
Non-accounting researchers have conducted
doi:10.1287/isre.12.4.384.9702
most of the existing research on semantic models,
both at the individual level and at the organization Cherrington, J. O., Denna, E. L., & Andros, D. P.
level. Because REA originated in an accounting (1996). Developing an event-based system: The
domain, non-accounting researchers have not case of IBM’s national employee disbursement
embraced it (perhaps because of their lack of system. Journal of Information Systems, 10(1),
awareness of the REA ontology). We hope that by 51–69.
making information about this enterprise ontology
David, J. S. (1995). An empirical analysis of REA
more available to non-accounting researchers who
accounting systems, productivity, and perceptions
are interested in semantically modeled informa-
of competitive advantage. Unpublished doctoral
tion systems we will encourage more interest and
dissertation, Michigan State University.
participation in REA research.
David, J. S., Dunn, C. L., & McCarthy, W. E.
(1999). Enterprise resource planning systems
reFerences research: The necessity of explicating and exam-
ining patters in symbolic form. Working paper,
Abrial, J. R. (1974). Data semantics. In J. W. Arizona State University.
Klimbie, & K. L. Koffeman (Eds.), Data Base
Management. Amsterdam: North Holland Pub- Dunn, C. L., Cherrington, J. O., & Hollander,
lishing Company, (pp. 1-60). A. S. (2005). Enterprise Information Systems: A
Pattern-based Approach, 3rd edition. New York:
Agarwal, R., Sinha, A. P., & Tanniru, M. (1996a). McGraw-Hill Irwin.
Cognitive fit in requirements modeling: A study of
object and process methodologies. Journal of Man- Dunn, C. L., & Grabski, S. V. (1998). The effect
agement Information Systems, 13(2), 137–162. of field independence on conceptual modeling
performance. Advances in Accounting Information
Agarwal, R., Sinha, A. P., & Tanniru, M. (1996b). Systems, 6, 65–77.
The role of prior experience and task character-
istics in object-oriented modeling: An empirical Dunn, C. L., & Grabski, S. V. (2000). Perceived
study. International Journal of Human-Computer semantic expressiveness of accounting systems
Studies, 45, 639–667. doi:10.1006/ijhc.1996.0072 and task accuracy effects. International Journal
of Accounting Information Systems, 1(2), 79–87.
Andros, D., Cherrington, J. O., & Denna, E. L. doi:10.1016/S1467-0895(00)00004-X
(1992). Reengineer your accounting the IBM way.
The Financial Executive, July/August, (pp. 28-31).
APICS. (1998). Defining enterprise resource
planning. http://www.apics.org/OtherServices/
articles/defining.htm.
203
Dunn, C. L., & Grabski, S. V. (2002). Empirical Geerts, G. L., & McCarthy, W. E. (2004). The
research in semantically modeled accounting ontological foundation of REA enterprise infor-
systems. In V. Arnold & S. G. Sutton (Eds.), Re- mation systems. Working paper, Michigan State
searching Accounting as an Information Systems University.
Discipline. Sarasota, FL: American Accounting
Gerard, G. J. (2005). The REA pattern, knowledge
Association (pp. 157-180).
structures, and conceptual modeling performance.
Dunn, C. L., & McCarthy, W. E. (1997). The Journal of Information Systems, 19, 57–77.
REA accounting model: Intellectual heritage and doi:10.2308/jis.2005.19.2.57
prospects for progress. Journal of Information
Gruber, T. (2008). Ontology. In Ling Liu and M.
Systems, 11(1), 31–51.
Tamer Özsu (Eds.) Encyclopedia of Database
Dunn, C. L., & McCarthy, W. E. (2000). Symbols Systems, Springer-Verlag. Found online at http://
used for economic storytelling: A progression from tomgruber.org/writing/ontology-definition-2007.
artifactual to more natural accounting systems. htm (accessed 10/5/07).
Working paper, Michigan State University.
Haugen, R., & McCarthy, W. E. (2000). REA:
Geerts, G. L., & McCarthy, W. E. (1991). Da- A semantic model for internet supply chain col-
tabase accounting systems. IT and Accounting: laboration. Presented at Business Object Com-
The Impact of Information Technology. In B. C. ponent Design and Implementation Workshop
Williams & B. J. Spaul (Eds.), London: Chapman VI: Enterprise Application Integration which is
& Hall, (pp. 159-183). part of The ACM Conference on Object-Oriented
Programming, Systems, Languages, and Applica-
Geerts, G. L., & McCarthy, W. E. (1999). An
tions, October 15-19, Minneapolis, Minnesota.
accounting object infrastructure for knowledge-
based enterprise models. IEEE Intelligent Systems Kim, Y. K., & March, S. T. (1995). Comparing
& Their Applications, (July-August), (pp. 89-94). data modeling formalisms. Communications of the
ACM, 38(6), 103–115. doi:10.1145/203241.203265
Geerts, G. L., & McCarthy, W. E. (2000). Aug-
mented intensional reasoning in knowledge- McCarthy, W. E. (1982). The REA accounting
based accounting systems. Journal of Informa- model: A generalized framework for accounting
tion Systems, 14(2), 127–150. doi:10.2308/ systems in a shared data environment. Accounting
jis.2000.14.2.127 Review, 57(3), 554–578.
Geerts, G. L., & McCarthy, W. E. (2001). Using Nakamura, H., & Johnson, R. E. (1998). Adap-
object templates from the REA accounting model tive Framework for the REA Accounting Model,
to engineer business processes and tasks. The Proceedings of the OOPSLA’98 Business Object
Review of Business Information Systems, 5(4), Workshop IV, http://jeffsutherland.com/oopsla98/
89–108. nakamura.html.
Geerts, G. L., & McCarthy, W. E. (2002). An onto- O’Leary, D. E. (2004). On the relationship between
logical analysis of the primitives of the extended- REA and SAP. International Journal of Accounting
REA enterprise information architecture. Interna- Information Systems, 5(1), 65–81. doi:10.1016/j.
tional Journal of Accounting Information Systems, accinf.2004.02.004
3, 1–16. doi:10.1016/S1467-0895(01)00020-3
204
Rockwell, S. R., & McCarthy, W. E. (1999). Weber, R. (1986). Data models research in ac-
REACH: automated database design integrat- counting: An evaluation of wholesale distribution
ing first-order theories, reconstructive expertise, software. Accounting Review, 61(3), 498–518.
and implementation heuristics for accounting
Weber, R. (1996). Are attributes entities? A
information systems. International Journal
study of database designers’ memory structures.
of Intelligent Systems in Accounting Finance
& Management, 8(3), 181–197. doi:10.1002/
doi:10.1287/isre.7.2.137
(SICI)1099-1174(199909)8:3<181::AID-
ISAF172>3.0.CO;2-E
Satoshi, H. (1999). The contents of interviews
with the project manager of FDWH in IBM Japan.
Proceedings of the 1999 SMAP Workshop, San Business Process: A term widely used in
Diego, CA. business to indicate anything from a single activ-
Scheer, A. W. (1998). Business Process Engineer- ity, such as such as printing a report, to a set of
ing: Reference Models for Industrial Enterprises. activities, such as an entire transaction cycle; in
Berlin: Springer-Verlag. this paper business process is used as a synonym
of transaction cycle. (p. 20)
Sinha, A. P., & Vessey, I. (1999). An empirical Enterprise Resource Planning System: An
investigation of entity-based and object-oriented enterprise wide group of software applications
data modeling. Proceedings of the Twentieth In- centered on an integrated database designed to
ternational Conference on Information Systems. P. support a business process view of the organiza-
De & J. DeGross (Eds.), Charlotte, North Carolina, tion and to balance the supply and demand for its
(pp. 229-244). resources; this software has multiple modules that
Swagerman, D. M., Dogger, N., & Maatman, S. may include manufacturing, distribution, person-
(2000). Electronic markets from a semiotic per- nel, payroll, and financials and is considered to
spective. Electronic Journal of Organizational provide the necessary infrastructure for electronic
Virtualness, 2(2), 22–42. commerce. (p. 2)
Ontologically-Based Information System:
Vessey, I., & Conger, S. A. (1994). Requirements An information system that is based upon a par-
specification: Learning object, process, and data ticular domain ontology, and the ontology provides
methodologies. Communications of the ACM, the semantics inherent within the system. These
37(5), 102–113. doi:10.1145/175290.175305 systems facilitate organizational productivity and
Wakayama, T., Kannapan, S., Khoong, C. M., inter-organizational communication. (p. 3)
Navathe, S., & Yates, J. (1998). Information and Process Level Model: A second level model
Process Integration in Enterprises: Rethinking in the REA ontology that documents the semantic
Documents. Norwell, MA: Kluwer Academic components of all the business process events.
Publishers. (p.20)
Resources-Events-Agents (REA) Ontology:
Watson, E. E., & Schneider, H. (1999). Using A domain ontology that defines constructs com-
ERP systems in education. Communications of mon to all enterprises and demonstrates how those
the Association for Information Systems, 1(9). constructs may be used to design a semantically
modeled enterprise database. (p. 3)
205
Semantically Modeled Database: A database research is needed comparing these two se-
that is a reflection of the reality of the activities mantic models with subsequent evaluations
in which an enterprise engages and the resources of commercially available ERP packages.
and people involved in those activities. The se- 2
The REA framework uses the term business
mantics are present in the conceptual model, but process to mean a set of related business
might not be readily apparent in the implemented events and other activities that are intended
database. (p. 3) to accomplish a strategic objective of an or-
Task Level Model: A third level model in the ganization. In this view, business processes
REA ontology is the most detailed level, which represent a high level of abstraction. Some
specifies all steps necessary for the enterprise to non-REA views define business processes as
accomplish the business events that were included singular activities that are performed within a
at the process level. (p. 6) business. For example, some views consider
Value Chain: The interconnection of business “process sale order” to be a business process.
processes via resources that flow between them, The REA view considers “sale order” to be
with value being added to the resources as they a business event that is made up of specific
flow from one process to the next. (p. 20) tasks, and it interacts with other business
events within the “Sales-Collection” busi-
ness process.
endnotes
1
An alternative semantic model has been
presented by Scheer (1998). Additional
206
207
Chapter 1.15
An Overview of Ontology-
Driven Data Integration1
Agustina Buccella
Universidad Nacional del Comahue, Argentina
Alejandra Cechich
Universidad Nacional del Comahue, Argentina
IntroductIon ering information through a common interface

becomes crucial in order to realize, for instance,
New software requirements have emerged because the full value of data contained in the databases
of innovation in technology, specially involving (Hass & Lin, 2002).
network aspects. The possibility enterprises, in- Thus, in the ‘90s the term Federated Database
stitutions and even common users can improve emerged to characterize techniques for proving an
their connectivity allowing them to work as they integrating data access, resulting in a set of dis-
are at the same time, generates an explosion in tributed, heterogeneous and autonomous databases
this area. Besides, nowadays it is very common (Busse, Kutsche, Leser & Weber, 1999; Litwin,
to hear that large enterprises fuse with others. Mark & Roussoupoulos, 1990; Sheth & Larson,
Therefore, requirements as interoperability and 1990). Here is where the concept of Data Integra-
integrability are part of any type of organization tion appears. This concept refers to the process
around the world. In general, large modern enter- of unifying data sharing some common semantics
prises use different database management systems but originated from unrelated sources. Several
to store and search their critical data. All of these aspects must be taken into account when working
databases are very important for an enterprise but with Federated Systems because the main charac-
the different interfaces they possibly have make teristics of these systems make more difficult the
difficult their administration. Therefore, recov- integration tasks. For example, the autonomy of the
information sources, their geographical distribu-
tion and the heterogeneity among them, are some
DOI: 10.4018/978-1-60566-242-8.ch051
An Overview of Ontology-Driven Data Integration
of the main problems we must face to perform sufficiently comprehensive to support translation
the integration. Autonomy means that users and of all relevant information sources; an ontology
applications can access data through a federated may support consistency management and rec-
system or by their own local system. Distribu- ognition of inconsistent data; etc. Then, the next
tion (Ozsu & Valduriez, 1999) refers to data (or section will analyze several systems using ontolo-
computers) spread among multiple sources and gies as a tool to solve data integration problems.
stored in a single computer system or in multiple
computer systems. These computer systems may
be geographically distributed but interconnected bAckGround
by a communication network. Finally, heteroge-
neity relates to different meanings that may be Recently, the term Federated Databases has
inferred from data stored in databases. In (Cui & evolved to Federated Information Systems be-
O’Brien, 2000), heterogeneity is classified into cause of the diversity of new information sources
four categories: structural, syntactical, system, involved in the federation, such as HTML pages,
and semantic. Structural heterogeneity deals with databases, filing, etc., either static or dynamic. A
inconsistencies produced by different data mod- useful classification of information systems based
els whereas syntactical heterogeneity deals with on the dimensions of distribution and heterogene-
consequences of using different languages and ity can be found in (Busse et al., 1999). Besides, this
data representations. On the other hand, system work defines the classical architecture of federated
heterogeneity deals with having different sup- systems (based on Sheth & Larson (1990)) which
porting hardware and operating systems. Finally, is widely referred by many researches. Figure 1
semantic heterogeneity (Cui & O’Brien, 2000) is shows this architecture.
one of the most complex problems faced by data in- In the figure, the wrapper layer involves a
tegration tasks. Each information source included number of modules belonging to a specific data
in the integration has its own interpretation and organization. These modules know how to retrieve
assumptions about the concepts involved in the data from the underlying sources hiding their data
domain. Therefore, it is very difficult to determine organizations. As the federated system is autono-
when two concepts belonging to different sources mous, local users may access local databases
are related. Some relations among concepts that through their local applications independently
semantic heterogeneity involves are: synonymous, from users of other systems. Otherwise, to access
when the sources use different terms to refer to the federated system, they need to use the user
the same concept; homonymous, when the sources interface layer.
use the same term to denote completely different The federated layer is one of the main com-
concepts; hyponym, when one source contains a ponents currently under analysis and study. Its
term less general than another in another source; importance comes from its responsibility to solve
and hypernym, when one source contains a term the problems related to semantic heterogeneity,
more general than another in another source; etc. as we previously introduced. So far, different
In this paper we will focus on the use of ontolo- approaches have been used to model this layer.
gies because of their advantages when using for In general they use ontologies as tools to solve
data integration. For example, an ontology may these semantic problems among different sources
provide a rich, predefined vocabulary that serves (Stephens, Gangam & Huhns, (2004); Le, Dieng-
as a stable conceptual interface to the databases Kuntz & Gandon, (2004); Giunchiglia, Yatskevich
and is independent of the database schemas; & Giunchiglia, (2005), Buccella, Cechich & Bris-
knowledge represented by the ontology may be aboa, 2003; Buccella, Cechich & Brisaboa, 2004).
208
Figure 1. Architecture of federated systems
ontoloGy-bAsed surveys based on ontology-based systems for data

dAtA InteGrAtIon integration can be found in the literature (Euzenat
& Shvaiko, 2007; Kalfoglou & Schorlemmer,
The term ontology was introduced by Gruber 2003; Wache et al., 2001). For example, the work
(1993) as an “explicit specification of a concep- presented by Euzenat & Shvaiko (2007) describe
tualization”. A conceptualization, in this defini- and analyze a widely set of ontology matching
tion, refers to an abstract model of how people proposals. Although it is focused on methodolo-
commonly think about a real thing in the world; gies to improve matchings, new approaches of
and explicit specification means that concepts complete integrated systems are also taked into
and relationships of the abstract model receive account. Besides, more information with links
explicit names and definitions. to several approaches can be found at Ontology-
An ontology gives the name and description Matching.org2. Another example is (Wache et al.,
of the domain specific entities by using predicates 2001) in which authors focus on some aspects of
that represent relationships between these entities. the use of ontologies, the language representation,
The ontology provides a vocabulary to represent mappings and tools. This work also classifies the
and communicate domain knowledge along with use of ontologies into three approaches: single
a set of relationships containing the vocabulary’s ontology approach, multiple ontology approach
terms at a conceptual level. Therefore, because and hybrid ontology approach, as Figure 2 shows.
of its potential to describe the semantics of in- We use this classification to categorize the use
formation sources and to solve the heterogeneity of ontologies by each proposal. At same time,
problems, the ontologies are being used for data with respect to the integration process and based
integration tasks. on the approaches aforementioned, two new ap-
Recent surveys have emerged describing proaches (Calvanese & Giacomo, 2005; Calva-
and analyzing proposals to ontology matching nese, Giacomo & Lenzerini, 2001) have emerged
(Euzenat & Shvaiko, 2007; Shvaiko & Euzenat, for specifying mappings in an integrated system.
2005; Rahm & Bernstein, 2001). But some of these One, of them, the global-as-view (GAV) approach,
surveys only analyze methodologies and tools is based on the definition of global concepts as
to build an integrated system, without analyzing views over the sources, that is, each concept of
how the whole system works. On the other hand, the global view is mapped to a query over the
209
Figure 2. Classification of the use of ontologies
sources. The other approach, local-as-view (LAV), ontologies. Changeability refers to the ability of
is based on the definition of the sources as views changing some structures within an information
over the global view. Thus, the information the source, without producing substantial changes in
sources contain is described in terms of this the system components. Finally, scalability refers
global view. to the possibility of easily adding new information
In this section we will focus on how ontology- sources to the integrated system.
based systems address the semantic heterogeneity In order to give a wide view of ontology-based
problems. We have investigated many systems, systems, we divide the approaches into two branch-
which follow some of the approaches of Figure es. On one branch, we briefly analyze systems
2, considering relevant aspects respect to the use proposed during the 90’s decade and no longer
of ontologies. They are, reusability, changeability under research. Table 1 shows our classification
and scalability. These aspects allow us analyze taking into account some of these systems. The
whether quality properties are improved due a columns of the table contain the three relevant as-
particular use of ontologies in a given system. pects within the use of ontologies aforementioned.
The use of ontologies refers to how ontologies A more extensive analysis of these systems can be
help solve data integration problems. Commonly, found in (Buccella, Cechich & Brisaboa, 2005a;
the systems show how ontologies are used to Buccella, Cechich & Brisaboa, 2005b).
solve the integration problems and how ontologies As we can see, SIMS (Ambite et al., 1997)
interact with other architectural components. In and Carnot (Woelk, Cannata, Huhns, Shen &
this aspect the systems describe its ontological Tomlinson, 1993) are scored as “Not supported”
components and the ways to solve the different in the three columns. In the first column this
semantic heterogeneity problems. value means that both ontologies are not reusable
Then, we characterize this feature by means of because both define a global ontology or a single
three quality characteristics: reusability, change- ontology approach in order to integrate data. The
ability and scalability. Reusability refers to the same happens with the second column, change-
ability of reusing the ontologies, that is, ontologies ability, because no support is given by these
defined to solve other problems can be used in a systems to bear changes in the information
system because of either the system supports dif- sources. When one source changes, the global
ferent ontological languages (and) or define local ontology must be rebuilt.
210
Table 1. Ontology-based systems during 90’s decade
Approach Ontology-based Systems Reusability Changeability Scalability

SIMS (Ambite et al., 1997)
Single Ontology Ap-
Carnot (Woelk, Cannata, Huhns, Shen & Not supported Not supported Not supported
proach
Tomlinson, 1993)
Multiple Ontology Ap- OBSERVER (Mena, Kashyap, Sheth &
Fully-supported Supported Supported
proach Illarramendi, 2000)
KRAFT (Preece, Hui, Gray, Jones & Cui,
Supported Semi-supported Semi-supported
1999)
Hybrid Ontology Ap-
COIN (Firat, Madnick & Grosof, 2002; Goh
proach Supported Supported Supported
et al., 1999)
InfoSleuth (Bayardo et al., 1997) Supported Supported Supported
Finally, scalability, is not supported because that most of the new systems do not follow the
again by adding a new information source, the single ontology approach. The inherent disad-
rebuilding of the global ontology is generated. vantages of this type of systems make difficult to
The problem of dealing with the global ontology deal with them. However, some systems such as
approach is that we must manage a global inte- MediWeb (Arruda, Lima & Baptista, 2002) and
grated ontology, which involves administration, DIA (Medcraft, Schiel & Baptista, 2003) apply this
maintenance, consistency and efficiency problems approach but without better results. Only in the
that are very hard to solve. case of DIA the changeability and scalability can
On the other hand, OBSERVER system (Mena, be improved because the global ontology does not
Kashyap, Sheth & Illarramendi, 2000) uses the need to be modified when a source is added; only
multiple ontologies approach to alleviate the an ontology-schema matching table is necessary
problems presented by the single approach. As to add a new database to the integrated system.
each ontology can be created independently, the As a new proposal within the multiple ontology
reusability is fully supported. However, change- approach we can cite OntoGrate (Dou & LePendu,
ability and scalability are semi-supported because 2006). Here, first order logic inference is used to
changes in local ontologies also generate changes find mappings among underlying ontologies. By a
in the mappings. set of rules and user assistance, databases (that are
Finally, the three last systems following the the sources of the system) are translated to source
hybrid ontology approach are scored better. ontologies. Then, these ontologies are merged
Although they propose different and novel ways by applying bridging axioms and a reasoner. In
to improve these characteristics, all these mecha- this system, reusability is not supported because
nisms conclude in good results. For example, in the ontologies are created based on the underly-
InfoSleuth (Bayardo et al., 1997), to add a resource ing databases by using a specific language and
to the system it only needs to have an interface to particularities. For changeability or scalability,
advertise its services and let other agents make the reasoning process must be re-executed over
use of them immediately. the information changed or added. The speed of
On the other branch, we analyze more recent this process will depend on how large the ontolo-
systems widely referred by researchers, although gies are.
some of them are still under development. We use Recently, new approaches have emerged by
the same characteristics in order to analyze the applying the advantages the hybrid ontology
evolution of the systems. For example, we notice approach provides. Some novel approaches are
211
MOBS (McCann, Doan, Varadarajan & Kramnik, the LAV approach, adding a new source to the
2003), AutoMed (Zamboulis & Poulovassilis, integrated system does not require changes in
2006) and Alasoud et.al (Alasoud, Haarslev & the global view. On the other hand, in the GAV
Shiri, 2005). In MOBS (Mass Collaboration to approach when a new source is added, changes
Build Systems) something different is proposed. over the global view are required. However, scal-
Instead of creating a global ontology based on ability over the global view is easier in this last
underlying domain ontologies, all of them are cre- approach. Finally, considering query processing,
ated at the same time. That is, in this first step the LAV requires reasoning mechanisms in order to
global ontology contains general concepts users answer them. Contrarily, in GAV conventional
want to recover instead of information extracted mechanisms can be implemented.
from underlying sources. Then, initial mappings In order to take advantage of the benefits of
are defined by using a random assignment or a both approaches, a new approach, called global-
schema matching tool. Finally, the mass collabora- local-as-view (GLAV), has been proposed by
tion of this approach is in the user feedback needed (Friedman, Levy & Millstein, 1999). Although
to readjust the system and to define the correct the query process is out of the scope of this work,
mappings. In this way, the ontologies are reusable we consider that a study of query capabilities of
because they can be used by this system or by each proposal shoud be performed. For example,
another one. But changeability and scalability can in (Calvanese, Giacomo & Lenzerini, 2001) two
be tedious because the user collaboration process case studies are presented in order to show the
must be re-run each time something is modified or need of suitable techniques for query answering.
added. In AutoMed and Alasoud et.al approaches
materialized views can be defined. In the case of
Alasoud et.al an integrated view (as a global on- Future trends
tology) is partially materialized to improve query
answering time, and to take advantages of fully Several systems compared here are still in a de-
materialized and fully virtual approaches. In this velopment stage and, as we have explained, some
way, the answers to queries are retrieved from problems should be solved in order to reach a good
materialized as well as virtual views. Besides, the integration. Recent proposals, which are based on
mappings between concepts in the source ontolo- the old ones, try to improve several aspects in order
gies are described in terms of the integrated view. to propose an integral solution. For example, new
Thus, adding a new ontology to this system does systems do not apply a global ontology approach
not require changes in the global view. such as SIMS did, due to problems as scalability
Considering the approach to specify mappings, and changeability.
the GAV approach is generally chosen by the pro- Other proposals applying the others ap-
posals. Only Alasoud et.al and OntoGrade systems proaches (multiple and hybrid), intent to solve
define something similar to the LAV approach. some characteristics of the global approach, but
Advantages and disadvantages of each ap- additional efforts have to be done to reach them.
proach have to be considered when an integration For example, reusability deserves more attention
process is initiated (Cali et.al, 2002). For example, because several ontologies created for the semantic
in the LAV approach restrictions over the sources web might be part (some day) of an integrated
can be defined easily. On the contrary, in the GAV system. In this way, thinking of particular ontol-
approach it is easier to define restrictions over the ogy with specific requirements, as in OntoGrate,
global view. With respect to scalability, following is not such a good idea.
212
There are several proposals (Euzenat & Sh- ontology-based systems. Among them, we should
vaiko, 2007), which were not analyzed here, only mention the use of automated tools for supporting
implementing methods for semantic matching. the manipulation of ontologies.
These proposals are complementary because they Of course, further characterization is needed
propose mechanisms to create the global view to completely understand the use of ontologies for
or the mappings in an efficient way. Euzenat & data integration. Hope our work will motivate the
Shvaiko (2007) presents a survey that is mainly reader to deeper immerse in this interesting world.
focused on ontology matching proposals. The
task of ontology matching involves the process
of finding relationships or correspondences reFerences
between entities of different ontologies. In this
paper these proposals are not considered because Alasoud, A., Haarslev, V., & Shiri, N. (2005). A
we do not focus on this type of problems but on hybrid approach for ontology integration. Pro-
the functionality of the whole integrated system. ceedings of the 31st VLDB Conference. Trond-
Nevertheless, semantic matching is a crucial area heim, Norway.
that must be analyzed to assess the processes the Ambite, J. L., Arens, Y., Ashish, N., Knoblock,
system involves. C. A., & collaborators (1997). The SIMS manual
Currently, the task of building an integrated 2.0. Technical Report, University of Southern,
system is not easy. By trying to improve some es- California. December 22. Retrieved August 01,
sential aspects, other equally important ones can 2007, from http://www.isi.edu/sims/papers/sims-
be forgotten. For example, in order to improve manual.ps.
methods for finding mappings, more expressive
ontologies are created. But aspects as understand- Arruda, L., Baptista, C., & Lima, C. (2002).
ability and complexity are not taken into account. MEDIWEB: A mediator-based environment
Furthermore, new systems must consider all these for data integration on the Web. Databases and
aspects in order to give an integral solution. Information Systems Integration. ICEIS, 34-41.
Baru, C. (1998). Features and requirements for
an XML view definition language: Lessons from
conclusIon XML information mediation. In W3CWorkshop
on Query Language (QL´98). Boston.
Today, semantic heterogeneity implies many
complex problems that are addressed by using Bayardo, R. J., Jr., Bohrer, W., Brice, R., Cichocki,
different approaches, including ontologies, which A., Fowler, J., Helal, A., et al. (1997). InfoSleuth:
give a higher degree of semantics to the treatment Agent-based semantic integration of information
of data involved in the integration. in open and dynamic environments. Proceedings
We have analyzed several systems accordingly ACM SIGMOD International Conference on
to the use of the ontologies and we have evaluated Management of Data, 195-206.
three important aspects also related with them –
Buccella, A., Cechich, A., & Brisaboa, N. R.
reusability, changeability and scalability. Each
(2003). An ontology approach to data integration.
system has implemented its own solution with
Journal of Computer Science and Technology,
advantages and disadvantages, but some ele-
3(2), 62–68.
ments in common and some original aspects can
be found. There are other important aspects we
could have considered for characterizing current
213
Buccella, A., Cechich, A., & Brisaboa, N. R. Euzenat, J., & Shvaiko, P. (2007). Ontology
(2004). A federated layer to integrate heteroge- matching. Heidelberg: Spinger-Verlag, (DE), isbn
neous knowledge. In VODCA’04 First Interna- 3-540-49611-4. (pp. 341).
tional Workshop on Views on Designing Complex
Firat, A., Madnick, S., & Grosof, B. (2002).
Architectures. Bertinoro, Italy. Electronic Notes in
Knowledge integration to overcome ontological
Theoretical Computer Science, Elsevier Science
heterogeneity: Challenges from financial informa-
B.V, 101-118.
tion systems. Twenty-Third International Confer-
Buccella, A., Cechich, A., & Brisaboa, N. R. ence on Information Systems, ICIS.
(2005a). Ontology-based data integration: Differ-
Friedman, M., Levy, A., & Millstein, T. (1999).
ent approaches and common features. Encyclope-
Navigational plans for data integration. AAAI/
dia of Database Technologies and Applications.
IAAI, 67-73.
Rivero L., Doorn J., and Ferraggine V. Editors.
Idea Group. Giunchiglia, F., Yatskevich, M., & Giunchiglia,
E. (2005). Efficient semantic matching. In A.
Buccella, A., Cechich, A., & Brisaboa, N. R.
Gomez-Perez and J. Euzenat, editors, ESWC 2005,
(2005b). Ontology-based data integration meth-
volume LNCS 3532, 272-289. Springer-Verlag.
ods: A framework for comparison. Revista Co-
lombiana de Computación 6(1). Goh, C. H. (1996). Representing and reasoning
about semantic conflicts in heterogeneous infor-
Busse, S., Kutsche, R. D., Leser, U., & Weber, H.
mation sources. Phd, MIT, Massachusetts Institute
(1999). Federated information systems: Concepts,
of Technology, Sloan School of Management.
terminology and architectures. Technical Report.
Nr. 99-9, TU Berlin. Goh, C. H., Bressan, S., Siegel, M., & Mad-
nick, S. E. (1999). Context interchange: New
Calí, A., Calvanese, D., Giacomo, D., & Len-
features and formalisms for the intelligent
zerini, M. (2002). On the expressive power of
integration of information. ACM Transac-
data integration systems. In Proceedings of the
tions on Information Systems, 17(3), 270–293.
21st Int. Conf. on Conceptual Modeling, ER, 2503
doi:10.1145/314516.314520
of Lecture Notes in Computer Science. Springer.
Gruber, T. (1993). A translation approach to por-
Calvanese, D., & Giacomo, G. D. (2005). Data
table ontology specifications. Knowledge Acquisi-
integration: A logic-based perspective. AI Maga-
tion, 5(2), 199–220. doi:10.1006/knac.1993.1008
zine, 26(1), 59–70.
Hass, L., & Lin, E. (2002). IBM federated da-
Calvanese, D., Giacomo, G. D., & Lenzerini, M.
tabase technology. Retrieved August 01, 2005,
(2001). A framework for ontology integration. In
from http://www-106.ibm.com/developerworks/
SWWS, 303-316.
db2/library/techarticle/0203haas/0203haas.html.
Cui, Z., & O’Brien, P. (2000). Domain ontology
Kalfoglou, Y., & Schorlemmer, M. (2003). Ontol-
management environment. In Proceedings of the
ogy mapping: The state of the art. The Knowledge
33rd Hawaii International Conference on System
Engineering Review, 18(1), 1–31. doi:10.1017/
Sciences.
S0269888903000651
Dou, D., & LePendu, P. (2006). Ontology-based
Integration for relational databases. SAC, 06,
461–466.
214
Le, B. T., Dieng-Kuntz, R., & Gandon, F. (2004). Shvaiko, P., & Euzenat, J. (2005). A survey of sche-
On ontology matching problems for building a ma-based matching approaches. Journal of Data
corporate Semantic Web in a multi-communities Semantics, IV, 146–171. doi:10.1007/11603412_5
organization. In ICEIS 2004 Software Agents and
Stephens, L., Gangam, A., & Huhns, M. (2004).
Internet Computing, 236-243.
Constructing consensus ontologies for the Seman-
Litwin, W., Mark, L., & Roussoupoulos, N. (1990). tic Web: A conceptual approach. In World Wide
Interoperability of multiple autonomous data- Web [Kluwer Academic Publishers.]. Internet and
bases. ACM Computing Surveys, 22(3), 267–293. Web Information Systems, 7, 421–442.
doi:10.1145/96602.96608
Wache, H., Vögele, T., Visser, U., Stuckenschmidt,
McCann, R., Doan, A., Varadarajan, V., & Kram- H., Schuster, G., Neumann, H., & Hübner, S.
nik, A. (2003). Building data integration systems (2001). Ontology-based integration of information
via mass collaboration. International Workshop - A survey of existing approaches. In Proceedings
on the Web and Databases. California: WebDB of IJCAI-01 Workshop: Ontologies and Informa-
tion Sharing, Seattle, WA, (pp. 108-117).
Medcraft, P., Schiel, U., & Baptista, P. (2003).
DIA: Data integration using agents. Databases and Woelk, D., Cannata, P., Huhns, M., Shen, W., &
Information Systems Integration. ICEIS, 79-86. Tomlinson, C. (1993). Using Carnot for enterprise
information integration. Second International
Mena, E., Kashyap, V., Sheth, A., & Illarramendi,
Conf. Parallel and Distributed Information Sys-
A. (2000). Observer: An approach for query
tems, 133-136.
processing in global information systems based
on interoperation across pre-existing ontologies. Zamboulis, L., & Poulovassilis, A. (2006). In-
Kluwer Academic Publishers, Boston, (pp. 1-49). formation sharing for the Semantic Web — A
schema transformation approach. In Proceedings
Ozsu, M. T., & Valduriez, P. (1999). Principles of
of DISWeb06, CAiSE06 Workshop Proceedings,
distributed database systems, (2nd ed.), Prentice
275-289.
Hall.
Preece, A., Hui, K., Gray, A., Jones, D., & Cui, Z.
(1999). The KRAFT architecture for knowledge
fusion and transformation. In 19th SGES Interna-
tional Conference on Knowledge-based Systems Data Integration: Process of unifying data
and Applied Artificial Intelligence (ES’99). Berlin: that share some common semantics but originate
Springer. from unrelated sources.
Rahm, E., & Bernstein, A. (2001). A survey of Distributed Information System: A set of
approaches to automatic schema matching. [New information systems physically distributed over
York: Springer-Verlag Inc.]. The VLDB Journal, multiple sites, which are connected with some
4(10), 334–350. doi:10.1007/s007780100057 kind of communication network.
Federated Database: Idem FIS, but the in-
Sheth, A. P., & Larson, J. A. (1990). Feder- formation systems only involve databases (i.e.
ated database systems for managing distrib- structured sources).
uted, heterogeneous and autonomous databases. Federated Information System (FIS): A set
ACM Computing Surveys, 22(3), 183–236. of autonomous, distributed and heterogeneous
doi:10.1145/96602.96604 information systems, which are operated together
to generate a useful answer to users.
215
Heterogeneous Information System: A set Ontology Matching: the process of finding

of information systems that differs in syntactical relationships or correspondences between entities
or logical aspects like hardware platforms, data of different ontologies.
models or semantics. Semantic Heterogeneity: Each information
Ontological Reusability: The ability of source has a specific vocabulary according to
creating ontologies that can be used in different its understanding of the world. The different
contexts or systems. interpretations of the terms within each of these
Ontological Changeability: The ability of vocabularies cause the semantic heterogeneity.
changing some structures of an information source
without producing substantial changes in the
ontological components of the integrated system. endnotes
Ontological Scalability: The ability of easily
adding new information sources without generate
1
This work is partially supported by the
substantial changes in the ontological components UNCOMA project 04/E059.
of the integrated system.
2
http://www.ontologymatching.org
Ontology: It provides a vocabulary to represent
and communicate knowledge about the domain
and a set of relationship containing the terms of
the vocabulary at a conceptual level.
216
217
Chapter 1.16
A Fundamental SOA Approach
to Rebuilding Enterprise
Architecture for a Local
Government after a Disaster
Zachary B. Wheeler
SDDM Technology, USA
AbstrAct SOA, utilizing the remaining local assets, XML

(extensible markup language), and Web services.
As a result of Hurricane Katrina, the destruction
of property, assets, documentation, and human
life in the Gulf Port has introduced a myriad of IntroductIon
challenging issues. These issues involve human,
social, government, and technological concerns. Disaster preparedness, response, and recovery
This chapter does not address the many imme- received a lot of attention immediately after the
diate human and social concerns brought forth terrorist attacks of 9/11 and eventually faded
from a natural disaster or major terrorist attack from the forefront of attention after the invasion
(NDMTA); this chapter addresses a small but of Iraq and the global war on terrorism. However,
significant problem of re-establishing or laying recent natural disasters such as the Indonesian
the groundwork for an enterprise architecture for Tsunami in 2004 and the devastating Hurricane
local government during the response phase of Katrina in Louisiana have refocused attention on
the disaster. Specifically, it addresses construct- these three prominent areas. Specifically, the lack
ing a high-level data model and fundamental of preparedness, inadequate response, and slow
recovery has burdened local, state, and federal
DOI: 10.4018/978-1-60566-330-2.ch015 governments as well as citizens.
A Fundamental SOA Approach to Rebuilding Enterprise Architecture for a Local Government after a Disaster
The presented enterprise approach and imple- typically accomplished by establishing various
mentation process covers an area that is void in the local agencies and departments. These departments
disaster preparedness and response phase; how- are structured to provide essential services for the
ever, it is applicable in each phase: preparedness, community. For instance, the fire department role
response, and recovery. It is recommended that is to help citizens in immediate danger due to fire,
the presented approach be included as part of the gas, or chemical hazard. The role of the health
disaster preparedness phase, implemented in the department is to establish policy, programs, and
response phase, and eventually expanded in the standards regarding health and health related issues.
recovery phase. The approach is unique because An additional role of the health department is to as-
the enterprise implementation takes place during sist citizens in obtaining basic health care services.
the actual response phase of the disaster and uti- Each established department or agency has a role in
lization of the fundamental SOA leads to further assisting the community and its residents by provid-
expansion during and after the recovery phase. ing relevant services. In a typical municipality, each
The approach introduced in this chapter takes agency has a database of information relating to
advantage of the Zachman framework system the citizens and the services provided to the citizen
model perspective by utilizing Web services on a by the agency. For instance, the police department
local level and introducing a practical but efficient maintains a database of criminals, criminal activity,
method for populating the initial data model. A and citizen complaints. The Department of Human
series of basic assumptions are introduced based Services maintains a database of child immunization
on information regarding the recent Gulf Port, records. In short, each agency maintains a database
Hurricane Andrew, Indonesian Tsunami, and 9/11 and application system to enter data, process data,
disaster events. These assumptions are based on and execute business rules. However, in the wake of
the physical, environmental, and technological an NDMTA, these systems along with other IT as-
conditions immediately after disaster strikes. sets are destroyed or rendered useless. For instance,
The assumptions are there will be limited or non- Hurricane Katrina destroyed most of New Orleans
existent landline and wireless communication, a including property, buildings, human life, landline
lack of ability to use generators for power source, and mobile communications, Internet services, in-
limited or nonexistent Internet and intranet, major tranet services, and essentially incapacitated local
IT system destruction, and the incapacitation of government. In the terror attacks of 9/11, the same
local government services. asset destruction was prevalent within a specified
This chapter addresses the problem of re- geographic area. Hurricane Andrew wreaked havoc
establishing or laying the groundwork for an among Florida communities and followed the same
enterprise architecture for local government during line of asset destruction and local government in-
the response phase of the disaster. Specifically, capacitation as Hurricane Katrina. In each of these
it addresses constructing a high-level data model cases, major response and rebuilding were needed
and fundamental SOA by utilizing the remaining to help reestablish public safety, government, and
local assets, XML, and Web services. services to the remaining citizens. This approach
suggests that reestablishing a basic framework for
IT services can be facilitated during the response
bAckGround phase of a disaster. In that regard, the proposed
approach is unique in that the role of rebuilding
The fundamental role of local government is to typically takes place during the recovery phase
protect the people, provide basic human services, (University of Florida, 1998).
and assist in strengthening communities. This is
218
The extended Zachman Framework system In the conclusion, the what, how, and where ques-
model perspective will be utilized to establish high- tions of the ISA will be answered.
level data elements for the model. The utilization
of Web services will be used to lay down a basic
framework for a fundamental service oriented mAIn tHrust oF tHe cHAPter
architecture that can be extended to an enterprise
level once essential government services have been basis for a conceptual data model
restored. In addition, a data collection process is
provided for initial population of the primary data The ISA system model perspective represents the
elements from the remaining survivors. system analyst role in information technology.
The system analyst is responsible for determining
the system model and Zachman the data elements and functions that represent the
business entities and processes. Zachman sug-
In the initial framework provided by Zachman gests introducing all of the entities; however, the
(1987), he identifies five different perspectives construction of all data elements, processes, and
of an enterprise architecture, three views of functions for a local government would be beyond
the enterprise, and introduces the six questions the scope of this chapter, therefore, a high-level
pertaining to an enterprise. The six questions are perspective for core data elements utilized during
what, how, where, who, when, and why. (Table 1) the response phase will be presented.
Zachman provides a clear and concise identi-
fication of the various views of an enterprise and Primary data elements
shows how each view is proper and correct. In
1992, the Zachman framework was extended by One of the main priorities of local government is
Zachman and Sowa (1992). In addition to answer- to provide services to the citizens of the com-
ing the final three questions, they introduce the munity. Regardless of the service provided, most
conceptual graph to represent the ISA and replace government agencies interact with its residents
the “model of the information system” with the and maintain some form of database of citizen
more generic system model reference for row 3 information. In a disaster area, the citizens of the
or the designer perspective. Hence, the various community are the disaster survivors. From a data
perspectives identified by Zachman are scope, acquisition perspective, we can obtain valuable
enterprise model, system model, technology information from the survivors and with this in-
model, and components. Our perspective will formation begin to develop a conceptual data
cover the system model or designer perspective. model for the emerging enterprise. Typically, the
conceptual data model does not show actual data
details of the entities. Instead, the conceptual data
Table 1. Zachman’s enterprise questions model provides a high-level entity view using the
entity relationship diagram (ERD) (Rob & Coro-
Zachman’s Six Enterprise Questions
nel, 2002). Entity details will be provided in
What? What entities are involved?
tabular format for clarity; however, the ERD will
How? How they are processed?
only show the entities and the defined relation-
Where? Where they are located?
ships. Utilizing the following assumptions, we
Who? Who works with the system?
can define each remaining citizen as unique:
When? When events occur?
(Table 2, Table 3, Table 4, and Table 5)
Why? Why these activities are taking place?
219
Table 2. Person entity details Table 5. Police entity object relating to person
PERSON POLICE
Unique_Id Not Null Unique_Id Not Null
Unique_ID_Type Not Null Unique_Id_Type Not Null
First_Name Not Null Arrested(Y,N) Not Null
Middle_Name Officer_Id Not Null
Last_Name Not Null Comments
Name_Suffix (i.e. Jr, Sr, etc….) Crime_Id Not Null
Date_of_Birth Not Null
Sex Not Null
Status(Living,Deceased) Not Null • Every entity has a name (names are not
Phone(Optional) unique).
Address_Id • Every entity has or had an associated ad-
dress (address is not unique).
• Every entity has a sex (unique).
Table 3. Address entity details • Every entity will have an identifying
unique number (ID card, green card, fed-
ADDRESS
eral employee identification number, social
Address_Id Not Null
security card, or driver’s license).
Street_Number Not Null
Prefix
Street_Name Not Null Note: Newborns or infants will be given a tempo-
Street_Type Not Null rary generic unique id if they do not have a SSN.
PostDir If we further assume that each remaining sur-
City Not Null vivor (citizen) has an associated address then we
State Not Null can define the following address entity.
Zip We are working under the assumption that local
Current_Address (Y,N) Not Null assets and asset information have been destroyed,
which includes the destruction of roads, streets,
Table 4. Essential local government agencies bridges, highways, and previously existing ad-
dresses. Thus, when the data collection process
Essential Department/Agencies begins, during the response phase, local officials,
for maintaining order and protect- or management can glean a geographic represen-
Police Department
ing the people from physical harm
tation of survivors and establish a basic address
for maintaining control and
administering of basic health ser- information repository. During the recovery phase,
Department of Health
vices including disease and disease old and new street, road, bridge, highway and
outbreak
address information will be added to the system
Emergency Medical for assisting in medical data collec-
Services tion and medical services thus creating a complete address reference or
for cleaning and clearing of debris,
even an address database. For instance, an entity
Department of Public
Works
corpses and other related health that contains parcel information (square, suffix,
hazards
and lot) and an instance that contains ownership
for maintaining fire, gas, and information (owner name, owner address, etc…)
Fire Department chemical controls and basic rescue
operations will be needed, however, during the response
220
phase, only the address entity defined above is data elements to begin the construction of our
necessary. The person and address entities are conceptual data model. In the overall enterprise,
termed primary data elements. entities can be identified and added as service
agencies are added or new requirements are
secondary data elements: determined.
extending the core elements
In the event of NMDATA, there must be a continu- Web servIces

ation of basic essential services for the remaining
citizens. These essential services required during The construction of our enterprise architecture,
a disaster, according to Davis (1998), are public from a technology perspective, relies on the
safety, public works, and health services. This
position is further bolstered by our knowledge of Table 7. EMS entity object relating to person
the recent events in the Gulf Port and the estab-
lishment of the following five essential services EMS
for that particular region. (Table 6, Table 7, Table Unique_Id Not Null
8. and Table 9) Unique_Id_Type Not Null
Based on the five essential departments, five Service_Provided_Id Not Null
basic data elements can be identified. These es- EMS_ID Not Null
sential data elements are termed secondary data Comments
elements. Although there are more possible data Service_Provided_Id Not Null
elements than presented here, an expanded view
of potential secondary elements is provided for
clarity. Table 8. Public works entity object relating to
Now that the primary and secondary elements person and address
have been identified, we have enough high-level
PUBLIC WORKS
Work_Order_Id Not Null
Table 6. Health services entity object relating to
Unique_Id
person
Unique_Id_Type
HEALTH Address_Id
Unique_Id Not Null Comments
Unique_Id_Type Not Null

Temperature Not Null
Table 9. Fire department entity object relating to
Eyes(Normal,Dialated) Not Null
person and address
Blood_Pressure_Systollic Not Null
Blood_Pressure_Diastollic Not Null FIRE
Heart_Rate Not Null Call_Id Not Null
Recommendations Response_Unit_Id Not Null
Comments Address_Id Not Null
Treatment Unique_Id
Medicine_Prescribed Unique_Id_Type
Disease_Id Comments
221
utilization of the data model, Web services, and it will receive a reply in format B. The
SOA. In our approach, we take advantage of WSDL has two basic strengths:
three different definitions of a Web service while ° It enforces the separation between the
saliently maintaining that a Web service, based interface and implementation.
on the Web services architecture, is considered a ° WSDL is inherently extensible.
software system (Guruge 2004).
uddI
• Definition 1: Web services are modular,
self-contained “applications” or applica- • A discovery mechanism used to discover
tion logic developed per a set of open stan- available services.
dards (Guruge, 2004).
• Definition 2: Web services are extensi-
ble markup language (XML) application Although a Web service is a particular imple-
mapped to programs, objects, or databases mentation of a protocol (SOAP), Web services
or to comprehensive business functions description language (WSDL) and UDDI, the Web
(Newcomer, 2002) service is composed of one or more independent
• Definition 3: A Web service is a particular services. A service represents a particular func-
implementation of a protocol (SOAP), Web tion of the system and has a well-defined, formal
services description language (WSDL), interface called its service contract that:
and universal description discovery and in-
tegration (UDDI) (Fermantle, 2002) where • Defines what the service does and
• SOAP • Separates the services externally acces-
• Uses a RPC or a request-response mecha- sible interface from the services technical
nism based on HTTP. implementation (Newcomer 2002).
• Utilizes an XML message format that con-
tains an address, possible header, and body. For instance, a Web service can contain a
• Contains one or more elements. service that performs the function of adding data,
° The elements are defined using another service that performs the function of re-
common interoperable data formats trieving data, and another service that performs
(integers, strings, and doubles). the function of generating reports for management.
° The parameters are maybe encoded A service can be either an atomic (simple) or a
as child elements of a common parent composite (complex) service. An atomic service
whose name indicates the operation and does not rely on other services and are usually
whose namespace indicates the service. associated with straightforward business trans-
• Can be sent over a common transport actions or with executing data queries and data
typically-HTTP. updates (Newcomer, 2002). A composite service
uses other services, has a well-defined service
Wsdl contract, is registered in the service registry, can
be looked up via the service registry, and can be in-
• Offers the ability to describe the inputs and voked like any other service provider (Newcomer,
outputs of a Web service. 2002). Regardless of the service type (atomic or
• Allows a Web service to publish the inter- composite), the services are required to satisfy the
face of a service, thus if a client sends a following basic requirements (Fermantle, 2002):
SOAP message in format A to the service,
222
• Technology neutral: Each service is non- services of person and address. An example is
technology dependent and can be invoked presented in the following screen shot for clarity.
through the standardized lowest common The PersonAddress Composite service will be
denominator technologies. used in the initial data collection process for the di-
• Loosely coupled: Each service has a life of saster survivors. The SOAP XML message format
its own, each service remains independent representation for the PersonAddress_Composite
of all other services, and each service does service is provided for clarity.
not have knowledge about other services.
• Support location transparency: Services POST /Primary_Core_Service/Ser-
should have their definition and location vice1.asmx HTTP/1.1
information stored in a repository such as Host: localhost
UDDI and is accessible by a variety of cli- Content-Type: text/xml;
ents that can locate and invoke the services charset=utf-8
irrespective of their location. Content-Length: length
SOAPAction: “http://tempuri.org/
the basic services Primary_Core_Service/Service1/
PersonAddress_Composite”
During the response phase and part of the recov- <?xml version=”1.0”
ery phase, several assumptions are made, such as encoding=”utf-8”?>
limited landlines, limited mobile communications, <soap:Envelope
and limited Internet and intranet services. The xmlns:xsi=”http://www.
main objective, however, is to form a basic frame- w3.org/2001/XMLSchema-in-
work using Zachman’s framework (system model stance” xmlns:xsd=”http://
perspective), Web services, and service-oriented www.w3.org/2001/XMLSchema”
architecture. By maintaining our focus on basic xmlns:soap=”http://schemas.xml-
services for the Web service, a foundation is cre- soap.org/soap/envelope/”>
ated for extending our Web services to a service <soap:Body>
oriented architecture later in the recovery phase. <PersonAddress_Composite
If we utilize the best practice approach of xmlns=”http://tempuri.org/Pri-
Krafzig, Banke, and Slama (2004), we can identify mary_Core_Service/Service1”>
two crucial basic service types: simple data-centric <Person>
services and logic-centric services. A data-centric <Person>
service is used to handle data manipulation, data <Person_ID>string</Person_
storage, and data retrieval (Krafizig, Banke, & ID>
Slama 2004). We can easily incorporate logic- <Person_Id_Type>string</
centric services at a later date to handle business Person_Id_Type>
processing and application logic. In a data centric <First_Name>string</First_
service, an entity can be encapsulated into a service Name>
(Krafizig et al., 2004). This encapsulation acts as <Middle_Name>string</Mid-
data layer and all services developed in the future dle_Name>
will have to access these services to access and <Last_Name>string</Last_
manipulate the data. In this chapter, the primary Name>
data elements are wrapped into services and then a <Name_Suffix>string</Name_
composite service is created that utilizes the simple Suffix>
223
<Date_Of_Birth>string</ sponse xmlns=”http://tempuri.

Date_Of_Birth> org/Primary_Core_Service/Ser-
<Persons_Sex>string</Per- vice1”>
sons_Sex> <PersonAddress_
<Living_Status>string</Liv- CompositeResult>string</Person-
ing_Status> Address_CompositeResult>
<Phone>string</Phone> </PersonAddress_CompositeRe-
</Person> sponse>
<Address> </soap:Body>
<Address_Id>int</Address_ </soap:Envelope>
Id>
<Street_Number>string</ Individual person and address services were
Street_Number> necessary for the initial data population and to
<Predir>string</Predir> make data available to entities and agencies as
<Street_Name>string</ they are developed. For instance, the police may
Street_Name> spot a crime taking place at a particular address
<Postdir>string</Postdir> thus they must be able to retrieve or add that
<Suite_Apt>string</Suite_ address to the database thereby identifying the
Apt> crime location. In another instance, the DMV will
<City>string</City> require basic person and address data for license
<State>string</State> issuance, fines, and motor vehicle infractions. The
<Zip>string</Zip> creation of individual and composites services,
<Current_Address>string</ using data centric services, for each of the es-
Current_Address> sential agencies can be generated for immediate
</Address> data collection and tracking purposes. Later in
</Person> the recovery phase, logic centric services can be
</PersonAddress_Composite> integrated to provide business rule processing. In
</soap:Body> the example, below the services are extended to
</soap:Envelope> include the Department of Health.
HTTP/1.1 200 OK The SOAP message format for the Health
Content-Type: text/xml; Service and PersonHealth_Service is provided
charset=utf-8 below for clarity
Content-Length: length
<?xml version=”1.0” POST /Primary_Core_Service/Ser-
encoding=”utf-8”?> vice1.asmx HTTP/1.1
<soap:Envelope Host: localhost
xmlns:xsi=”http://www. Content-Type: text/xml;
w3.org/2001/XMLSchema-in- charset=utf-8
stance” xmlns:xsd=”http:// Content-Length: length
www.w3.org/2001/XMLSchema” SOAPAction: “http://tempuri.org/
xmlns:soap=”http://schemas.xml- Primary_Core_Service/Service1/
soap.org/soap/envelope/”> Health_Service”
<soap:Body> <?xml version=”1.0”
<PersonAddress_CompositeRe- encoding=”utf-8”?>
224
<soap:Envelope Content-Length: length

xmlns:xsi=”http://www. <?xml version=”1.0”
w3.org/2001/XMLSchema-in- encoding=”utf-8”?>
stance” xmlns:xsd=”http:// <soap:Envelope
www.w3.org/2001/XMLSchema” xmlns:xsi=”http://www.
xmlns:soap=”http://schemas.xml- w3.org/2001/XMLSchema-in-
soap.org/soap/envelope/”> stance” xmlns:xsd=”http://
<soap:Body> www.w3.org/2001/XMLSchema”
<Health_Service xmlns=”http:// xmlns:soap=”http://schemas.xml-
tempuri.org/Primary_Core_Ser- soap.org/soap/envelope/”>
vice/Service1”> <soap:Body>
<Health> <Health_ServiceResponse
<Health_Id>int</Health_Id> xmlns=”http://tempuri.org/Pri-
<Person_Id>string</Person_ mary_Core_Service/Service1”>
Id> <Health_ServiceResult>
<Person_Id_Type>string</Per- <Health_Id>int</Health_Id>
son_Id_Type> <Person_Id>string</Person_
<Persons_ Id>
Temperature>double</Persons_Tem- <Person_Id_Type>string</Per-
perature> son_Id_Type>
<Persons_BP_ <Persons_
Systollic>double</Persons_BP_ Temperature>double</Persons_Tem-
Systollic> perature>
<Persons_BP_ <Persons_BP_
Disstollic>double</Persons_BP_ Systollic>double</Persons_BP_
Disstollic> Systollic>
<Eyes>string</Eyes> <Persons_BP_
<Persons_Heart_Rate>int</ Disstollic>double</Persons_BP_
Persons_Heart_Rate> Disstollic>
<Recommendations>string</ <Eyes>string</Eyes>
Recommendations> <Persons_Heart_Rate>int</
<Treatment>string</Treat- Persons_Heart_Rate>
ment> <Recommendations>string</
<Comments>string</Comments> Recommendations>
<MedicinePrescribed>string</ <Treatment>string</Treat-
MedicinePrescribed> ment>
<Disease_Id>int</Disease_Id> <Comments>string</Comments>
</Health> <MedicinePrescribed>string</
</Health_Service> MedicinePrescribed>
</soap:Body> <Disease_Id>int</Disease_Id>
</soap:Envelope> </Health_ServiceResult>
HTTP/1.1 200 OK </Health_ServiceResponse>
Content-Type: text/xml; </soap:Body>
charset=utf-8 </soap:Envelope>
225
Composite Person_Health Service </Person>

POST /Primary_Core_Service/Ser- <Health>
vice1.asmx HTTP/1.1 <Health_Id>int</Health_Id>
Host: localhost <Person_Id>string</Person_
Content-Type: text/xml; Id>
charset=utf-8 <Person_Id_Type>string</
Content-Length: length Person_Id_Type>
SOAPAction: “http://tempuri.org/ <Persons_
Primary_Core_Service/Service1/ Temperature>double</Persons_Tem-
PersonHealth_Service” perature>
<?xml version=”1.0” <Persons_BP_
encoding=”utf-8”?> Systollic>double</Persons_BP_
<soap:Envelope Systollic>
xmlns:xsi=”http://www. <Persons_BP_
w3.org/2001/XMLSchema-in- Disstollic>double</Persons_BP_
stance” xmlns:xsd=”http:// Disstollic>
www.w3.org/2001/XMLSchema” <Eyes>string</Eyes>
xmlns:soap=”http://schemas.xml- <Persons_Heart_Rate>int</
soap.org/soap/envelope/”> Persons_Heart_Rate>
<soap:Body> <Recommendations>string</
<PersonHealth_Service Recommendations>
xmlns=”http://tempuri.org/Pri- <Treatment>string</Treat-
mary_Core_Service/Service1”> ment>
<PersonHealth> <Comments>string</Comments>
<Person>
<Person_ID>string</Person_ <MedicinePrescribed>string</Med-
ID> icinePrescribed>
<Person_Id_Type>string</ <Disease_Id>int</Disease_
Person_Id_Type> Id>
<First_Name>string</First_ </Health>
Name> </PersonHealth>
<Middle_Name>string</Mid- </PersonHealth_Service>
dle_Name> </soap:Body>
<Last_Name>string</Last_ </soap:Envelope>
Name> HTTP/1.1 200 OK
<Name_Suffix>string</Name_ Content-Type: text/xml;
Suffix> charset=utf-8
<Date_Of_Birth>string</ Content-Length: length
Date_Of_Birth> <?xml version=”1.0”
<Persons_Sex>string</Per- encoding=”utf-8”?>
sons_Sex> <soap:Envelope
<Living_Status>string</Liv- xmlns:xsi=”http://www.
ing_Status> w3.org/2001/XMLSchema-in-
<Phone>string</Phone> stance” xmlns:xsd=”http://
226
www.w3.org/2001/XMLSchema” (health, personhealth), and the basic data centric

xmlns:soap=”http://schemas.xml- services are created.
soap.org/soap/envelope/”> The Web service is implemented on a single
<soap:Body> or multiple servers capable of operating as a Web
<PersonHealth_ServiceResponse server for IIS or Linux. We define this as a local or
xmlns=”http://tempuri.org/Pri- single instance installation. Based on assumptions
mary_Core_Service/Service1”> with regards to limited Internet and intranet ca-
<PersonHealth_ pabilities, local instance installation is necessary.
ServiceResult>string</Person- If mobile units are used as mechanisms for data
Health_ServiceResult> collection or multiple servers are setup at different
</PersonHealth_ServiceRe- data collection points then data reconciliation can
sponse> take place at the end of the day. The Web services
</soap:Body> approach is crucial because of the ability to easily
</soap:Envelope> implement and extend the Web services across
the Internet or intranet once landline and wireless
services become readily available and reliable.
deFInItIon oF soA WItH data collection

Web servIces
In our approach, data collection is crucial. Based
SOA is a design model with a deeply rooted on our assumptions, local assets have been lost or
concept of encapsulation application logic with destroyed and city residents have been dispersed
services that interact via a common communica- or congregated into localized specified areas (i.e.,
tions protocol. When Web services are used to convention center, stadium, etc.). On the surface,
establish this communications framework, they data collection from disaster survivors can appear
basically represent a Web based implementation to be a daunting task as survivors may initially
of an SOA (Krafizig et al., 2004). feel hysteria, shock, and confusion. However,
We can begin to visualize the development SDDM technology has observed and identified
and basis of a service-oriented architecture for three stages during which primary data collection
our enterprise. Our services can be put into action can and should be accomplished.
during the response phase of the disaster. This The three stages identified for data collection
may appear implausible based on our assump- are food relief, medical relief, and post evacuation.
tions; however, in the next section we discuss A structured and orderly method of dispensing
the implementation and data collection process. food and medical supplies to survivors will allow
designated staff to capture primary data about
disaster survivors using a local installation. For
ImPlementAtIon And brevity and clarity, we outline a data collection
dAtA collectIon process based on two of the three stages.
Implementation
In our approach, a single Web service is created,

the primary data entities (persons, address, per-
sonaddress), one essential agency health services
227
Proposed structured data For local installation, we do not need a services

collection Process repository (Krafizig et al., 2004), however, for
the extension of the services to the enterprise a
In order to maximize data collection and minimize services repository will be needed for identifying
staff efforts SDDM technology proposes the fol- the location of services and identifying their func-
lowing structured approach. tions. The service bus for this approach may rely
on the technology implementation, for instance
Food and Medical Relief enterprise Java Bean, .NET, or IBM MQseries. We
make these statements to show that our approach
• Survivor enters food dispensing station. adheres to the definition of a services-oriented
• Survivor provides primary data (see pri- architecture given by Krafiz (2004).
mary data) to staff.
• Staff enters primary data into data collec- definition of service-
tion device. oriented Architecture
• Staff saves data.
• (Optional) staff issues identification card A service-oriented architecture (SOA) is a software
with photo (data is barcode scanned). architecture that is based on the key concepts of
• Staff instructs survivor to food dispensing an application front-end, service, service reposi-
station. tory, and service bus.
• Survivor obtains care package. In fact, Krafzig et al. identifies three expansion
• Staff walks survivor to medical treatment stages of SOA: fundamental SOA, network SOA,
station. and process-enabled SOA. Based on our assump-
• Medical staff retrieves primary data from tions, we take full advantage of the characteristics
server. of a Fundamental SOA (Krafizig et al., 2004)
• (Optional) medical swipes id thus retriev- identified by Krafzig, and imply the extension
ing data. to a full SOA at later stages during the disasters
• Medical staff collects and records basic vi- recovery period and beyond.
tal signs.
• Medical staff provides basic medical care
(if needed). cHArActerIstIcs oF A
• Survivor exits process. FundAmentAl soA
servIce-orIented • A fundamental SOA consists of two layers:

ArcHItecture the basic layer and the enterprise layer.
• Enables two or more applications to share
In our approach, we have not explicitly defined business logic and live data.
the application front-end because the application • Provides a strong platform for large enter-
front-end can be a Web application utilizing the prise application landscapes.
local installation of the Web server or it can be
a windows application and still utilize the Web
services on the local server. The only requirement Enables the enterprise to start small on the technical
with regards for the application front-end is that side and focus on other critical success factors.
it must be user interactive for data collection
purposes.
228
Web servIce trAnsActIons on local data stores and Web servers will decrease
(locAlly) significantly. A more dispersed and hopefully more
robust enterprise will develop while continuing
The ability to apply all or nothing data commit to build on our basic services for the primary
process is paramount in our SOA both on a local agencies. In short, a distributed government and
and wide area network (WAN) scenario. This information technology environment will evolve
all or nothing data commit process is called a that will require distributed transactions process
transaction. The ability to model and implement and distributed transaction systems (DTS). Al-
transactions is a frustrating and difficult task with though there are several distributed transactions
regards to services. According to Newcomer, the system specifications and models to choose from:
difficulty lies in the loosely coupled interfaces of (Figure 1)
services (Newcomer & Lomow, 2004).
On a local level, the two-phase commit transac- • WS-composite application framework.
tion process is ideal as the services are relatively • WS-transactions
close to each other (Newcomer et al., 2004) (they
exist on the same server) and a single uniform data The WS—transaction specification developed
repository is used for the initial data collection. by BEA, IBM, and Microsoft are recommended
Data collection can take place on local servers for our approach. The WS-transaction specifica-
during the response and certain parts of the recov- tion, based on a WS-coordination (WS-C), has the
ery phases and data reconciliation, replication or ability to utilize the two-phase commit process
integration accomplished at specified time frames defined for our local data collection process. In
throughout theses two phases. However, with the fact, the WS-transaction is composed of a family
expansion of our SOA during the recovery and of specifications (Newcomer et al., 2004)
rebuilding phase service transactions will play
an important role. • WS-AtomicTransactions (WS-AT): A
two-phase commit protocol for Web ser-
soA expansion and Web service vices interoperability.
transactions (distributed) • WS-BusinessActivity (WS-BA): An open
nested protocol for long running business
As resources and services expand, the fundamental processes.
SOA will expand to include other services, busi-
ness process and data repositories and the reliance
Figure 1. Fundamental SOA Representation
229
• WS-Coordination (WS-C): A pluggable Smart card technology will facilitate data collec-
coordination framework supporting WS- tion and data integrity across the municipality. A
AT and WS-BA. brief description of this technology is presented
for clarity. (Figure 2)
For a more detailed explanation of the imple-
mentation of the WS-transaction, we defer to Integrating smart card and smart
Newcomers et al.’s Understanding SOA with Web card technology
services (2004).
Once the primary data has been collected, the
soA expansion and other survivor data can be recorded onto a smart card
Information systems (SC). Smart card technology has been around since
1968 when German inventors Jurgen Dethloff
The fundamental SOA by design is expandable to and Helmut Grotrupp applied for the first ICC
be inclusive of other services, business processes, related patents with similar applications in Ja-
and data repositories while utilizing the services pan in 1970 and France 1974 (Litronic, 2003) In
developed for the response and recovery phase. 1984, the French Postal and Telecommunications
Over the course of the recovery and rebuilding began using the smart card for telephones and
phases, independent information systems will be by 1996, 150 million smart cards were in use for
developed by vendors. These independent systems this purpose. Over the past 20 years, smart cards
will be integrated into the enterprise (SOA) using
Web service integration tools, XML, and XML Figure 2. Initial high-level Web services infra-
translation. structure
The creation of our enterprise using the ser-
vice-oriented architecture will allow a smoother
integration transition due to the inherent nature
of Web services and XML. Various services,
interfaces, data wrappers, application wrappers,
and application programming interfaces (API’s)
can be developed to extract and utilize data and
business processes within the framework.
Future trends And reseArcH
SDDM technology is in the process of research-

ing Web service security, citizen privacy issues,
and the implementation of smart card technol-
ogy. The maturity of the Web and Web services
over the past three years have produced several
security specifications worth researching and
implementing to provide for tighter and stronger
security protocols. In addition, citizen privacy
issues will play a major role with the introduction
and implementation of smart card technology.
230
have been used throughout Europe successfully encryption algorithms to make the application
for a variety of purposes. They range from stor- software and data unreadable.
ing bank account information, to storing health Although smart cards remain relatively the
care and health related information, as well as same, they do not perform the same functions.
transportation information (Jacquot, 2003). In the In fact, the function of the smart card is based on
United States, smart cards are beginning to garner two factors: the chip type and the interface used
interest since the events of 9/11 and the increase to communicate with the card reader. In short,
in identity theft. Current U.S. applications for there are three chip types: memory only, wired
the smart card are identity security authentica- logic, and microcontroller.
tion, personal identification, and transportation
services. For thoroughness, a brief description of • Memory-only integrated circuit chip
smart cards and smart card technology is provided. cards (including serial protected mem-
The majority of the background information was ory chip cards): Memory-only cards are
taken directly from the U.S. Government Smart “electronic magnetic stripes” and provide
Card Handbook (GSA Card Handbook, 2004). little more security than a magnetic stripe
card. Two advantages they have over mag-
What is a smart card? netic stripe cards are: (a) they have a high-
er data capacity (up to 16 kilobits (Kbits)
A smart card is a simple plastic card that resembles compared with 80 bytes per track), and (b)
a credit card in size. It is considered a smart card the read/write device is much less expen-
because a small microprocessor and memory is sive. The memory-only chip cards do not
embedded inside the card. A smart is card is used contain logic or perform calculations; they
for storing. In addition to storing data a smart card simply store data. Serial-protected mem-
can also store applications and perform mostly ory chip cards have a security feature not
basic and a few complex calculations. The U.S. found in the memory-only chip card; they
Government Smart Card Handbook provides an can contain a hardwired memory that can-
excellent description of a smart card and we will not be overwritten.
use that description as our basis for the definition • Wired logic integrated circuit chip
of a smart card.Definition: Smart Card: A smart cards: A wired logic chip card contains
card is a credit card-sized device that contains one a logic-based state machine that provides
or more integrated circuits (ICs) and also may encryption and authenticated access to
employ one or more of the following machine- the memory and its contents. Wired logic
readable technologies: magnetic stripe, bar code cards provide a static file system support-
(linear or two-dimensional), contactless radio ing multiple applications, with optional en-
frequency transmitters, biometric information, crypted access to memory contents. Their
encryption and authentication, or photo identifica- file systems and command set can only be
tion. The integrated circuit chip (ICC) embedded changed by redesigning the logic of the IC
in the smart card can act as a microcontroller or (integrated circuit).
computer. Data are stored in the chip’s memory • Secure microcontroller integrated cir-
and can be accessed to complete various process- cuit chip cards: Microcontroller cards
ing applications. The memory also contains the contain a microcontroller, an operating
microcontroller chip operating system (COS), system, and read/write memory that can
communications software, and can also contain be updated many times. The secure micro-
controller chip card contains and executes
231
logic and calculations and stores data in The memory only card could be used for both
accordance with its operating system. The human and non-human asset data capture dur-
microcontroller card is like a miniature ing the response phase; however, since we have
PC one can carry in a wallet. All it needs developed a framework for the future enterprise
to operate is power and a communication and we view the survivor data as dynamic and
terminal. consistently changing due to the addition of
agencies and services. Thus, we recommend the
smart card Interfaces contact multi-application cards for issuance to
the survivors. A multi-application card has the
There are two primary types of chip card in- advantage of having multiple applications on a
terfaces—contact and contactless. The terms single card and each application can be managed
“contact” and “contactless” describe the means by a different agency. Thus, when the department
by which electrical power is supplied to the in- of motor vehicles comes online with services the
tegrated circuit chip (ICC) and by which data is citizen can simply show the card to motor vehicle
transferred from the ICC to an interface (or card staff (MVS), MVS will read the primary data off
acceptance) device (reader). Cards may offer both of the card (name, dob, address, etc….), initiate
contact and contactless interfaces by using two the immediate process, update the DMV entity(s)
separate chips (sometimes called hybrid cards) or utilizing data centric and logic centric services
by using a dual-interface chip (sometimes called and finally update the smart card with DMV
“combi” cards). data. Thus we have shown how two separate and
distinct agencies (DOH, DMV) with different
contact smart card functions and data requirements are linked by a
survivors primary data and how information can
• Contact smart cards: A contact smart be added, updated, retrieved and shared using a
card requires insertion into a smart card smart card. A high-level overview is provided in
reader with a direct connection to a con- the following diagram.
ductive micromodule on the surface of the With regard to public safety assets, although the
card. maintenance of data will change over time, it will
• Contactless smart card. remain static for most of the lifetime of the asset.
Integrating smart cards during the response phase
will produce immediate returns for emergency
If the smart card does not have a contact pad then workers and government. For instance, the abil-
the connection between the reader and the card is ity to determine and track food rationing (based
done via radio frequency (RF). Contactless smart on food dispensation), the ability to immediately
cards can exchange data without having to make view a survivors vital signs, and possible treat-
contact with the reader using radio frequency (RF). ments and the ability to identify public and public
Hence, contactless smart card communication safety assets (i.e., street lights, catch basins, down
can take within a perimeter of 10 centimeters or power lines, roadways, alleys, etc).
3.94 inches.
smart card Authentication/
combo smart card Authorization
A combo or hybrid card can have both contact and A natural question that arises with the utilization
contactless communication with the card reader. or implementation of smart card technology with
232
our approach is the issue of authentication and It is recommended that the Unique ID (SSN,
authorization. To be fair, authentication has re- drivers license number, green card number, etc.)
ceived the most interest. We define authentication defined in our high level data model be used as
and authorization utilizing Wikipedia dictionary the PIN. Thus, the authentication level provides
(Wikipedia Dictionary): a uniform consistency of the citizen and the de-
fined levels of authorization would apply to the
• Authenticaion is the act of establishing particular (application) service provided (DMV,
or confirming something (or someone) as emergency medical services, etc.).
authentic, that is, that claims made by or
about the thing is true. Authenticaion of
an object may mean confirming it prov- conclusIon
enance. Authentication of a person often
consists of verifying their identity. This chapter strongly emphasizes an implementa-
• Authorization is the process of verifying tion approach that rests on the Zachman system
that a known person has the authority to model perspective. This approach was formulated
perform a certain operation. after the visual confusion of the aftermath of
Hurricane Katrina and the lack of preparedness
With regards to authentication, there exists thereof. It is important to note in the conclusion
three basic types or categories which are: that this approach be considered in the prepared-
ness phase as to circumvent or assist in answering
• Something the user is (e.g., fingerprint or organizational and logistical challenges at the lo-
retinal pattern, DNA sequence (there are cal, state, and federal government level. Natural
assorted definitions of what is sufficient), questions will arise, for instance:
voice pattern (again several definitions),
signature recognition, or other biometric • Where will the back bone of the enterprise
identifier. be located?
• Something the user has (e.g., ID card, secu- • Which agencies and business processes are
rity token, software token, or cell phone). next to come online?
• Something the user knows (e.g., a pass- • Who will be in control of the fundamental
word, a pass phrase, or a personal identifi- enterprise (SOA)?
cation number (PIN)).
The author hopes that viewing this approach
The three authentication categories provide as part of the disaster preparedness phase will
a basis for single factor or multiple factor au- answer many of the organizational and logistical
thentication. In single factor authentication only challenges that face government during and after
one of the categories is used for authentication the disaster.
purposes. If more than one category used for We have attempted to establish a basic
authentication then the authentication process is framework for the development of an enterprise
called: multiple-factored authentication. In our architecture for local government immediately
approach we suggest a two factor authentication after a natural disaster or major terrorist attack.
process using: As a basis for our development, we have applied
Zachman’s system model perspective and defined
• Something the user has (smart card). two basic types of data elements: primary data
• Something the user knows (PIN). elements and secondary data elements. These
233
elements were then wrapped into simple services GSA. U.G.S.A., (2004). Government Smart Card
and utilized by composite services to establish Handbook.
a Web services environment on a local server
Guruge, A. (2004). Web services theory and
level. The ability to extend the local server level
practice. Burlington, MA: Elsevier Digital Press.
to a broader audience among local government
through the Internet or intranet is obvious. In the Jacquot, J. N. (2003). Application Note 104: Smart
future SDDM technology will continue to refine Cards. Exfo Photonic White Paper.
the concepts and explore the areas of process and
Krafzig, D., Banke, K., & Slama, D. (2004).
functionality in more detail. The expectation is
Enterprise SOA: Service-oriented architecture
that this approach will be deemed practical and
best practices (The Coad Series). Upper Saddle
useful as a model for assisting local government
River, NJ: Prentice Hall.
to gain valuable information from survivors,
assist survivors in medical need, and eventually Litronic, S. (2003). Introduction to smart cards:
provide the basis for a an enterprise utilizing a A White Paper. Retrieved September 2005, from
service-oriented architecture in the rebuilding of www.litronic.com
local government.
Newcomer, E. (2002). Understanding Web servic-
es XML, WSDL, SOAP, and UDDI. Indianapolis,
IN: Addison Wesley.
reFerences
Newcomer, E., & Lomow, G. (2004). Under-
Davis, T. (1998). The role of government in a di- standing SOA with Web services. Indianapolis,
saster. The Disaster Handbook 1998 National Edi- IN: Addison Wesley.
tion Institute of Food and Agricultural Sciences.
Chapter 3, Section 3.7: University of Florida. Rob, P., & Coronel, C. (2002). Database systems:
Design, implementation, & management. Boston:
Erl, T. (2004). Service-oriented architecture: a Course Technology.
field guide to integrating XML and Web services.
Upper Saddle River, NJ: Prentice Hall. Wikipedia, Defining Authentication and Au-
thorization, from http://en.wikipedia.org/wiki/
Fermantle, P. (2002). Enterprise services: Examin- authentication
ing the emerging field of Web services and how it is
integrated into existing enterprise infrastructures. Zachman, J. A. (1987). A framework for informa-
Communications of the ACM, 45(10). tion systems architecture. IBM Systems Journal,
26(3).
This work was previously published in Services and Business Computing Solutions with XML: Applications for Quality Man-
agement and Best Processes, edited by Patrick Hung, pp. 269-287, copyright 2009 by Business Science Reference (an imprint
of IGI Global).
234
235
Chapter 1.17
An Overview of Executive
Information Systems
Gary P. Moynihan
The University of Alabama, USA
AbstrAct able to better understand how EIS has adapted to

meet the requirements of senior management in an
An executive information system (EIS) is a software enterprise system environment.
system designed to support the informational needs
of senior management. The EIS is characterized
by an easy to use and maintainable graphical user IntroductIon
interface; integrated capabilities for data access,
analysis, and control; analysis and report generation Executives are upper-level managers who exert a
across multiple files; and on-request “drill down” strong influence on the direction and activities of
capability. Most existing management information the entire organization (e.g. McLeod and Schell,
systems provide an enormous quantity of detailed 2001). An executive information system (EIS) is
status reports. However, they lack the capability of a computer-based system designed to support the
providing summarized levels of information, in an unique informational needs of these very senior
appropriate format, for higher levels of management. managers. These systems are designed to simplify
This problem has continued despite the emergence the user’s interface with the computer, facilitate
of enterprise resource planning systems. By under- retrieval and manipulation of data from different
standing the concept and functionality of traditional sources, and display results in a single presentation.
executive information systems, readers will also be Finally, executive information systems provide the
capability to highlight exceptions, and explore data
at progressively lower levels of detail.
DOI: 10.4018/978-1-59904-859-8.ch006
An Overview of Executive Information Systems
The concept of the executive information emphasized. For example, based on Rockart and
system can be traced to 1982, when Rockart and DeLong’s (1988) original discussion, Turban and
Treacy introduced this term to describe an emerg- Aronson (2001) further define an executive sup-
ing category of information systems. They noted port system as “a comprehensive support system
that there are four critical components to the EIS that goes beyond EIS to include communication,
concept. First, there is the Executive, the human office automation, analysis support, and business
element in the EIS environment. The executive re- intelligence” (p. 308). An important feature, es-
quires timely information, quick inquiry response, pecially to upper-level management, is access to
and systemic ease of use. Information supports external data. Thus, where core EIS functionality
both critical success factor analysis and leading focuses on the processing and presentation of
indicators of potential problems. The third com- information, an ESS implies other communica-
ponent is the system architecture that encompasses tions support capabilities more oriented toward
linkages to the relevant, and the processing of this the second fundamental EIS approach mentioned
data into critical information. The final component above. These communications support capabilities
is the organizational structure that manages the may include e-mail, office automation functions
databases and systems, and maintains information like electronic calendars, and linkages to stock
security. All four components must be integrated market news and industry trends.
for the EIS to be successful. Further, EISs were initially developed to sup-
To do this, an EIS combines two complimen- port a small set of high-level executives within an
tary approaches. According to McNurlin and organization. These systems normally served up to
Sprague (2002), “at its heart, an EIS should filter, 10 or 15 senior executives. The success of these
extract, and compress a broad range of up-to-date initial implementations has led to an expansion of
internal and external information. It should call these systems to support mid-level managers with
attention to variances from plan and also monitor cross-functional information needs. This type of
and highlight critical success factors of the indi- system served up to sixty executives and manag-
vidual executive user” (p. 386). This perspective ers, and has been termed an “extended EIS” In
defines an EIS as a structured reporting system to some cases, EIS, have been adapted to the entire
meet the unique needs of executive management. organization (McLeod and Schell, 2001). Rockart
The second fundamental approach of the EIS, as and DeLong (1988) first observed this migration.
identified by McNurlin and Sprague (2002), is as For this paper, the term “executive information
a human communications support, such that. “the systems” will be used across this spectrum of ap-
managers make requests, give instructions, and ask plications and levels of use, with consideration of
questions to selected members of this network to how the EIS concept has evolved in the process.
get people going on the desired action” (p. 387).
Alter (2002) confirms this by emphasizing the
networks of internal and external contacts used bAckGround
to gather information about specific issues of
current importance, as opposed to only utilizing Historically, the traditional management informa-
formalized information systems. tion systems (MIS) within a company have been
The terms executive information system and developed independently of each other. These
executive support system (ESS) are often used systems typically supported marketing, manufac-
interchangeably (e.g. Laudon and Laudon, 1998). turing, distribution, personnel, accounting, order
Other authors make a distinction between the two, entry, inventory control, and purchasing. They
based upon the fundamental approach used or are used to support organizational planning and
236
control, and are capable of producing both sum- data accessed from the appropriate data source.
mary and very detailed reports. Senior managers The presentation manager provides the formatted
have typically tried to utilize these MIS reports executive information in the form of reports to
to identify exceptions and trends that require the user, as well as a means of system access and
management attention. However, in attempting to inquiry. Thus, they only addressed the earliest
assimilate outputs beyond the limited functional phases of decision-making (i.e. isolating problems
areas, MIS systems are frequently deficient in sup- and raising questions). The information output
porting the company-wide perspective required by was considered to be restrictive and inflexible.
executive decision-makers. Although these MIS Review of the literature has noted consider-
systems may be subsequently interfaced, there able work on the development of decision rules
was no true integration. There is a wide variety and computer-based support systems to aid in
of screens and reports provided. Each system entire decision-making process. Decision support
focuses on individual areas of concern, and each systems (DSS) are software systems that utilize
may have its own format under separate security sophisticated algorithmic approaches to solve
access. Timliness of the resulting information problems. Within the DSS, the data processor
may vary based on the specific system execution (i.e. the database management system) controls
and the data that it uses. These reports frequently the access and update of data from external files
take the form of an endless stream of computer and datasources. The modelbase contains the spe-
print-outs or online reports, which the executive cific analytical methods used for processing the
has little or no time to review. There is a limited accessed data. Although decision support systems
capability of providing summarized levels of were originally intended to serve the needs of up-
information, in an appropriate format, for higher per management, they became to tools of analysts
levels of management. Even when the reams of and lower-level managers. A further evolution was
data are condensed and summarized, relevant needed to better meet executive needs. As noted
information is frequently difficult to extract or in Figure 1, the integration of DSS capabilities
understand. Too often decisions were a reaction within the MRS framework led to the emergence
to an event that has already occurred. of the true executive information system.
Briefing books, sometimes referred to as man-
agement reporting systems (MRS), were an early
Figure 1. Executive information system archi-
effort to address these problems by automating
tecture
data acquisition from a wide variety of corporate
systems and databases, and providing online output
in the form of fixed-format reports (e.g. Laudon
and Laudon, 1998). The purpose of a MRS was
to integrate, select, and reduce raw data to valu-
able information; create and distribute finished
quality reports and charts; and to ensure reporting
of consistent information across the corporation.
Management reporting systems tended to be MIS-
based transaction systems linked to a presentation
manager interface. These levels represent the ma-
jor functional components of the MRS, where the
transaction system tracks the status of operations
and reports performance relative to plan based on
237
Insertion of the DSS level into this framework manufacturing companies also shared these same
provides a means whereby users can better de- requirements. Perhaps the most important aspect
velop their plans, specify what activities should is that an EIS must meet the information needs
take place, and better organize the operational of the executive user. Determining precisely what
information for executive inquiry. For example, information should be included in the system
Turban and Aronson (2002) describe a pharma- continues to be a major topic of research in this
ceutical company’s product managers utilizing area (e.g. Rainer and Watson, 1995; Walstrom
an EIS-based DSS module to predict the end-of- and Wilson, 1997).
month order status. Similarly, the commercially- The literature particularly notes executives’
available DecisionWeb EIS software provides reliance on critical success factors (CSFs). Criti-
ad-hoc analysis for determining trends and vari- cal success factors consist of a relatively small
ances. As noted in Figure 1, this DSS integration number of easily identified operational goals that
may be executed either by using the presentation are believed to ensure the success of an organiza-
manager as a selection interface, or automatic tion. These CSFs may be derived by the industry,
access from the source database(s). The results the specific firm, the manager, and the broader
of the DSS analysis are then formatted through business environment. For example, McLeod and
the presentation manager for display to the user. Schell (2001) identify “styling, an efficient dealer
Although, DSS and EIS capabilities were initially network, and tight control of manufacturing cost”
considered to be separate and distinct, as reported as critical success factors within the automobile
in the literature, Alter (2002) noted an accelerating industry (P. 332). Rockart and Treacy (1982)
trend during the 1990s towards their integration. argued that the information requirements of an or-
This is frequently expressed through the use of a ganization, and particularly of that organization’s
spreadsheet facility. executives, are shaped by these critical success
factors. The CSFs then determine a smaller data
executive requirements set on which resulting information systems can
focus to best meet the executives’ requirements.
Considerable research has been conducted on the The EIS is subsequently utilized to monitor
informational needs and decision making pro- each of the CSFs according to five categories of
cesses used by executives. Watson et al. (1993) information (Turban and Aronson, 2001). Key
provided a comprehensive survey of the available problem narratives are reports that highlight
research at that time. Executives’ busy schedules overall performance, major problems, and the
preclude either long training sessions or long un- potential underlying reasons for these problems.
interrupted time for system use. Systems must be Associated narrative explanations are often linked
intuitive and easy to use, with very quick response. to supporting tables and graphs. Highlight charts
They must allow “environmental scanning” by display summary information based on the user’s
accessing and integrates a broad range of inter- indicated perspective and judgment. Since they
nal and external data, and support management are personalized to the user’s concerns, they visu-
by exception by highlighting potential problems ally highlight the performance status of priority
and tracking critical data. The EIS should store areas versus respective CSFs. Key factors provide
aggregated data for rapid access, efficiently index specific measures of CSFs, referred to as key per-
and retrieve information, and sort screen data ac- formance indicators (KPIs). These corporate-level
cording to user defined criteria. Although earlier displays usually employ management by excep-
research focused on large corporations, Walters tion to investigate measures of CSFs identified as
et al. (2003) confirmed that executives in smaller problems, or potential problems, on the highlight
238
chart. Typical key performance indicators include pictorial menus that can be learned intuitively,
profitability measures, financial ratios, turnover with variances and exceptions highlighted in
rate, sales growth, market share, and demographic color” (p. 611). Users frequently access data and
data. Detailed KPI responsibility reports indicate make selections via touch screen, keyboard, or
the detailed performance of or business areas criti- mouse. Navigation to deeper levels of detail may
cal to the success of the organization. be conducted based on their own selections, or by
following pre-defined paths. The screen displays
Functionality frequently incorporate management by exception
by comparing the planned (or budgeted) perfor-
Consistent with these defined executive user mance with the actual performance of an operation.
requirements, an EIS promotes a proactive an- Accordingly, the executive’s attention is focused
ticipatory style of management by which problem on exceptions to standards. Significant exceptions
indicators are closely monitored based on their (i.e. those beyond a pre-identified threshold) are
CSFs, and timely corrective action implemented automatically identified by the EIS and highlighted
prior to serious problem occurrence. This type (or otherwise brought to the executive user’s atten-
of system may be further defined by having the tion. This saves significant time in sifting through
following characteristics: the data for exception conditions.
This last capability allows the pointing and
1. An easy to use and maintainable graphical clicking on a specific data field for which the
user interface. user desires an additional level of detail. As a
2. Integrated capabilities for data access, analy- result, the components of that data field are then
sis, and control. displayed. EIS systems have a top-level menu,
3. Analysis and report generation across mul- often referred to as an “executive dashboard”
tiple data sources. or “executive scorecard” (Liang and Miranda,
4. On-request “drill down” capability to lower 2001). The intent of the dashboard is to provide
levels of detail. a single-screen display of relevant and critical
5. Depiction of organizational critical success business metrics and/or statuses, frequently fea-
factors. turing stoplight indicators. It is configured with a
6. Functionality for decision support, ad-hoc “hotspot” button at the intersection of the report-
queries and what-if analysis. ing categories and the organization identifiers,
7. Sophisticated tools for system navigation. indicating the appropriate status. (See Figure 2.)
8. Data analysis features. Red, yellow, and green lights are indicated based
9. Advanced report generation. on comparison to preset thresholds, where green
10. Access to a variety of external data sources. indicates a nominal condition, yellow indicates a
warning, and red identifies performance outside of
The resulting combination of capabilities the pre-established boundaries. (Some EIS systems
provides a powerful tool for senior management. also provide audio signals to alert the executive
They can navigate through data hierarchies, view to arriving information.) White indicates that no
data from different dimensions, analyze it, and information is currently available. The colors
create reports tailored to their specific informa- thus allow the user to prioritize his actions, and
tional needs. address potential problems requiring immediate
As noted by Laudon and Laudon (1998), most attention. This enables more efficient and effec-
EIS systems provide “colorful presentations and tive decision-making.
239
Figure 2. Example top-level selection matrix
Clicking on any of these stoplights allows drill are able to track data across multiple months.
down to the desired level of reporting, in order to Drill-down capability exists for visibility to even
indicate the source of any problems. (See Figure further levels of detail.
3.) This concept allows users at various manage- An EIS normally includes tools for modeling
rial to see data details relevant to their position and analysis within its DSS layer. These frequently
and viewing responsibility. For example, the include built-in functions for statistical (e.g.
dashboard may provide a set of KPIs (e.g. revenue trending, regression, and Monte Carlo analysis)
forecasts, gross profit, or inventory levels) with and financial models (such as profitability analy-
drill-down capabilities for each KPI monitored sis, marginal utility, and benefit-cost analysis).
(Bose, 2006).Users at lower levels can view Other functions may include goal-seeking (i.e.
data details specific to their individual business specify the values for the output data and the
function, while users at higher levels can view system will determine the required value of the
summaries across multiple functions. All levels input variable) and forecasting models. This may
Figure 3. Sample subordinate display
240
be deployed through the use Excel or similar to their position and viewing responsibility. Users
spreadsheet program linkages in order to create at the lowest levels view data details specific to
graphic comparisons (e.g. by time period; by their functional area, while users at higher levels
product: by region; or by price range). Costlier initially view summaries across subordinate or-
systems include more sophisticated specialty ganizations, but can drill-down for visibility to
analysis built-in programs as part of their on-line further levels of detail. There are several strategies
analytical processing (OLAP) capabilities. OLAP in addressing this problem. Some organizations
supports decision-making based on analysis of adopt a multi-level security system utilizing the
multi-dimensional aggregated (i.e. summary) capabilities of an EIS application development
data. Examples of these include Pilot Software’s software. An alternate approach utilizes a multi-
Decision Support Suite, Informix’s MetaCube level, centralized database system to support data
Product Suite, and Cognos’ PowerPlay. To aid in having different classifications and users having
the decision-making process, interactive support different clearances.
functions are normally available which allow
the user to initiate ad hoc queries and “what-if” development Approach
analysis. What-if analysis is a particularly useful
tool for managers, whereby they can project the The EIS can be coded directly using a conventional
effect of modifications to the input data or other programming language, analogous to any deci-
parameters will have on a resulting decision. sion support system. Accepted DSS development
Construction of an EIS implies the need for a methodology identifies four primary phases: data
security system that will control access to sensitive acquisition, system design, system construction,
information. The EIS, its data and software, must and verification and validation. During the data
be protected at all times from unauthorized access. acquisition phase, the key concepts, relationships
Accepted backup and recovery procedures need to and sources are identified. One major purpose of
be incorporated. Additional security features also the general system design phase is to establish
need to be incorporated that will restrict access to the overall definitions and descriptions of the
data and information outside a specific user’s or- proposed system. This is represented through the
ganization and/or subordinate offices. Multi-level development of a requirements definition (RD)
security for data access needs to be established document. Laudon and Laudon (1998) provide
for any EIS that is used by multiple executives. a series of recommended steps for determining
In addition to prohibiting access to the system by EIS requirements:
unauthorized personnel, the security system must
be multi-level in nature, to restrict access within 1. Identify a set of issue-generating critical
a variety of classifications (Moynihan, 1993b). events.
The security system should also be as transparent 2. Obtain, from the executive, his assessment
as possible to the executive user. Multiple levels of the impact of the critical events on his
of user-input passwords should be avoided Most goals and determine a set of critical issues.
senior managers prefer “one-button” access to 3. Determine three to five indicators which can
their systems. Security supports the integrity of the be used to track each critical issue.
data by ensuring that the data cannot be tampered 4. Obtain, from the executive, a list of potential
with, viewed by outside parties, or that the data information sources for the indicators.
cannot be lost. The security feature enforces the 5. Identify exception heuristics for each
drill-down functionality such that users at various indicator.
levels of the organization see data details relevant
241
During the subsequent detailed design phase, drill-down. Utilization of a COTS diminishes
the acquired data and candidate mathematical EIS development cost due to the availability of
models is organized within the RD context. This such product-specific data manipulation tools.
may provoke further investigation into support- The literature review indicates that subsequent
ing models and data. A functional system design maintenance costs for a COT-based system would
document is then developed identifying the overall be substantially less than in-house development
system architecture, processing considerations, (e.g. Watson et al., 1993). A variety of such EIS
and definition of displays and reports. The identi- application software shells are now commercially
fied requirements and functionality can be utilized available, which allow customization to individual
for software selection. The general design is then company needs.
“fine-tuned” to map to the capabilities of the soft- Executive information systems tend to be
ware. This step precedes actual programming of unique in that the implementation (and even
the EIS system. Upon completion of the system’s post-implementation) issues often overshadow
construction, a thorough verification and valida- the actual system development. According to
tion is conducted. Traditional software engineering Rockart and Delong (1988), “the fragmented
techniques, i.e., using a set of predetermined tests, nature of executive work, the high degree of
are generally used to verify the integrity of the environmental uncertainty at this level of the
system. The validation process establishes that organization, and the political ramifications of
the system’s functionality addresses the original providing top management with more and bet-
engineering problem and the user’s needs. ter information, as well as other factors, make
Because executive’s needs may change implementing EIS a special challenge” (p. 158).
rapidly, some developers recommend utilizing While data management may not be considered to
prototyping, also referred to as rapid, application be a significant problem when an EIS prototype
development, to develop an EIS. In this approach, is first installed, the practical barriers to provid-
a series of preliminary working versions, each ing executives with the data that they need can
having progressively more functionality, are con- be a major road block post-implementation. As
structed for demonstration and evaluation. Due such, the system developers must contend with
to the executive user’s expectations, developers a number of issues relating to data availability,
must be certain that the system will work before ownership, infrastructure, integrity, integration
demonstrating it to the user. Each stage of the and management (Moynihan, 1993b). Once the
prototype must have an intuitive user interface, EIS application goes online, the data owners
i.e. one that the executive can learn very rapidly. are responsible for insuring the accuracy of the
Alternatively, a commercial off-the-shelf data. If the data are too old, too detailed, or too
(COTS) product can be utilized. While early EIS summary, or simply wrong, the executives may
systems were programmed exclusively using make the wrong decisions. Procedures need to be
internal company resources, by the mid-1980s instituted in order to insure that the appropriate
Comshare and Pilot Software introduced the data are collected on time, refreshed when nec-
first COTS development products (Watson and essary, consolidated where applicable, and made
Carte, 2000). A COTS product is not an EIS per available in the appropriate format. An underlying
se, but rather a set of software tools with which problem with traditional data management is the
to construct an EIS. They generally include the difficulty of integrating data originally designed
capability for constructing standard executive to meet the needs of isolated applications. This
information features, such as picture menus, problem of decentralized development has lead to
icons, hotspots, key performance indicators, and discrepancies in data definitions, data redundancy,
242
and synchronization problems. The resulting lack tential for the system is beneficial and should be
of accessible, quality data often limits executive encouraged. However, the introduction of major
access to information, and hinders system main- modifications too early can result in an unforeseen,
tenance (Moynihan, 1993b). yet interrelated, impact elsewhere in the EIS.
Rather than providing direct linkage to Therefore a strict policy should be established as
source data, one approach to EIS construction early as possible, regarding the type and extent of
recommends the establishment of a separate EIS changes that are permitted before implementation,
database. As noted by McNurlin and Sprague and how post-implementation modifications will
(2002), there is concern that the executive would be handled. The subsequent addition of numerous
need too much detailed knowledge to access, in- enhancements can degrade system efficiency.
terpret, and use the raw source data. Instead, the Judgment must be applied in order to balance
desired data should be extracted from these data system capabilities versus response rate. Rapid
sources, formatted and incorporated into the EIS response is considered vital to the executive user.
database. This information should be organized
into a hierarchical fashion such that highlights are
indicated in the top layer. Most raw data sources trAdItIonAl eIs APPlIcAtIons
contain mostly current data, with little historical
context. In order to be able to identify trends, A variety of executive information system appli-
the EIS database needs to incorporate time series cations have been developed for private industry.
data. Further, the executives need to track exter- By necessity, very large corporations were among
nal sources of information. An EIS that accesses the first to realize the advantages that an EIS
internet-based sources gives executives a greatly could provide. One such representative example
improved ability to assess competitive conditions. is Conoco’s EIS (Belcher and Watson, 1993).
Conversely, this access has its limitations; i.e. Senior management was interviewed regarding
internet-based sources may be of questionable ac- their specific needs, and statistics tabulated on
curacy and reliability. Those data definitions that utilization of specific software applications. The
are closely defined and controlled in an EIS, are system was designed and constructed by in-house
not controlled in these external internet sources. staff with a heavy emphasis on the use of financial
This may lead to confusion as outside informa- and refining capacity models. The completed EIS
tion is merged with the well-defined EIS database has continued to grow in usage and capability,
(McNurlin and Sprague, 2002). including 75 decision support applications which
It is also to be noted that, as executive user were used in some form by over 4,000 Conoco
personnel, become familiar with an EIS and its employees throughout the world (Belcher and
capabilities, these users inevitably begin to iden- Watson, 1993). The cited benefits of the sys-
tify new system “requirements”. Sometimes, this tem included increased productivity, improved
occurs before the system is even formally imple- decision-making, and cost savings.
mented. This user community is used to getting its Pratt and Whitney’s Commercial Engine
own way. Despite the flexibility inherent in most Business purchased Commander, a commercial
EIS systems, some of these requests may involve EIS software package from Comshare, to view
modifications well beyond those originally speci- metrics on customer service and jet engine prod-
fied in the design. Some requests are relatively uct performance (Laudon and Laudon 1998).
easy to accommodate, others are not (e.g. those The Commander software shell was customized
requiring the establishment of a new database). to track critical quality and reliability measures
The fact that users are discovering increasing po- (e.g. parts availability, spares inventories, and
243
delivery statuses) for each engine model by the were a variety of Program Office and functional
specific customer. As with many such COTS directorate systems. In March 1992, the MSFC
shells, Commander has graphics facilities, ex- Center Director appointed a steering committee
ception reporting/ variance highlighting and an and working group to begin efforts toward a
easy-to-use system navigation capability. The center-wide EIS (Moynihan, 1993a). The resulting
resulting EIS was used to provide drill down system linked existing communication networks
capabilities to determine the reasons for specific to support both Windows-based and MacIntosh-
engine repairs, and where quality improvements based workstations then used at the Center, with the
should be implemented. The system was initially capability to also support a planned UNIX-based
utilized by approximately 25 executives, but was platform. Based on an analysis of MSFC executive
planned for significantly expanded use (Laudon user requirements, the HOLOS EIS application
and Laudon, 1998). Pizza Uno, a food service shell was purchased. HOLOS provided a compre-
chain, similarly employed the Pilot Software hensive set of tools for constructing standard EIS
OLAP-based (Online Analytical Processing) features, such as picture menus, icons, hotspots,
shell to integrate its existing corporate intranets key performance indicators and drill-down. The
(Turban and Aronson, 2001). The resulting EIS software also allowed multi-dimensional mod-
was characterized by a user-friendly graphical eling with associated calculations, statistical
user interface and an ability to conduct multidi- analysis, regression, and financial- and time-based
mensional analyses across different product lines. functions (Moynihan, 1993a).
For example, marketing executives utilized the Watson and Carte (2000) completed a compre-
system to measure the results of test marketing a hensive survey on the use of EISs in the federal
new menu and how it relates to labor and other government environment. Among the organizations
costs (Turban and Aronson, 2001). they cited include the Internal Revenue Service,
A number of EIS systems have also been the Government Printing Office, Department of
applied to the government environment. One of Commerce, Department of Defense, and the Gen-
the first was an EIS developed for the New York eral Services Administration. Liang and Miranda
State Office of General Services beginning in (2001) note that non-federal public sector EIS
1988. With a budget of $500 M and over 4000 applications remain limited but are anticipated to
employees, the agency needed to consolidate and grow. Among the state and city governments they
integrate its data and information on the “design, cite with active EIS development projects are:
construction, and maintenance of state buildings, Utah, Texas, San Francisco, Albuquerque, and
food and laundry services to both correctional Philadelphia. Recently, the State of Minnesota De-
facilities and health-related institutions, statewide partment of Finance has pursued the development
vehicle management, and centralized printing” of an executive information system. The purpose
(Laudon and Laudon, 1998, p. 612). The system of the system is to permit executive users to access
provided conventional EIS capabilities of status pre-defined views of data stored on legacy state
monitoring, comparison of budgets to actuals, information systems. Conventional EIS capabili-
and financial estimates, with drill-down to view ties (e.g. graphical and tabular formats, drill-down,
details in any category. The EIS was fielded on and exception reporting) have been implemented
a set of networked PCs linked to the agency’s in conjunction with Web-based access (Liang and
mainframe computer. Miranda, 2001). Planned development would
Another early public sector system was the increase user flexibility and functionality with the
center-wide EIS for NASA’s Marshall Space ability to develop what-if analyses, cost projections,
Flight Center (MSFC). Prior to this effort, there and scenario sharing/update.
244
relevAnce to generated from a wide range of interrelated busi-

enterPrIse systems ness activities performed at different levels within
the enterprise” (p. 43). Effective management thus
The advent, and subsequent adoption, of enterprise requires an understanding of how each discrete
resource planning systems would appear to have business action affects the entire organization.
addressed much of the underlying rationale for EIS Conventional ERP provides snapshots of organi-
development. As noted by Umble et al. (2003), zational performance but do not adequately sup-
ERP provides the twin benefits of integrating the port continuous planning activities increasingly
information processing of many departments in necessary for enterprise management. EIS has
order to create a coordinated corporate response, evolved to become a complementary component
and creating a consistent, common set of databases. to ERP systems.
It may also include integration with the informa- As suggested by McLeod Schell (2001), Fig-
tion systems of other companies. These overall ure 3 depicts how enterprise systems provide the
enterprise systems accumulate all transaction foundation for various business area systems (e.g.
data from sales, purchasing, manufacturing, and manufacturing information systems, and financial
other business functions, then provide coordinated information systems). One aspect of the enterprise
and integrated reporting of this data. This solves system is to gather and distribute data to all of the
such traditional EIS development problems of business processes within the organization. At
disparate databases and conflicting data defini- the top of this architecture resides the executive
tions. However, senior management may still information system. Executives and managers still
be inundated with status report output from the need to make decisions based on data within this
enterprise system. The primary mission of the framework. Thus, the EIS provides an overlay to
traditional EIS still remains in the ERP environ- the ERP system such that it the processes enter-
ment: to extract, filter, compress and track critical prise data into decision-making information for
executive information (e.g. Liang and Miranda, senior management.
2001; Walters et al., 2003). During the late 1990’s, the independent EIS
As noted by Bose (2006), “managing enterprise concept described previously began to be inte-
performance is an important, yet difficult process grated with more cost-effective enterprise-based
due to its complexity. The process involves moni- systems. Martinek and Szikova (2005) have dis-
toring the strategic focus of the enterprise, whose cussed such an extended system to support printed
performance is measured from the analysis of data circuit board production in Europe. Umble et al
Figure 4. EIS within the enterprise system framework
245
(2003) detail a successful implementation at Huck includes this internet support function in its En-
International, Inc. during 1998 and 1999. Huck terprise Software Suite. Vendors are increasingly
purchased the Baan ERP system to replace a legacy beginning to use the term enterprise information
MRPII hosted on HP3000hardware platform at ten system, or “business intelligence” (BI), to reflect
international sites. In addition to the expected per- this emerging role of EIS (Negash and Gray,
sonnel reductions, cost of information processing, 2003). With the advances in data warehousing
better inventory and order control, and improved and Web-networks, the traditional independent
cash management, the capabilities for improved EIS is continuing to evolve into the concept of the
management decision-making were identified and enterprise (or corporate) portal for business intel-
added as an enhancement to the system. ligence purposes. This portal concept is intended
This type of integration serves to better lever- to better integrate internal company applications
age existing ERP capabilities. For this reason, with a variety of external applications available
EIS has begun to transform into an acronym for via the internet.
enterprise information systems. This connotes an Ease of use has always been a primary re-
organization-wide system that provides consistent quirement for executive information systems.
and integrated managerial information from a Traditionally, this has inferred an intuitive graphi-
company point of view. Different users across cal user interface. To further improve the user
the enterprise can utilize the system for different interface, software vendors have begun to provide
purposes, as well as serving the purposes of top capabilities to quickly customize the EIS for the
executives. Increasingly DSS and traditional EIS specific executive user through the inclusion
capabilities are routinely added to commercially- of a toolbox option for analytical and graphical
available ERP software products. In the SAP ERP objects (Turban and Aronson, 2001). This allows
software package, these capabilities are referred easy customization capabilities. Pilot Decision
to as “strategic enterprise management systems” Suite and Comshare’s DecisionWeb provide two
(Turban and Aronson, 2001). examples of this emerging capability.
However, “ease of use” is also beginning to
imply that the system has sufficient intelligence to
Future dIrectIons automate some (if not all) aspects of the process.
Although EISs are effective in monitoring present
As the EIS concept has expanded and adapted to status, they can neither interpret nor explain. This
serve more levels of management utilizing enter- necessitates a combining of EIS functionality with
prise systems and databases, the term executive artificial intelligence capabilities. As these intel-
information system has begun to lose its relevance. ligent EIS system expand, the task of drill-down
Data warehouse products can provide data in will be progressively automated. By querying a
user-friendly, graphics-oriented query systems, “Why” function, the executive will initiate an
consistent with EIS requirements. These data intelligent agent to conduct the drill-down and
warehouses also provide filtered enterprise-wide ascertain the answer. This will depict the logic
data. The traditional linkage of an EIS to multiple stream to locate and evaluate the significance of
subject area databases, or specific applications key indicators influenced the results. Simple trend-
databases, is being replaced by a data warehouse ing or comparisons can already be determined by
interface. Further, information delivery via the such agents, with automatic alerts provided based
corporate internet is evolving into Web-enabled on comparison to a threshold value. Turban and
support systems to better access internet-based Aronson (2001) report on such a deployment of
information. For example, the SAS Institute EIS-based intelligent agents for improved environ-
246
mental scanning, i.e. to monitor specific websites mation output. Future applications of intelligent
for relevant news and pricing data. The collected EIS will automatically incorporate explanations,
data will then be routed for quantitative analysis analyses, and forecasts that will further enhance
by DSS components of the EIS. Results will be the executive’s understanding of this output. They
customized for each executive, and communicated will anticipate potential business problems by
as an e-mail alert. examining leading indicators and alerting users to
As an extension of the ESS aspect, it has been future possibilities, as well as suggesting strategies
proposed that linkages be established to groupware and solutions when provided specific constraints.
to facilitate communication among executives to
solve common problems. (Groupware refers to
collaborative software and technologies beyond reFerences
e-mail or videoconferencing. Lotus Notes is a
common example.) Other authors have proposed Alter, S. (2002). Information Systems: Founda-
architectures for distributed intelligent EISs tion of E-Business (4th ed.). Upper Saddle River,
composed of an integrated framework of intel- NJ: Prentice Hall.
ligent agents, expert systems, neural networks Belcher, L. W., & Watson, H. J. (1993). Assessing
and conventional DSS to support the executive the value of Conoco’s EIS. MIS Quarterly, 17(9),
decision-making process. For example, Bose 239–253. doi:10.2307/249770
(2006) discusses such a framework as the basis
for business intelligence-based enterprise perfor- Bose, R. (2006). Understanding management data
mance management. systems for enterprise performance management.
Industrial Management & Data Systems, 106(1),
43–54. doi:10.1108/02635570610640988
conclusIon Laudon, K. C., & Laudon, J. P. (1998). Manage-
ment Information Systems: New Approaches to
Much of the value of executive information sys-
Organization & Technology (5th ed.) Upper Saddle
tems come from their flexibility. Executives are
free to determine the problems that they need to
address, using the system as an extension of their Liang, L. Y., & Miranda, R. (2001). Dashboards
own thinking process. The most apparent benefits and scorecards: executive information systems for
of EIS are their ability to analyze, compare, and the public sector. Government Finance Review,
highlight trends, frequently through the use of 17(6), 14–19.
graphics. EIS becomes a reality when executives
Marteinek, P., & Szikora, B. (2005). Integrated
access vital information, analyze trends with
enterprise resource planning system. In 28th In-
integrated diagnostics, and can model multiple
ternational Seminar on Electronics Technology
alternatives. To be truly successful, the EIS sys-
(pp. 417-421). IEEE.
tem components must require a minimal level of
support, offer a fast development cycle, and are McLeod, R., & Schell, G. (2002). Management
responsive and adaptable to the changing needs Information Systems (8th ed.). Upper Saddle River,
of executives. NJ: Prentice Hall.
The progressive incorporation of EIS capa-
McNurlin, B. C., & Sprague, R. H. (2002). Infor-
bilities, within the enterprise system, reduces the
mation Systems in Practice (5th ed.). Upper Saddle
quantity of data that an executive must filter while
dramatically increasing the quality of the infor-
247
Moynihan, G. P. (1993a). Development of an Walters, B. A., Jiang, J. J., & Klein, G. (2003). Stra-
executive information system for Marshall Space tegic information and strategic decision making:
Flight Center. In B.Y. Kang & J.U. Choi (Eds.) the EIS/CEO interface in smaller manufacturing
Proceedings of the Second International Confer- companies. Information & Management, 40(6),
ence of the Decision Sciences Institute, Vol. 2, 487–495. doi:10.1016/S0378-7206(02)00063-0
Managing the Global Economy: A Decision Sci-
Watson, H., Rainer, R. K., & Koh, C. (1993).
ences Perspective (pp. 606-608), Seoul, Korea:
Executive information systems: a framework for
Sungrim Press.
development and a survey of current practices.
Moynihan, G. P. (1993b). An executive informa- MIS Quarterly, 17(3), 13–29. doi:10.2307/249771
tion system: Planning for post-implementation at
Watson, H. J., & Carte, T. A. (2000). Executive
NASA. Journal of Systems Management, 44(7),
information systems in government organizations.
8–31.
Public Productivity and Management Review,
Negash, S., & Gray, P. (2003). Business intelli- 23(3), 371–382. doi:10.2307/3380725
gence. In 9th American Conference on Information
Systems (pp. 3190-3199).
Rainer, R. K., & Watson, H. J. (1995). The keys to key terms And deFInItIons
executive information systems success. Journal of
Management Information Systems, 12(2), 83–98. Critical Success Factors (CSFs): A relatively
small number of easily identified operational
Rockart, J. F., & DeLong, D. W. (1988). Executive goals that are believed to ensure the success of
Support Systems: The Emergence of Top Man- an organization.
agement Computer Use. Homewood, IL: Down Drill Down Capability: EIS capability to
Jones-Irwin. display details of summary-level information by
Rockart, J. F., & Treacy, M. E. (1982). The CEO allowing the pointing and clicking on a specific
goes on-line. Harvard Business Review, 60(1), data field for which the user desires an additional
84–88. level of detail. As a result, the components of that
data field are then displayed
Turban, E., & Aronson, J. (2001). Decision Support Enterprise Information System: The linkage
Systems and Intelligent Systems. Upper Saddle of EIS and ERP to provide an organization-wide
River, NJ: Prentice Hall. system that provides consistent and integrated
Umble, E. J., Haft, R. R., & Umble, M. M. (2003). managerial information from a company point
Enterprise resource planning: implementation of view.
procedures and critical success factors. Euro- Executive Dashboard (also referred to as
pean Journal of Operational Research, 146(2), executive scorecard): Top-level EIS screen that
241–257. doi:10.1016/S0377-2217(02)00547-7 provides a single-screen display of relevant and
critical business metrics and/or statuses.
Walstrom, K. A., & Wilson, R. L. (1997). An ex- Executive Information System (EIS):
amination of executive information system (EIS) Computer-based system designed to support the
users. Information & Management, 32(1), 75–83. informational needs of senior management.
doi:10.1016/S0378-7206(97)00010-4 Executive Support System (ESS): Often
synonymous with executive information system,
it connotes an expansion of basic EIS functional-
248
ity to include extended communication support Management Reporting System (MRS):

often through inclusion of office automation and Early effort to address executive requirements by
external data access. automating data acquisition from a wide variety
Key Performance Indicators (KPIs): Spe- of corporate systems and databases, and providing
cific measures of critical success factors. online output in the form of fixed-format reports.
249
250
Chapter 1.18
Making Sense of e-HRM:
Transformation, Technology
and Power Relations
Steve Foster
University of Hertfordshire and NorthgateArinso, UK
AbstrAct formational change. Research suggests that power

relations between key groups of stakeholders, in
Several organisations have adopted e-HRM technol- particular HR Managers and line managers, may
ogy as a platform for achieving transformational influence these frames and shape attitudes to technol-
change, improving HR operational processes, al- ogy. This approach may also provide the basis for
lowing distributed access to employees / managers strategies to manage e-HRM related change more
and providing better decision support. However, effectively. Using a grounded theory approach, the
as a consultant working in this field, the author research, currently work in progress in support of
regularly encounters organisations that fail to take a professional doctorate (DBA), investigates how
advantage of the transformational potential of e- United Kingdom public sector organisations make
HRM, particularly those in the United Kingdom sense of, plan for and implement HR technology.
public sector. This chapter argues that the concepts
of sense-making and technological frames may ex-
plain the inertia experienced in some organisations. tHe trAnsFormAtIve
It contends that the analysis of technological frame eFFect oF e-Hrm
domains provides a valuable lens for understand-
ing and interpreting e-HRM, where high levels of A wide range of Human Resources (HR) pro-
frame incongruence may act as a barrier to trans- cesses and information can now be managed and
devolved to line managers and employees through
web-based technologies using e-HRM (‘electronic
DOI: 10.4018/978-1-60566-304-3.ch001
Making Sense of e-HRM
Human Resource Management’), with potentially the term ‘transformation’ itself is problematic,
significant benefits in terms of cost reduction and since it is often mis-used and abused, covering a
improved service levels (Ashton, 2001; CedarCre- range of activities from simple process redesign
stone, 2006). Organisations make a significant through to dramatic change. One definition that
investment of time and resource in implement- seems to capture the general spirit of current usage
ing e-HRM, with more than $1.5bn pa being is “the process of recreating or reinventing the HR
spent in Western Europe on related software and function—such as re-engineering, restructuring,
implementation (Lykkegaard, 2007). At least 91% implementing new systems or a new HR service
of midsize and large US organisations claim to delivery model, outsourcing or co-sourcing—with
be using web based HR technology in some way the specific intent of enhancing HR’s contribution
(Keebler & Rhodes, 2002).As a consultant work- to the business” (Mercer, 2007).
ing in the e-HRM field, the author has observed Although a recent study by the UK’s pro-
that organisations typically choose one of three fessional HR body, the Chartered Institute for
paths when implementing HR technology: Personnel and Development (CIPD, 2006), pro-
vided evidence that organisations are beginning
• Replication: Simply re-creating the to use technology to enable HR transformation,
content and functions of the existing in practice, the main focus of many software
system(s)—this is typically an IT ‘refresh’ implementation projects remains administrative
activity, often because existing technol- efficiency and HR operational cost reduction
ogy becomes non-viable or is simply out rather than strategic, transformational outcomes
of date. There is generally no intention to (Ball, 2001). As Broderick & Boudreau (1992)
restructure HR/Payroll services. note, “most organisational investments in HR
• Enhancement: New systems provide ad- information technology support only a narrow
ditional functionality (such as self-service, range of administrative decisions” (p.9) and the
improved integration and reporting) but author’s own experience suggests that the state-
are intended to have only an evolutionary ment remains largely true. Projects often pursue a
impact on the overall HR service delivery replication or enhancement strategy, implementing
model. basic functionality during the first phase, then fail
• Transformational: Involving a revolu- to capitalise on the full transformative potential
tionary restructuring of HR service de- of the investment in subsequent phases.
livery, including the use of service cen- Research literature typically explains the
tres, outsourcing and business partnering. impact of e-HRM technology on HR transforma-
E-HRM effectively becomes an enabler for tion under three, non-exclusive themes. The first
these new approaches. of these sees technology as creating more time
to be strategic—as HR processes become more
Reddington, Williamson & Withers (2005) efficient, HR functions are able to devote greater
claim that the greatest benefits of e-HRM arise time to strategic matters. For example, Snell,
when transformational outcomes are pursued and Stuebner & Lepak (2002), suggest that e-HRM
clearly, HR technology has a role to play in support- technology enables organisations to become not
ing the transformation effort. As Boroughs, Palmer only cost-efficient, but also strategic and flexible,
and Hunter (2008) observe, “The development shifting the emphasis towards people management
of human resources is bound inextricably to the processes. The introduction of initiatives such as
technology that serves it” (p.3). While HR trans- HR Business Partners, a concept championed by
formation appears to be the Holy Grail of e-HRM, Ulrich (1997), is therefore enabled through the
251
automation of transactional work. Several other transformation, acting as a stimulus for a fresh
writers, including Davenport (1993); Gourley approach to Human Resource practices and new
& Connolly (1996); Hannon et al (1996); Liff employment relationships. Kovach et al (2002)
(1997) and Tyson & Selbie (2004) also refer to conclude that the engagement of line managers
this transitioning theme. Clearly, HR functions in implementing HR technology exposes line
need to devote time to strategic activity—Lawler managers to HR issues and gives them better
& Mohrman (2003) found that successful HR appreciation of HR practices, with significant
functions tend to invest more time in planning, outcomes for HR transformation. The introduction
organisational design and development, career of technology therefore makes a statement about
planning and management development. How- transformational intent and becomes a powerful
ever, we can be sure that the link is co-relational vehicle for driving change.
not causal—simply having more time will not
make the function more strategic, even though
most studies are light on explaining how the e-Hrm And tHe uk
transition will take place. PublIc sector
Another common theme is that transformation
is achieved through the greater use of strategic The UK Public Sector generally has a poor repu-
information, which refocuses the HR function. tation for Human Resource Management, being
While technology has a strong capability for associated with low status, low influence roles
‘Automating’, it also has the capacity for ‘In- and representing an ‘enclave’ in the personnel
formating’ (a term first used by Zuboff in 1988), profession (Lupton & Shaw, 2001). While some
where informating refers to generating new research (Kelly & Gennard, 1996) found that some
forms of information to empower managers and Personnel1 Directors have developed a strategic
provide data that HR specialists can strategically role in managing the consequences of government
act upon. Othman (2003) for example, refers to changes, in others, the HR role has declined, as
the ‘informated workplace’. Lawler & Mohrman industrial relations issues have seemingly become
(2003) found that the greater use of information less critical. Evidence seems to support the idea
technology may be associated with HR being that the HR function in the public sector is rela-
more of a business partner, arguing that “..it is tively immature and less sophisticated than its
much easier for HR to gather strategic data and private sector counterparts (Harris, 2002).
analyze them in ways that can contribute to form- As a result of the Gershon Report (Gershon,
ing and implementing business strategy” (p.21). 2004), there has been pressure on the UK Public
At its boldest, technology provides opportunities Sector to become more efficient, by reducing over-
for virtual and networked organisations, linking staffing and inefficiency. Transformation of the
e-learning to knowledge management and the HR function has therefore become a key priority
potential for new HR business models. in recent years, awakening an interest in technol-
A third theme is the symbolic nature of ogy. Although e-HRM has been widely used by
technology. Kossek et al. (1994) have noted the private sector over recent years, many public
the role of technology in strategic positioning, sector organisations are now embarking on a pro-
that can “practically and symbolically represent gramme of investment in e-HRM as part of their
the transformation of Human Resources into a transformation strategy, providing an excellent
strategic business partner” (p.137). Likewise, research platform for investigating perceptions
Tansley, Newell & Williams (2001) contend that of technology, its role and the approaches taken
technology plays a critical part in driving HR towards implementation.
252
tHe e-Hrm lAndscAPe tion and are aimed at reducing costs and achieving
HR operational efficiencies through automation.
One barrier to greater understanding of HR The second group, shown in the upper half, are
technology is that a multiplicity of terminology ‘Human Capital’ functions, aimed at supporting
is used—HRIS, HRMS, web portals, e-HR and, people management activities such as performance
of course, e-HRM, all of which serve to confuse management, skills profiling and analytics. Be-
potential users. However, for the purposes of this cause of the wider impact on business outcomes,
chapter, the term ‘e-HRM’ will refer to a fully these functions are more strategically oriented.
integrated, organisation-wide electronic network It has been suggested elsewhere (Foster, 2006)
of HR related data, information, services, data- that the term ‘e-Human Capital Management’, or
bases, tools, applications and transactions, best ‘e-HCM’ should be used to describe these more
summarised as: sophisticated applications of HR technology.
This model, developed by the author, has
The application of any technology enabling man- proved to be highly useful as a strategic planning
agers and employees to have direct access to HR tool to demonstrate the ‘art of the possible’. In
and other workplace services for communication, reality, the impact of HR technology is likely to be
performance reporting, team management, knowl- some combination of process and human capital
edge management and learning…in addition to functions, analysis of which is beyond this chapter.
administrative applications. (Watson-Wyatt, 2002) However, the breadth of literature on these themes
is rich enough to suggest that definitions of e-HRM
Functionality may be categorised into two which focus only on operational or administrative
areas, each of which has different but significant outcomes are likely to be inadequate.
transformational potential. The first of these (Pro- While organisations are becoming more so-
cess Technology—lower half) provides support phisticated and ambitious in their use of e-HRM,
for basic transactions, now well-established and many HR professionals and line managers con-
the foundation of all commercial HR technology. tinue to lack a clear knowledge of its transforma-
These functions include managing payroll, per- tive effect. For some, technology is seen as little
sonal / organisational data and routine administra- more than an ‘electronic filing cabinet’, a basic
Figure 1. e-HRM landscape
253
repository of data, with little understanding of its labelling and categorising to stabilise the stream-
potential for enabling change. As we have seen ing of experience, connecting the abstract with
from the e-HRM landscape model (Figure 1), there the concrete”. This idea of creating ‘order out of
is now a bewildering choice of functionality avail- chaos’ will be familiar to anyone who has been
able, either as a module in a company-wide ERP involved in a technology project, where goals are
solution, as standalone ‘best-of-breed’ software or discovered through social processes involving
as a series of modules in traditional HR software. argument and debate (Daft, 1986).
Arguably, e-HRM technology is advancing faster If we accept that the use of e-HRM is funda-
than many organisations are able to comprehend mental to HR transformation, then being able to
and implement what is available, raising ques- comprehend, plan for and ultimately implement
tions about how organisations make sense of what e-HRM becomes a core organisational capability,
technology can offer. This has critical implications not just for HR teams, but for all business lead-
for the approach to e-HRM. ers. Various explanations are available as to why
organisations differ in their ability to make sense
of e-HRM. One useful concept here is the idea of
mAkInG sense oF e-Hrm ‘absorptive capacity’. Zahra and George (2002),
adopting the earlier work of Cohen and Leventhal
The concept of sense-making has its roots in the (1990), define absorptive capacity as the acquisi-
personal construct theories of Kelly (1955), which tion, assimilation, transformation and exploitation
see our understanding of the world as a social proof knowledge to produce a new organisational
cess. Organisations are essentially a collection of capability. Those with higher absorptive capacity
people trying to make sense of what is happening (i.e. those better able to make sense of its potential)
around them through interpretations, translating will naturally be better able to exploit technology2.
events, developing models for understanding and Likewise, group learning processes may also be
assembling conceptual schemes—mapping out seen as important mechanisms in understanding
their world to create an intelligible whole (Daft & and developing technological knowledge. Bonda-
Weick, 1984). E-HRM takes place exclusively in rouk (2006) argues that acceptance of technology
an organizational setting, with technologies that is related to group learning processes, which can
are often new and unfamiliar. In order to interact only be achieved through complex interactions.
with technology, people first have to make sense Using Kolb’s (1984) learning theory, Bondarouk
of it; in this sense-making process, users develop describes a 5 step action-oriented group learning
assumptions, expectations and knowledge of the approach where individuals start to use technol-
technology, which then serve to shape subse- ogy in a haphazard way and gradually, through
quent actions toward it. These perceptions have shared understanding, mutual adjustment occurs
a powerful influence on expectations—Ginzberg in which the use of technology is agreed.
(1981) showed how users’ expectations of tech-
nology influenced their perceptions of it and how
cognitive and ‘micro-level processes’ became tecHnoloGIcAl FrAmes
important to understanding these perceptions. The
mental models formed when exploring technol- Orlikowski & Gash (1994) argue that the concept
ogy are important sense-making devices during of ‘technological frames’ offers a useful analytic
processes of organizational change (Bartunek perspective for explaining and anticipating actions
& Moch, 1987). As Weick (2005 p.411) states and meaning. Frames are related to sense-making
“sense-making starts with chaos and involves and personal constructs, being cognitive structures
254
or mental models that are held by individuals, reference between stakeholders shape the plan-
typically operating in the background with both ning and development of e-HRM, both before
facilitating and constraining effects. These indi- and during implementation. It suggests that this
vidual frames of reference have been described understanding is critical to an organisation’s ability
as “a built-up repertoire of tacit knowledge that to plan for and ultimately deliver HR transforma-
is used to impose structure upon, and impart tion. If stakeholders are unable to make sense
meaning to, otherwise ambiguous social and of technology and agree common (congruent)
situational information to facilitate understand- technological frames, as a minimum there will
ing” (Gioia, 1986 p.56). They have a powerful be issues of resistance and potentially significant
effect on people’s assumptions, expectations and barriers to progressing beyond administration
knowledge about the purpose, context, importance towards transformational HR.
and role of technology. Frames can be both posi-
tive and negative—where they are positive, there
is shared understanding and mutuality, whereas reseArcH metHodoloGy
negative frames can become “psychic prisons”
that inhibit learning because people “cannot Approach
look at old problems in a new light and attack
old challenges with different and more powerful The underlying methodology of the research is
tools—they cannot reframe” (1991 p.4).E-HRM based on the formative work on grounded theory
implementation projects are often intensely po- by Glaser & Strauss (1967). Grounded theory is an
litical in nature, due to the challenges associated inductive approach to theory development that at-
with the cost of investment and the changes to tempts to account for observed behaviour through
ways of working that are involved. Strong frame a series of theoretical propositions. In this case,
congruence may therefore be seen as a critical the focus is on how organisations make sense of
pre-requisite of project success and where frame e-HRM, its use as a transformational tool, how it
incongruence exists, difficulties and conflict may is perceived by HR and line managers and how
arise. Orlikowski & Gash (1994), researching the those perceptions are manifested through to final
introduction of Lotus Notes into a consultancy implementation and use.
organisation, found that three ‘domains’ charac-
terised interpretations of technology: research sample & Approach
• Technology in use: People’s understand- Several methods were used to collect data, the
ing of how the technology will be used on primary method being a series of semi-structured
a day-to-day basis. interviews. Research also included documentary
• Nature of technology: People’s images of analysis, workshops, participatory observation
the technology and their understanding of and the maintenance of a reflective diary. Inter-
its capabilities and functionality. views were aimed at obtaining an insight into
• Technology strategy: Their understand- the sense-making processes of human resource
ing of why the organisation acquired and professionals and line managers with regard to
implemented the technology. the use of e-HRM technology. Participants were
drawn from public sector organisations at dif-
The current research is aimed at exploring ferent stages of their implementation of e-HRM
how various organisational actors make sense of technology, a mix of consultancy clients of the
e-HRM by examining how the various frames of author or obtained via direct approaches through
255
networking at conferences, events etc. Participants Theme 1: Risk-averse, bureaucratic and under-
either used, or expected to use, a range of software valued—An HR function under pressure.
products to meet their e-HRM needs. Some initial It was no surprise to find that interviews sup-
screening was performed against participants to ported the stereotypical idea of a public sector HR
understand the current stage of their project and function, struggling to establish a credible repu-
whether reasonable access could be granted. The tation and highly concerned with demonstrating
organisations participating were: value. The HR functions studied were typically
working under great pressure, with sub-optimal
• 6 Local Councils (with extended access to resources, often failing to deliver a good service.
one Council) Line managers were especially critical of the
• 1 regional police constabulary quality of HR service delivered:They don’t give a
• 2 Government agencies s**t [Council, Line Manager]And:We get a c**p
• 1 Health Authority service [Council, Line Manager]All too often,
interviews portrayed organisations as highly rule-
To date, 41 interviews have been conducted, bound and bureaucratic. Line managers frequently
including HR professionals, HR Directors and line expressed a concern that the HR function was
managers on an individual basis, plus a series of defensive, risk averse and often uninterested in
group workshop discussions with line managers. strategic people development issues:
Interviews lasted between 45 minutes and 2 hours
and were recorded and transcribed for analysis. We’re probably a bit more slaves to the rules of the
Interviewees were encouraged to engage in a broad- processes here. I think probably we made things
ranging discussion in order to provoke responses, a bit difficult for ourselves in areas—we do tend
rather than the interviewer being a ‘speaking to tie ourselves up in knots and I do think it sets
questionnaire’. Views were invited on the context us back, a different way with more freedom would
of e-HRM, including an insight into what progress be good. [Council, Line Manager]
had been made with technology, concerns relat-
ing to its introduction and their perceptions of the Play it safe, rather than make a decision which
impact of technology. They were also asked about needs to be made. We’re less bold in making deci-
the role of the HR function and the people manage- sions, we’re good at administrative decisions but
ment capability of the organisation. Line manager less bold on big decisions. [Council, HR Manager]
access has proved to be challenging—experience
suggests that HR managers are protective of their HR is much more like a sort of emergency plumber
‘customers’ and initially act as gatekeepers. They service, there to give us help when things go wrong,
are often reluctant to give permission to meet line such as where we have a long-term sickness issue
managers until they have at least been through the or for sorting out the back-office processes like
research process themselves. salaries and sickness. [Council, Line Manager]
I’m not sure we’ve all got a shared understand-

key tHemes ing of what the hell they’re supposed to be doing.
I think that’s the problem. For me, it’s a people
A prototype template analysis approach was used business; we should have a very strategic view
to analyse interviews. Although originally intended of what the role of our HR support is. [Council,
to provide context for the technology discussion, Line Manager]
two key themes arose consistently throughout:
256
Strategic means assessing the policies, making That’s part of the culture change, we’ve been
sure the council operates in a consistent way trying to give managers more accountability and
and like any organisation employing people, that trust. It’s been a hard year for the centre, because
people abide by it. [Council, HR Manager] they feel they have to check everything—we said
we’re not going to be checking everything in fu-
One recurring theme was that line managers ture. Some of the managers think everything will
seemed to appreciate the contribution of individual get checked. [Government Agency, HR Manager]
HR people but were dismissive of the function as a
whole. A complex relationship exists between line Although not stated so explicitly, other HR
managers and functional HR staff that is beyond managers expressed an aspiration that improved
this chapter to explore. line management capability may be an additional
Theme 2: Line Managers are perceived as outcome of the e-HRM investment, where tech-
lacking people management capability.A second nology acts as the ‘symbolic’ representation of
major theme was a series of strong concerns about a new way of working (see Tansley, Newell &
the capability of line managers in their people Williams, 2001).
management role. HR managers expressed this
in terms of a frustration with line management’s
need for constant ‘hand-holding’ whenever people tHeory develoPment
management issues arose. A common view was
that: Following the structure proposed by Orlikowski
& Gash (1994), research findings were analysed
Some managers would like HR to come along from the perspective of an interpretative techno-
and take away the people management function. logical frames model using the prototype template
[Council, HR Manager] analysis structure. The three core technological
domains form the basis for comparing differing
I see it time and time again, one recent example views about the use, nature and strategy of technol-
is where a senior manager was in a terrible state, ogy, revealing varying degrees of frame congru-
she had never thought that as part of being a ence and some significant frame incongruence.
manager, she would find herself in an employment Technology in use: People’s understanding of how
tribunal, she didn’t have the skills to manage the the technology will be used on a day-to-day basis.
situation. [Council HR Manager] This domain addresses the operational perception
of e-HRM—that is, how technology is being (or
Won’t face up to their responsibilities, disciplinary will be) used to support HR delivery, particularly
processes can take forever because the managers in administrative areas. Without exception, HR
don’t like disciplining, grasping the nettle and managers were enthusiastic about the automating
dealing with it. [Council Line Manager] impact of e-HRM:
There are some managers for whom it’s impossible, What we’re really hoping in terms of the new
they’re managing huge numbers of people, there computer system we’re introducing is that it will
is legislation which requires them to work within take away some of that pressure, particularly
boundaries, they don’t have the luxury to be good on some of the administrative things which will
people managers. [Council, HR manager] now be done automatically. We are relying very
much on this new system doing that. [Council,
HR Manager]
257
Managers will see the benefits quickly… they’re I think it’s good for processing between the manag-
not stupid, they will see it. There’ll be a bit of ers and HR but I think between the staff and the
work, but a lot of gain. So I’m hoping that there managers it’s cut the tie a little bit [Government
will be a balance and the fact that there is a vast Agency, Line Manager]
amount of information available to them that they
will actually appreciate that overall it is a good I think they want the accountability but they don’t
thing for them. [Council, HR Manager] want to fully own the responsibility that goes with
it. I think managers do kind of want it both ways.
There was an attempt at trying to tell us what the The organisation is saying one thing and allow-
benefits were but I don’t think I’ve reaped those ing another. We have to follow it through, let’s
benefits yet. [Government Agency, Line Manager] say it much firmly, much more clearly. [Council
HR Manager]
We were all trained some time ago, it may be that
I haven’t picked up on some of this. I think it’s I’m not sure [managers] would want too many
quite difficult. [Council, Line Manager] fancy functions because they don’t have very much
freedom to act anyway… more sophisticated tools
I remember leaving the training session and think- wouldn’t be that much use to them. [Council, Line
ing, well that’s another load of extra work for me Manager]
for no perceived benefits. I don’t mind putting extra
effort in if I’m going to get something tangible Although there is a broad level of agreement
at the end of it, and I just thought well here we about the potential of e-HRM for administrative
go, we’ve a great deal more to do. [Government improvements, with high expectations about how
Agency, Line Manager] it will be delivered, there appears to be a mismatch
of views about e-HRM in practice, revealing
There was quite a bit of resentment initially about shortcomings in the way that the new systems
the system and I have to say my team were quite had been implemented.
sympathetic, saying this is going to give you a Nature of Technology: people’s images of the
lot more work and feeling sorry for me and they technology and their understanding of its capa-
were kind. [Government Agency, Line Manager] bilities and functionality.This domain deals with
the overall capabilities and potential of e-HRM
Let’s be honest, however you dress self-service up, and the perceived impact on the roles of HR and
you’re giving them an additional task to do that line managers. HR managers generally held a
they didn’t do before, that work doesn’t actually go highly positive view of e-HRM capabilities, with
away and if you transfer it over you can fool yourself several describing it in terms of a ‘magic bullet’,
into thinking you’ve actually made some efficiency solving a variety of process problems. This is not
savings and you haven’t. [Council, HR Manager] uncommon in IT projects and can lead to highly
unrealistic expectations (Markus & Benjamin,
I always preferred the old system, where once 1997). For example:
a year you get a manual form and confirmed.
This system doesn’t pick up those people who Self-service is going through meteoric develop-
have forgotten or can’t be bothered to tell you ment, isn’t it, in terms of the potential of what it
they’ve moved home whereas it it’s just relying on can do, there’s nothing that we can’t do through
remembering to go in. It all sounds a bit dodgy. self-service. [Council, HR Manager]
[Council, Line Manager]
258
It would be the manager’s friend—basically the With the system now, we can do compliance reports
manager would relate completely to that system on who has done appraisals, who hasn’t, who has
and it would help him or her actually manage recorded absence and who hasn’t, and they are
their staff. It gives them all the information they regular kind of reports… it goes to the Directors,
need, members might say we need to know the it goes to members, they go to the chief executive.
turnover in your area. In future, no matter what [Council HR Manager]
the personnel question is that comes up—I want
the manager to feel confident they can interrogate I don’t think that technology is seen as the tool
the system easily and out it comes in a report. for policing us. Managers can get a breakdown
[Council, HR Manager] of absence and there are lots of reports available
for monitoring so in that sense they are getting
It’s partly about being an enabler, a large percent- something out of the system. Staff may not have
age of the personnel staff are out there profession- cottoned on to that yet but I personally think it’s
ally qualified, and I would like them to be using great that we’ve got the information there and
that as much as possible which means having we can do something about it. [Council, Line
fewer of them and fewer of them passing bits of Manager]
paper, also it’s much more motivating for them.
[Council, HR Manager] It will take us out of the role of chasing manag-
ers, having to police them, instead the system
There were two drivers really, one was definitely should help them and only that small minority of
efficiency to get the cost of the service down and managers that aren’t capable will say the system
the other driver was to get a more modern and is coercing them, that the technology isn’t deliver-
professional service to meet the needs of the ing. [Council, HR Manager]
business. They were the two catalysts of change
really, efficiency and modernising the service. However, interviews revealed major frame
[Government Agency, HR Manager] incongruence with regard to the capabilities and
outcomes of e-HRM; whereas HR managers saw
I think a year ago, some managers would have a step change in their capability, line managers
thought ‘I don’t know what HR does’, but they often did not recognise or acknowledge the trans-
are now trying to use HR technology to link into formational outcomes or where e-HRM might
their systems, their processes which in turn makes change or enhance their line roles.
it better for everybody. [Council, HR Manager] Technology Strategy: People’s understanding
of why the organisation acquired and implemented
We do other things around reporting so that on the the technology.This domain addresses the more
personnel intranet site you can set an alert which strategic aspects of e-HRM and is particularly
tells you for example when the latest turnover relevant to transformational objectives. Unless
figures for your department are available so I’ll actors have a common point of reference and
do a report, put it on the intranet and you get an clear understanding of the transformational effect
alert telling you about it. [Council, HR Manager] of e-HRM, a process-focused strategy is likely,
so strong frame congruence about the strategic
When I ask them what they’ve done about absence, objectives of e-HRM would appear to be a pre-
with this new system, they can’t say they don’t requisite. However, we find very different frames
know. [Government Agency, HR Manager] of reference among the two key stakeholder
groups. Some HR Managers took the view that
259
e-HRM would lead to a ‘time-shift’transformation, The whole system coming has completely changed
where the automation or devolvement of adminis- their perceptions they are now beginning to see the
trative workload would allow the HR function to value of it, to take away the tedious systems and
concentrate on strategic issues. Others saw a more they’ll see it pay back for all the work they’ve put
active, direct transformational impact of e-HRM, in, I see self service as a starting point. [Govern-
where the technology itself would enable higher ment Agency, HR Manager]
levels of managerial capability:
The system won’t make us strategic—because we
Eventually, we want to reduce HR administration, have to believe we are already strategic. [Council,
but also it’s about strategic change, giving line HR Manager]
managers the tools to be more effective, that’s what
it boils down to, putting things online, Personal No, I think that’s rubbish. I think because of the
Development Reviews, sickness etc. we’re giving nature of HR you are driven operationally. I think
you all that technology, they will have to deal the biggest barrier is professionals not having
with it in future. [Police Authority/HR manager] the capability to be strategic and really focusing.
[Council HR Manager]
There is a chain—the technology can drive changes
in the way that managers operate and the way we My feeling is that time will just be swallowed up
recruit—I also get really enthusiastic about this. with day to day work. I’ve never yet seen—and
[Council HR Manager] one of the reasons why I’ve been a bit of a wet
blanket in terms of supporting the system—I’ve
It was always seen as a way of helping manag- never yet seen a successful personnel system and
ers do their job better with better management I’ve seen three. The reason is they all get watered
information, understanding learning and devel- down and Personnel say “we can’t do this” and
opment interventions, looking at their record of managers don’t get engaged and the managers
attendance, all of those things, so it was more say that it does nothing for them. [Council, Line
a tool to do your job but not the end in itself. Manager]
[Government Agency, HR Manager]
I don’t think we’ve thought it through. We know
The other aspect to it is that you’d hope that not we’re going to save some costs, but beyond that
only will it allow us to take out posts across the we don’t really know what to expect. [Council,
organisation, but if set up properly it should allow HR Manager]
mangers to manage their teams in a better way.
[Council Line Manager] One finding of concern was that HR Busi-
ness Partners (those staff directly supporting line
Yes, the good ones use the information or know managers in an internal consultancy role) tended
the information is there and will ask us to get to avoided involvement in e-HRM projects, see-
more involved in professional issues rather than ing little or no relevance to technology beyond
processes. The less good ones will still complain administration. It has been difficult to test this
about having to do personnel work. [Council, technological frame, since several planned in-
HR Manager] terviews with Business Partners were cancelled
at short notice. It may be that this subset of HR
managers underestimates or misunderstands the
potential transformational impact of technology,
260
which would also restrict the scope of transfor- HR system? A poor track record of strategic HR
mation programmes. Attempts will be made to engagement is more likely to lead to a belief that
explore this frame further as research progresses. technology will also be poorly implemented. If
this is true, then an analysis of e-HRM limited
to the three technological domains may provide
AnAlysIs oF FIndInGs an inadequate explanation of sense-making
processes. We must also examine the context in
Interviews reveal that HR managers and line which technology operates, in particular the power
managers have very different views of technol- relations between groups of stakeholders, form-
ogy in terms of its day-to-day use, capability and ing a three-way relationship to technology. This
strategic value. Using the technological frames is consistent with Klein and Kleinman (2002),
model developed by Orlikowski and Gash (ibid.), who argue that power relations play a key role in
we can compare attitudes along the three core the social shaping of technology. As a result, the
technological sense-making domains. Table 1 following proposition is made:
summarizes the key themes from interviews in Context is critical: technological frames are
tabular form against these domains: moderated by the power relations between groups
There is an irony lurking here—while the of actors, in this case, the mutual opinions of HR
HR function regards technology as an enabler of teams and line managers.One particularly reveal-
transformation and a tool to enhance its strategic ing interview with a Local Council Line Manager
impact, if the line managers it serves doubt the highlighted the need for transformational e-HRM
general strategic capability, why should they decisions to be taken in the context of a clear HR
trust HR management to introduce a strategic strategy, by a credible HR function:
Table 1. Technological frames for HR managers and line managers
STAKEHOLDERS
Domain How HR Managers see e-HRM How Line managers see e-HRM
Technology takes lot of time to get working and needs
an investment from users to get benefits; it’s just one
Managers will understand and use the system and
more thing to learn
Technology in use will intuitively grasp the system
HR is ‘dumping its dirty work’
The manager’s friend: alerts, warnings, guidance
Those with experience feel that that the system doesn’t
help them much
Initially see technology as a way of reducing the Technology should bring about process improvement
cost of HR operations, reduced headcount “These things never work”.
Technology seen as a magic bullet, expert system, Technology not likely to be ‘sold’ to managers prop-
giving them greater control over non-compliant erly.
Nature of technology
managers Assume that better management information will be
Believe that e-HRM will make managers better at delivered
their roles Cynicism that it will lead to meta-regulation from the
Believe e-HRM will produce better HR information centre
Strategic aspiration often limited to administration
and HR services
Desire to make long-term shift in HR function (time Technology is an opportunity (but only one among
shift) many other strategies)
Technology strategy
Doubt that technology will bring about a shift in HR needs to prove its operational capability before
strategic focus they can be trusted to implement technology
Technology is unlikely to impact on professional
HR roles
261
Without a clear vision of where the authority needs frame congruence is with regard to line manager
to be, we’re struggling to fit with HR initiatives, capability—line managers believe their colleagues
it’s still an authority with glib statements about don’t take it seriously and HR managers agree
what it wants to do. If it embeds properly, the in- with this, seeing technology as a means of iden-
formation will be invaluable in helping to develop tifying poor managers. One conclusion from this
a strategic approach, but without the information, is that transformational objectives with regard
no, it will struggle to be any more than something to technology are more likely to be supported
that enhances processes. [Council, Line Manager] and shared where good levels of credibility and
capability are in place.
Table 2 sets out these additional ‘power rela- When viewed in conjunction with the three core
tions’ domains, examining how HR managers technological domains, power relations domains
and line managers see one another. For example, offer a new insight into the context of e-HRM
whereas HR managers are likely to believe that implementation. (see Figure 2)
e-HRM will improve strategic information, line
managers often doubt that the HR function has
the ability to deliver strategic e-HRM and deliver
transformative change. Sadly, one key area of
Table 2. Power relations: Moderating frames
Domain HR Manager View Line Manager View

HR function Under valued HR is bureaucratic and policy driven
strategic capability Lack good information or know how to use it Value individual HR people but not the function as a
Technology will improve strategic information whole
No clear idea of the role and strategy of HR
Doubt that HR can deliver strategic e-HRM
Management Managers need a lot of ‘hand holding to manage people People management sometimes seen as a ‘necessary
capability Managers don’t understand their people management evil’, secondary to operations
responsibilities Colleagues won’t face up to their people management
Technology will improve managerial capability + responsibilities
weed out the weak managers No coherent view on technology
Figure 2. Relationship of core frames, moderating frames and technology strategy
262
dIscussIon And conclusIon technology replication or enhancement strategies

rather than transformational ones.
The technology to bring about transformational An investigation of e-HRM is in itself a sense-
HR change has existed for some time. While some making process and the concept of technological
organisations have succeeded in e-enabling their frames is proving to be a valuable tool for analysing
HR organisation, others have created bold plans and categorising stakeholder attitudes. Research
but have been unable to bring them to fruition. continues into this area and it would benefit from
One reason for this is that software capability is greater input by line managers, to understand
often far ahead of organisational ability to take better their relative technological frames and
make sense of and take advantage of its potential. explore the impact of power relations. Of par-
As a result, greater understanding is needed about ticular value would be to identify the conditions
the process through which organisations define that lead to increased or decreased congruence,
and understand e-HRM, as well as how they plan so that transformational objectives could be more
for its use. precisely stated and realised. Frame analysis may
This chapter argues that this process is not also prove to be a useful tool in tracking changes
simply about understanding technology in terms in the meanings ascribed to technology over time.
of the three core domains (use, nature and strat- To address this, this research will continue into
egy), but that power relations between groups of 2009 with further interviews planned with line
stakeholders also moderate attitudes towards it. So, managers. It is hoped that this work will make
where the HR function doubts that line managers an important contribution to real-world usage of
are capable of managing people effectively, or e-HRM and the development of tools to support
line managers doubt that HR knows how to get change management initiatives.
the best out of technology, there will be an impact
on expectations of what technology can achieve.
While there appears to be basic frame congruence reFerences
about e-HRM as an administrative tool, there is
likewise significant frame incongruence on the rel- Ashton, C. (2001). eHR Transforming the HR
evance of e-HRM for achieving transformational function: Business Intelligence.
outcomes. These include a major gap between Ball, K. (2001). The use of human resource in-
what HR feels it has achieved and the reality of formation systems: a survey. Personnel Review,
those managers that are using it, who can tend to 30(5/66), 677-693.
feel under-prepared and under-whelmed by their
experience. It appears that HR managers and line Bartunek, J. M., & Moch, M. (1987). First
managers lack a common language or frame of order, second order and third order change
reference for discussions about technology—as and organization development interventions:
long as there is mutual disrespect between the two A cognitive approach. The Journal of Ap-
groups about their capabilities, these discussions plied Behavioral Science, 23(4), 483–500.
will always be strained. By examining the differ- doi:10.1177/002188638702300404
ent, often polarised attitudes of line managers and Bolman, L. G., & Deal, T. E. (1991). Reframing
HR teams towards technology, we can cast light Organizations: Artistry, Choice and Leadership.
on which frames of reference are shared within San Francisco, California: Jossey-Bass.
and across groups and with what consequences. It
may be that, ironically, organisations which do not
already have strong HR capability tend to pursue
263
Bondarouk, T. V. (2006). Action-oriented group Gioia, D. (1986). The Thinking Organization. San
learning in the implementation of information Francisco, California: Jossey-Bass.
technologies: results from three case studies.
Glaser, B. G., & Strauss, A. L. (1967). The discov-
European Journal of Information Systems, 15,
ery of grounded theory: Strategies for qualitative
42–53. doi:10.1057/palgrave.ejis.3000608
research. New Tork NY: Aldine.
Boroughs, A., Palmer, L., & Hunter, I. (2008). HR
Gourley, S., & Connolly, P. (1996). HRM and com-
Transformation Technology: Delivering Systems
puterised information systems—have we missed
to Support the New HR Model. Aldershot: Gower.
a link? Paper presented at conference—strategic
Broderick, R., & Boudreau, J. W. (1992). Human direction of HRM.
resource management, information technology
Hannon, J., Jelf, G., & Brandes, D. (1996). Human
and the competitive edge. The Executive., Spring,
resource information systems: operational issues
6(2), 7-17.
and strategic considerations in a global environ-
CedarCrestone. (2006). CedarCrestone 2006 Hu- ment. International Journal of Human Resource
man Capital Survey: Cedar Crestone. Management, 7(1).
CIPD. (2006). Technology in HR: How to get the Harris, L. C. (2002). The future for the HRM
most out of technology in people management. function in local government: everything has
Wimbledon: CIPD. changed- but has anything changed? Strategic
Change, 11, 369–378. doi:10.1002/jsc.612
Cohen, W., & Leventhal, D. A. (1990). Absorp-
tive capacity: A new perspective on learning and Keebler, T. J., & Rhodes, D. W. (2002). E-HR
innovation. Administrative Science Quarterly, 35, becoming the ‘path of least resistance’. Employ-
128–152. doi:10.2307/2393553 ment Relations Today, 29(2), 57–66. doi:10.1002/
ert.10041
Daft, R. L. (1986). Organization Theory and
Design (2nd ed.). St. Paul, MN: West. Kelly, G. A. (1955). The Psychology of Personal
Constructs Vol 1 and 2. New York: Norton.
Daft, R. L., & Weick, K. E. (1984). Toward a
model of organizations as interpretation systems. Kelly, & Gennard, J. (1996). The role of Person-
Academy of Management Review, 9(2), 284–295. nel Directors on the board of directors. Personnel
doi:10.2307/258441 Review, 25(1), 7-24.
Davenport, T. H. (1993). Process Innovation. Klein, H. K., & Kleinman, D. (2002). The social
Boston M.A: Harvard Business School Press. construction of technology: structural consider-
ations. Science, Technology & Human Values,
Foster, S. (2006). A high tech future. Payroll Man-
27(1), 28–52.
ager’s Review, November 2006, (pp. 38-40).
Kolb, D. A. (1984). Experiential learning. Experi-
Gershon, P. (2004). Releasing Resources to the
ence as the source of learning and development.
Frontline: Independent Review of Public Sector
Englewood Cliffs, New Jersey: Prentice-Hall.
Efficiency. In H. Treasury (Ed.).
Kossek, E. E., Young, W., Gash, D. C., & Nichol,
Ginzberg, M. J. (1981). Early diagnosis of MIS
V. (1994). Waiting for innovation in the Human
implementation failure: Promising results and un-
Resources Department: Godot implements a
answered questions. Management Science, 27(4),
Human Resource Information System. Human
459–478. doi:10.1287/mnsc.27.4.459
Resource Management, Spring, 33(1), 135-139.
264
Kovach, K., Hughes, A., Fagan, P., & Maggitti, P. Snell, S. A., Stuebner, D., & Lepak, D. P. (2002).
(2002). Administrative and strategic advantages Virtual HR departments: Getting out of the middle.
of HRIS. Employment Relations Today, 29(2), In D. B. G. R. L. Henneman (Ed.), Human Re-
43–48. doi:10.1002/ert.10039 source Management in virtual organisations (pp.
81-101). Greenwich: Information Age Publishing.
Lawler, E. E., & Mohrman, S. (2003). HR as a
strategic partner—what does it take to make it Tansley, C., Newell, S., & Williams, H. (2001).
happen? Human Resource Planning, 26(3), 15–29. Effecting HRM-style practices through an in-
tegrated human resource information system:
Liff, S. (1997). Constructing HR information
An e-greenfield site? Personnel Review, 30(3),
systems. Human Resource Management Jour-
351–370. doi:10.1108/00483480110385870
nal, 7(2), 18–30. doi:10.1111/j.1748-8583.1997.
tb00279.x Tyson, S., & Selbie, D. (2004). People processing
systems and human resource strategy. Interna-
Lupton, B., & Shaw, S. (2001). Are public sector
tional Journal of HR Development and Manage-
personnel managers the profession’s poor rela-
ment, 4(2), 117–127.
tions? Human Resource Management Journal,
11(3), 23–38. doi:10.1111/j.1748-8583.2001. Ulrich, D. (1997). Human Resource Champions:
tb00043.x The Next Agenda for Adding Value and Delivering
Results. Boston MA: Harvard Business School
Lykkegaard, B. (2007). Western European Human
Press.
Capital Management and Payroll Applications
Forecast, 2007-2011: IDC. Watson-Wyatt. (2002). B2E/e-HR Survey Results.
Markus, M. L., & Benjamin, R. I. (1997). The Weick, K., Sutcliffe, K. M., & Obstfeld, D. (2005).
magic bullet theory in IT enabled transformation. Organising and the process of sensemaking. Or-
Sloan Management Review, (Winter): 55–58. ganization Science, 16(4), 409–421. doi:10.1287/
orsc.1050.0133
Mercer. (2007). HR Transformation 2.0: It’s all
about the business. Zahra, S. A., & George, G. (2002). Absorptive
capacity: A review, reconceptualisation and ex-
Orlikowski, W., & Gash, C. (1994). Techno-
tension. Academy of Management Review, 27(2),
logical frames: making sense of information
185–203. doi:10.2307/4134351
technology in organisations. ACM Transac-
tions on Information Systems, 12(2), 174–207. Zuboff, S. (1988). In the age of the smart machine.
doi:10.1145/196734.196745 New York: Basic Books.
Othman, R. (2003). On developing the informated
work place: HRM issues in Malaysia. Human
Resource Management Review, 13, 393–406. key terms And deFInItIons
doi:10.1016/S1053-4822(03)00042-1
e-Human Capital Management (e-HCM):
Reddington, M., Williamson, M., & Withers, Technologies that are specifically focused on
M. (2005). Transforming HR: Creating value supporting the people management aspects of an
through people. Oxford: Elesevier:Butterworth- organisation, providing managers with the tools
Heinemann. to support this activity.
Enterprise Resource Planning: (ERP):
Systems such as SAP and Oracle, covering a
265
wide range of business functions such as Finance, out meaning and assembling conceptual schemes.
Manufacturing, Logistics and Procurement. These These sense-making processes are thought to be
large systems typically have a Human Resources critical in influencing the decisions that are made
module. about e-HRM.
HR Transformation: “The process of rec- Technological Frames: Cognitive structures
reating or reinventing the HR function—such as or mental models that are held by individuals
re-engineering, restructuring, implementing new that impose structure on and provide meaning
systems or a new HR service delivery model, out- for social and situational information to facilitate
sourcing or co-sourcing—with the specific intent understanding. They influence our assumptions,
of enhancing HR’s contribution to the business” expectations, and knowledge and can be positive
(Mercer, 2007). and negative.
Human Capital Technology: Systems or parts
of systems that support the management of people,
such as providing decision support, planning and endnotes
employee data.
Process Technology: Systems or parts of
1
Many UK Public Sector organisations still
systems that re aimed at supporting transactional refer to the Human Resources function as
HR processes such as payroll, personal data man- ‘Personnel’. The significance of this is be-
agement, enrolment etc. yond discussion in this chapter.
Sense Making: People in organisations try
2
I see the concept of absorptive capacity as
to understand what is happening around them analogous to a sponge, that becomes full of
by making interpretations, translating events, water and has no further capacity for ideas
developing models for understanding or bringing until more capacity becomes available.
This work was previously published in Handbook of Research on E-Transformation and Human Resources Management Tech-
nologies: Organizational Outcomes and Challenges, edited by Tanya Bondarouk, Huub Ruel, Karine Guiderdoni-Jourdain and
Ewan Oiry, pp. 1-19, copyright 2009 by Information Science Reference (an imprint of IGI Global).
266
267
Chapter 1.19
Mix, Match, Rediscovery:
A Mashup Experiment of Knowledge
Organization in an Enterprise
Environment
Justin Meza
HP Labs, USA
Qin Zhu
HP Labs, USA
AbstrAct playing HP Labs Technical Reports. The findings

from our project will be included and some best
Knowledge is the fact or knowing something practices for creating enterprise mashups will be
from experience or via association. Knowledge given. The future of enterprise mashups will be
organization is the systematic management and discussed as well.
organization of knowledge (Hodge, 2000). With
the advent of Web 2.0, Mashups have become a
hot new thing on the Web. A mashup is a Web IntroductIon
site or a Web application that combines content
from more than one source and delivers it in an Knowledge is knowing facts, information, or
integrated way (Fichter, 2006). In this article, we knowing how to do something from experience
will first explore the concept of mashups and look or via association. Knowledge organization is
at the components of a mashup. We will provide the systematic management and organization of
an overview of various mashups on the Internet. knowledge (Hodge, 2000). The new technology
We will look at literature about knowledge and the of mashup can be used to organize, harvest, share,
knowledge organization. Then, we will elaborate and leverage knowledge.
on our experiment of a mashup in an enterprise With the endless development of new tech-
environment. We will describe how we mixed nologies, more and more information—in a wide
the content from two sets of sources and created variety of formats—continues to become increas-
a new source: a novel way of organizing and dis- ingly available on the Web, making it one of the
Mix, Match, Rediscovery
most information-rich infrastructures ever built. During the early stages of Web content integra-
With the advent of Web 2.0, mashups, which are tion, when the Web was still young, developers
combinations of content from different sources, were required to manually collect and hand code
have become a hot topic on the Web, gaining information into an HTML page. In information
increasing popularity. According to Program- services, the service provider may need to put
mableweb.com (http://www.programmableweb. the collected data from various sources into a
com/), there are currently 3.09 unique mashups central place such as a database to do further
being created each day, as programmers, Web data processing, such as data normalization, and
developers, and hobbyists continue to blend data to provide an indexing service. Under Web 2.0,
and services from different sources to produce O’Reilly (2005) points out the Web as platform.
entirely unique sets of content. Mashups use the Web itself as the programming
and content delivery platform. Mashups access
What is a mashup? data or information directly from Web sites and
programmatically create dynamically new Web
A mashup is a Web site or a Web application that sites or Web applications, presenting their data in
combines content from more than one source to a more integrated or useful way. Once the mashup
produce something entirely new (Fichter, 2006). process has been established, it will continue to
Sometimes the sources of the content may not automatically read, process, present, and update
be directly related. The newly created Web site the content and data collected from different
resulting from a mashup may provide something sources for the end user in the form of a Web
new or meaningful compared with the content site. Compared to the early stages of Web content
from each individual source. The word mashup integration, there is nearly no manual interven-
originated in the music industry, where it referred tion at any stage of the process. Similarly, in the
to the mixing and blending of two or more tracks information services field, there has traditionally
of recorded music or instruments in order to cre- never existed a central place to aggregate and
ate an entirely new sound track. The concept of normalize data from different sources.
mashups on the Web is not new. Since the start of To begin creating a mashup, there are three
the Web, people have been gathering information necessary components: content or data sources,
from various sources and mashing it together with a mashup algorithm or process, and a presenta-
HTML code to present it in a newer, more useful tion platform.
form: Web pages. In information services and the
field of knowledge organization, the concept of • Content and data sources:These are the
mashups is not new either. In these fields there foundation of any mashup; the raw objects
has historically been a tradition of information the mashup will work with. Without con-
service providers and knowledge organization tent and data, a mashup has no definition.
vendors working with various sources, such as Content and data may be obtained via APIs,
publishers, to collect publication and citation Web feeds, or screen scraping techniques.
information, combine the collected data, and Recently, many providers have taken steps
provide abstracts and indexing services. From the to make their content and data more read-
creation of abstracts and the population of indices, ily accessible. Some have developed ap-
end users are able to find not only publications plication programming interfaces (APIs)
from a specific source, but also publications from to give developers access to their content
a specific area, by a specific author, and so forth. and data via Web protocols such as REST
(Representational State Transfer) and also
268
through Web services. Others provide Web mashup include the reduced overhead and
feeds (simple XML documents for content more seamless user experience. Deploying
syndication) in popular formats such as RSS mashups using a combination of client and
or Atom as a means of accessing their content server techniques allows for the greatest
and data. Through the use of APIs and Web flexibility by melding reduced overhead
feeds, developers are given access to content and interactive user experience on the cli-
and data from the system backend and can ent side with the capability of performing
gather and retrieve this data programmati- complicated data retrieval and manipulation
cally. But not all data sources provide such on the server side.
means of programmatically accessing their • Presentation platform: This is where end
content. To get content from these provid- users will see the finished product of a
ers, a technique known as screen scraping mashup and interact with mashed up con-
must be deployed. Screen scraping is the tent. Often, the presentation platform itself
process of retrieving a raw set of mixed will interact with the end user, execute the
data and extracting and formatting it to a mashup process, and then generate mashed
data structure of the developer’s liking to up content—all in one—as is the case with
eventually be processed and mashed up. a client side mashup.
Comparatively, APIs and Web feeds pro-
vide a more reliable source of information,
as they are provided directly by the data’s mAsHuPs on tHe Web
originator. Screen scraping is more or less
a passive mode of data retrieval, adapting With the rapid growth of mashups on the Internet,
to the layout and content of its source Web most can be classified under the following genres:
site. Most modern Web sites tend to change map mashups, photo and video mashups, search
frequently and employ complicated layouts mashups, and news mashups.
in order to keep up with changing artistic A large number of mashups present today
trends, making screen scraping a difficult are map mashups. This may be due in part to the
process to maintain for any extended amount readily available set of API tools and API access
of time. to online map service providers such as Google
• Mashup algorithm or process: This is Maps, Yahoo Maps, Microsoft Virtual Earth, and
where all the magic happens, and where the AOL MapQuest. These tools make it possible for
mashup gets its name: during this design step, programmers, Web developers, and hobbyists
content and data from different sources are to easily mash up data and content with online
cohesively and seamlessly integrated. The maps. Map mashup developers must first obtain
mashup process may reside on the server or geographic information, such as latitude and lon-
on the client side of a given platform or be a gitude, for their data set and then mash the data
combination of the two. On the server side, up into a customized online map. One example
mashups can be deployed using server side of a large-scale map mashup is HousingMaps.
scripting techniques such as CGI, PHP, ASP, com, which mashes up real estate listings from
and so forth. On the client side, mashups Craigslist with Google Maps. A second example,
can be generated directly from an end user’s Chicagocrime.org, combines data from the Chi-
browser through scripting technologies such cago Police Department’s Citizen ICAM Web site
as JavaScript. Some advantages of deploy- with data from Google Maps.
ing a client side as opposed to a server side
269
When Google first launched their Map service, knowledge can be defined as a mix of framed
there existed no API to capture its data. Later, experience, values, contextual information, and
one was added, and other online services soon expert insight that provides a framework for
followed suit. In fact, there exist APIs for a variety evaluating and incorporating new experiences and
of services. Some developers mash up photo APIs information. Wilson (2002) states that knowledge
and video APIs with photo and video information. is what we know. Knowledge involves the mental
Others mash up search APIs or consumer market processes of comprehension, understanding, and
place APIs, news source Web feeds and more. learning that go on in the mind and only in the
Technologies such as RSS and Atom feeds for mind, however much they involve interaction with
news stories make it possible to have personalized the world outside the mind, and interaction with
news content delivered directly to the end user. others. From various definitions of knowledge,
In general, mashups can be put into two distinct basically knowledge means knowing facts and
categories: visual mashups and nonvisual mashups information, and knowing how to do something.
(Merrill, 2006). Visual mashups blend data and Knowledge is gained through past experience,
content from different sources and present it in a such as study or investigation.
graphical interface, producing a tangible result for Knowledge often is embedded in documents
the end user. Though nonvisual mashups may not or repositories and in organizational routines,
be as visually appealing, they can nevertheless be processes, practices, and norms. Knowledge is
used to create an integration point and be further also about meaning in the sense that it is context-
utilized for business process, data integration, and specific (Huber, Davenport, & King, 1998).
content integration. How can knowledge be accessible and usable?
Increasingly, individuals and companies are Knowledge organization addresses this question
realizing the potential for use of mashups on a busi- to some extent.
ness intranet or within an enterprise environment. What is knowledge organization? Knowledge
Indeed, although mashups can be creatively used organization is the management and organization
in an amateur realm by developers and hobbyists, of knowledge in a systematic way in order to sup-
they have distinct benefits for businesses as well, port easier access to and retrieval of knowledge
especially in the areas of business process, data (Hodge, 2000). It also encompasses creating,
integration, and content integration. In this article, sharing, and leveraging knowledge. Knowledge
we will describe our experiment with a mashup organization enables the management of a col-
for knowledge organization under an enterprise lection of materials, information, and documents,
environment. which represent recorded knowledge. It forms a
bridge between the user’s information need and
knowledge and the material in the collection. As the user browses,
knowledge organization searches, and navigates, knowledge organization
guides the user toward discovery. Knowledge
What is knowledge? The Random House Diction- organization also helps to answer questions about
ary of the English Language defines knowledge the scope of a collection (Hodge, 2000).
as “acquaintance with facts, truths, or principles, Knowledge Organization has become a wide
as from study or investigation.” Webster’s New interdisciplinary field. It has been broader than
Collegiate Dictionary defines knowledge as “the Library and Information Science focused. Ac-
fact or condition of knowing something with cording to Hjerland (2003), there are also social
familiarity gained through experience or associa- and intellectual forms of knowledge organiza-
tion.” According to Davenport and Prusak (1998), tion. Social forms of knowledge organization are
270
related to professional training, disciplines, and process the information, and exchange the results
social groups. The social perspective includes the with other programs. Thus, it will be a universal
social systems of genres and documents and the linked Web for knowledge exchange. Semantic
social system of knowledge producers, knowl- Web as a structure for programmatically access
edge intermediaries, and knowledge users. The and process will open up the knowledge and work-
social system of documents, genres, and agents ings of human kind to the next level. Potentially,
makes available a very complicated structure of it can assist the evolution of human knowledge.
potential subject access points, which may be From the review of knowledge organization,
used in information retrieval. This reflects what we can see, knowledge organization helps the
Huber et al. (1998) describe about knowledge: it user harvest and share knowledge. And finally,
becomes embedded in organizational norms and knowledge organization assists the reuse and
is context-sensitive. leveraging of knowledge to create new knowl-
Both Hodge (2000) and Hjerland (2003) men- edge and stimulate innovation. By exploring and
tion semantic approaches to knowledge organiza- getting familiar with existing knowledge, people
tion. Hodge (2000) states semantic networks and will have more chances to create new knowledge
ontologies as schemes for knowledge organiza- and innovation. Just like what Koestler (1964)
tion. Hjerland (2003) describes the basic unit in describes: “ uncovers, selects, re-shuffles, com-
knowledge organization is the semantic relation bines, synthesizes already existing facts, ideas,
between two concepts. But how to decide how faculties, skills. The more familiar the parts, the
things are grouped together is not an easy question. more striking the new whole.”
This leads to the points brought up by Wilde
(2006), that in any complex environment, it is our experiment
unlikely that it is possible to design and build one
unified knowledge organization system that will In our experiment, we combined HP Labs techni-
meet all requirements and be used by all people. cal report information with HP’s intranet People
Instead, systems often evolve out of practical Finder, and created a mashup: the Technical Re-
needs. He points out that the constant changes port Contextual Viewer. With this combination,
make it challenging to formalize and organize we were able to put recently published technical
knowledge. It will be hard to redesign the system reports under the lab and departmental organiza-
to better reflect the changes once the knowledge tional structure with a tag cloud section for each
organization system and the underlying formaliza- level of the organizational tree based on keyword
tion get bigger and more complex. He proposes a citing frequency within the technical reports. By
mashup model, loosely coupled distributed system scanning the tag cloud, end users are provided with
architecture, for knowledge organization. He also a sense of recent research and developments in
introduces Semantic Web technologies for shared labs and departments, a novel way of organizing
understanding of the different domains covered by and displaying HP Labs technical reports within
the different components for knowledge organiza- an enterprise environment.
tion. The Semantic Web was first thought up by
Tim Berners-Lee. According to Berners-Lee and motivation
Lassila (2001), for the Semantic Web, computers
can access to structured collections of information Initially, this project was begun out of the desire
and sets of rules to be used for automated reason- to allow easier, and more importantly quicker, ac-
ing. By this, people can create many programs cess to already-available data without the hassle of
to collect Web content from diverse sources, processing that data by hand. To us, this meant at
271
the very least providing a larger view of the picture, rectory. It allows employees to search for other
but we extended this goal to include a mechanism employees’ contact information while connected
which allowed our end users to narrow down their to the company intranet. People Finder is a Web
queries by sifting through an already optimized application consisting of a simple text box used for
and established infrastructure: the organizational inputting employee search terms that upon data
tree. Again, like in most mashups, the focus was entry dynamically changes to a list box contain-
on harvesting data which was already available ing possible hits from the Enterprise Directory.
for our use but was just not yet connected in a Bearing an intuitive interface, People Finder can
meaningful way. be accessed only from within the internal Web.
Data within the Enterprise Directory is stored
technical reports and the in LDIF (LDAP Data Interchange Format), which
technical report rss Feed presents entries as a group of labeled attributes
and separates entries with blank lines. To query
HP Labs technical reports are research papers, the database, one needs only perform a simple
reports, and articles published by HP Labs. HTTP call to the address of the database, speci-
They provide a means of letting researchers ac- fying which constraints are to be placed on the
cess various research developments, keep pace query results.
with the progress of research, and collaborate The HP Enterprise Directory is available for
on projects internally. They are high-quality, company-wide use and even has documentation
scholarly research papers. Externally available and source code available for sample queries in
technical reports are recognized and indexed by various languages. These important resources
major information services and search engines, are what made People Finder and our Technical
such as Thomson Scientific, Elsevier, Google, Report Viewer possible.
and so forth. Technical reports are managed by
the technical report publication department, who linking It All together
also provides a technical report RSS feed for re-
cent technical reports. The technical report RSS In any mashup application, it is essential that
feed updates researchers with recently published all forms of data speak harmoniously with one
technical reports and provides an alternative way another; often this requires a translator of sorts
for researchers to keep pace with the progress to ensure all forms of data can interact and work
of their research and to collaborate on various together. This translator most often comes in the
projects internally. form of a programming language. The target pro-
gramming language for a mashup is what takes
the various mixed forms of data and translates
tHe enterPrIse dIrectory them into structures which can be manipulated
within the language.
The HP Enterprise Directory is a large internally In our application, we simply took data from the
stored and managed LDAP (Lightweight Data user pertaining to the scope of data to a view based
Access Protocol) database used to store basic on their input from a Web page form, retrieved,
employee information. This information can later extracted, and organized the RSS feed data, and
be retrieved through simple calls to the database queried the LDAP database, retrieving a dataset
from a desktop application, a Web application, containing entries pertaining to the end user’s
or manually through a browser. People Finder scope. When all the data to be mashed up had been
is the frontend interface to the Enterprise Di- extracted, we identified the overlap between our
272
Figure 1. Workflow for a generalized mashup
Data
Source A
Mashup Utility: Generated

retrieve, extract, Content
organize, combine
Data Source User Input

B
two datasets: the names of authors. Our LDAP database stored variations of employees’ names,
query returned a list of employee names under a allowing us to have a high rate of success when
particular branch of the organizational structure, locating their corresponding tech reports.
and the HP Labs technical report RSS feed con- In the instance where an author publishes a
tained the relevant technical reports associated technical report under a completely different name
with those names. By selectively displaying only from the forms listed in the Enterprise Directory,
those technical reports from the RSS feed that the Technical Report Viewer degrades gracefully:
corresponded to our LDAP queries, we were able it notifies the end user of the problem and presents
to deliver the desired content to our end users. alternative options for tracking down the requested
Mashups by their very nature must be adaptive information. Disadvantages to using data sources
to their data. While it would be nice to design a without specific APIs include the possibility of
mashup solely with the end task in mind, it’s often on-the-fly format changes. Mashups which are
necessary to meet halfway between what the data tactful enough to anticipate these changes and
present is able to offer and what the designer has degrade gracefully to help the user work through
in mind to accomplish. In our own mashup, extra them instead of merely crashing are the norm in
caution was required when dealing with multiple current technology. Figure 1 is a flow chart il-
forms of the names used as keys to match authors lustrating a high-level mashup description.
in the LDAP database to their corresponding In the figure above, the mashup utility gathers
technical reports in the RSS feed. Sometimes data from two sources, extracts the meaningful
authors will publish works under a different name entries, organizes them so they are easier to work
from that which they use at work. In our case, we with, and finally combines them in some fashion.
were lucky that the Enterprise Directory LDAP Content is then generated for the end user with an
273
Figure 2. Sample breadcrumb navigation and gathered LDAP data. Note that all breadcrumb links and
researcher data is generated at page access time through inference and user input.
optional process of allowing users to customize the mat below for easy navigation and traversal; if any
data presented to them by querying a different data of these in turn manage employees, that is noted.
set and running the mashup on the new results. It is important to understand that all information
pertaining to the directory structure is inferred
from the Enterprise Directory and is generated
our mAsHuP: on the fly, making it robust against changes in the
tHe tecHnIcAl rePort organizational structure.
contextuAl vIeWer Then the technical reports for the current scope
are presented with links to their authors and the
Upon navigating to the Technical Report Con- paper’s abstracts (links to the papers themselves
textual Viewer frontend, the user is shown recent can be accessed from their respective abstracts).
technical report information at a lab-wide scope, These are organized with the most recent techni-
the largest view possible. At the top of the ap- cal reports appearing first.
plication are two breadcrumb navigation views
showing a linked trail of where the user has When a user navigates to more specific scopes,
traveled, both from a departmental standpoint a tag cloud appears underneath the technical report
as well as by the names of employees and their listings, featuring in weighted font sized relevant
managers. Breadcrumb navigations show a user keywords from the current abstracts. Clicking on
where they have been and how they got to a page one of the tags presents the user with a view of
by creating a trail of linked words denoting the technical report abstracts containing the chosen
sequential order of page accesses for a single ses- keyword.
sion, and can be especially useful in hierarchically
organized Web sites. keyword tag cloud
Underneath the breadcrumb navigation is the
name and a link to relevant contact information for Tag clouds are a stereotypically classified Web 2.0
the employee where the Technical Report Viewer’s concept, although the notion they encapsulate has
scope is rooted. If the employee manages other been around for a much longer time. The idea of
employees, their names are listed in a tabular for- organizing data in such a manner that the most
Figure 3. Keyword tag clouds provide both an easy-to-see glimpse of trends and links to related data.
274
relevant portions are more visually accessible has access information? Second is Web accessibility:
always been a key design focus in many media. Is the tag cloud compliant with Web accessibil-
Tag clouds bring this concept to the digital realm, ity standards? Third is how to calculate the term
and do so in a dynamic fashion. frequency. Fourth is how to set the font size range
Several general paradigms may be presented for different term frequencies: As the frequency
when dealing with meaningful tag clouds. changes, the font size range may need to adjust or
be set to accommodate the increase in frequency.
• Capturing keywords: In our application,
tag cloud keywords were captured by tally- our Findings
ing the number of occurrences of each word
present in all abstracts. The mashup we created solved a specific busi-
• Filtering out noise: Once all possible ness problem: quickly and easily finding recently
keywords have been tallied, meaningless published HP Labs technical reports under a
ones (such as articles and prepositions) must lab and departmental organizational structure,
necessarily be sifted out. and providing a way to quickly scan the topics
• Choosing candidates: After noisy data of technical reports as well. One of the primary
has been filtered out, what remains is the reasons for creating enterprise mashups is to solve
keyword dataset, from which it is necessary a specific business problem. Because there was
to choose the N words with the highest hit already data from People Finder at the intranet
counts. level and data from the technical report publica-
• Formatting the tags: In order to present tions system, there was no need to put data from
the tag cloud information in a meaningful these two sources in a third place. The Technical
way, it is often helpful to scale the words Report Contextual Viewer mashup was built to
relative to their frequency of appearance. combine the data from these two sources. Our
This can often be performed by generating mashup was created for data integration and
an absolute weight for each keyword based content integration, two of the key areas of benefit
on how it ranks with every other keyword for enterprise mashups. The Technical Report
and then normalizing this number with Contextual Viewer mashup is a lightweight in-
respect to how it should appear within the tegration, as it relies on data from People Finder
presentation platform (in most cases, a Web and the technical report RSS feed. There is an
browser). intrinsic serviced dependency. In addition, the
Technical Report Contextual Viewer mashup ex-
The tag cloud provides a unique view of the ists only as an add-on to existing technical report
data encapsulated within technical reports. By data to provide a different view; the mashup will
glancing over the tag cloud, one can easily spot not replace the regular technical report process
the most frequently cited abstract keywords. or delivery of data.
Weighting the font size of keywords gives end Our experiment indicates that creating an
users a sense of recent research and developments enterprise mashup might involve the following
in labs and departments, especially key areas. process: gathering data from different sources,
Tag clouds provide a useful kind of visual identifying the match point or the overlap area
display for information and data. There are a few for mashing up, processing the data and generat-
open-ended questions to consider when generating mashed content, and presenting a generated
ing tag clouds. First is usability: Is the tag cloud a mashed content output, as displayed in the chart
user-friendly feature? Does it help users find and below:
275
We also identified some best practices for the • Process data and generate mashed con-
process of creating a mashup, especially within tent: As mentioned in the preceding stage,
an enterprise environment. sometimes you may need to do further data
processing and manipulation. You may need
• Gather data from different sources: There to think about how to leverage the existing
is a huge amount of information on the In- services and interfaces for the data. You
ternet. There is a fair amount of information may consider how to handle the data process
on an intranet as well. During the mashup and where to put the data process, such as
preparation stage, one must analyze the on the client side or on the server side. You
sources available and find out which ones are may think about the scalability for future
mashable, easy to extract data and content growth of the mashup. Another consider-
from, and which ones may need a lot of work ation is keeping the mashup as a lightweight
to retrieve the relevant data. integration. You can lower the overhead of
Remember to keep in mind the motiva- IT and business process by eliminating any
tion for creating a mashup: to solve some unnecessary data processing and integration.
business problem or issue. Narrow your Finally, some internal data may be restricted
scope of data sources to those within or confidential. In that case, you will need to
your enterprise purpose, making the pro- consider how to blend and separate secured
cess of identification of data and content data. Governance and standards will come
more efficient and relevant, and in turn into play. There will be policies on who and
allowing the mashup to most succinctly how to classify data sources, who and how
serve the business process and purpose. can access the data source, and who and
Once a set of sources has been identified, how can build and view mashup products.
the owners of the sources may need to be From technology point of view, some se-
contacted for permission. Even if a service is curity architecture should be deployed and
open for internal use, it might still be a good implemented for user access and resource
idea to notify its owner of its intended use. crawling.
If the service was to change in any way in • Present generated mashed content: Once
the future, its owner might take the initiative you have the mashed up content, you need
to notify the developer of the mashup. For to present the content in a user-friendly
an external service provider such as one on way. If there are existing organization and
the Internet, it is always a good idea to read institution standards, you are well advised to
any applicable license agreements for data follow the standards and make the mashup
access restrictions and API documentation. more professional.
• Identify the match point or the overlap
piece for mashing up: Once the relevant To present the secured data in the mashup,
sets of data have been identified, a common you need to have a mechanism to check the user’s
match point or overlap piece must be found credentials.
to link them together. Sometimes, this match
point might not be obvious or some data
processing might need to be done to enable Future develoPment
one to blend and mix the data to create a
more integrated and seamless mashup. With the further development of mashups, indi-
viduals and companies are increasingly realizing
276
the promise and potential use of this technology. age existing knowledge to create new knowledge
Some companies are working on tools to make and stimulate innovation.
it easier to develop mashups, especially for non-
programmers. As mentioned before, enterprise
mashups are primary intended to solve certain reFerences
business problems. As business users and domain
experts understand their needs better than do
programmers, the tool will help them to create Berners-Lee, T., & Lassila, O. (2001, May). The
mashups and instant applications to solve their Semantic Web. Scientific American Magazine,
business problems. More and more mashup tools 284, 34-43.
will be developed and deployed in the enterprise Daveport, T. H., & Prusak, L. (1998). Working
environment. knowledge. Boston: Harvard Business School
As for the role of mashups in the enterprise Press.
environment, we can foresee more innovative and
creative uses of mashups to lower IT infrastructure Fichter, D. (2006). Doing the monster mashup.
Online, 30(4), 48-50.
and business process costs, and to solve business
problems. A number of enterprise mashup use Goodman, E., & Moed, A. (2006). Community in
cases are presented here: A mashup combines mashups: The case of personal geodata. Retrieved
a knowledge system—such as publications, re- August 12, 2007, from http://mashworks.net/im-
ports, and documents—with an expert system ages/5/59/Goodman_Moed_2006.pdf
to create a personalized knowledge system or Gruman, G. (2006). Enterprise mashups.
an expert-based knowledge system. A mashup InfoWorld, 28(31). Retrieved August 12,
interlays publications, reports, and documents 2007, from http://www.infoworld.com/
with organizational structure or project teams to article/06/07/28/31FEmashup_1.html
create an organization portfolio or project port-
Hjerland, B. (2003). Fundamentals of knowledge
folio. A mashup blends sale data with sales staff
organization. Knowledge Organization, 30(2),
information to provide further analysis of the sales
87-111.
force. A mashup mixes sale data with geographic
information and product information to provide Hodge, G. (2000). Systems of knowledge orga-
a view of the market per product and per region, nization for digital libraries: Beyond traditional
helping the company to form and improve its authority files. Washington, DC: The Digital
marketing strategy. A mashup combines supply Library Federation.
requests with business units and product lines to Hof, R. D. (2005, July 25). Mix, match, and mu-
provide further analysis of produce process and tate. Business Week, 3942,72.
production efficiency. The list can go on.
All in all, the mashup is an old concept Huber, G. P., Davenport, T. H., & King, D.
(1998). Some perspectives on organizational
equipped with new technologies under the new
memory. Task Force on Organizational Memory.
Web 2.0 environment. More and more enterprise
In F. Burstein, G. Huber, M. Mandviwalla, J.
mashups will be developed to solve business prob-
Morrison, & L. Olfman (Eds.), Presented at the
lems and help business process. Using mashups to 31st Annual Hawaii International Conference on
organize knowledge will help put knowledge in System Sciences.
context, harvest knowledge, and share knowledge.
Therefore, it will help reuse knowledge and lever- Jhingran, A. (2006). Enterprise information mash-
ups: Integrating information, simply. VLDB’06,
32, 3-4.
277
Koestler, A. (1964). The act of creation. London: O’Reilly, T. (2005). What Is Web 2.0? Retrieved
Arkana, Penguin Group. August 12, 2007 from http://www.oreillynet.
com/pub/a/oreilly/tim/news/2005/09/30/what-
Linthicum, D. (2007). Enterprise mashups
is-web-20.html
meet SOA. InfoWorld, 29(8). Retrieved August
12, 2007, from http://www.infoworld.com/ Sage. A. P. (2003). Knowledge management. In
article/07/02/19/08FEsaassoa_1.html Access Science. Retrieved August 12, 2007, from
http://dx.doi.org/10.1036/1097-8542.YB030820
Marks, P. (2006). Mashup Web sites are a dream
come true for hackers. New Scientist, 190(2551), Smiragila, R. P. (2002). The progress of theory in
28-30. knowledge organization. Library Trends, 50(3),
330.
Merrill, D. (2006). Mashups: The new breed
of Web app. Retrieved August 12, 2007 from Wilde, E. (2006). Knowledge organization mash-
http://www-128.ibm.com/developerworks/xml/ ups. Retrieved August 12, 2007 from http://dret.
library/x-mashups.html net/netdret/docs/wilde-tikrep245.pdf
Mou, Y., Zhang. S., & Cao. J. (2004). Providing Wilson, T. (2002). The nonsense of “knowledge
knowledge support in business process: A context management.” Information Research, 8(1). Re-
based approach. In Proceedings of the 2004 IEEE trieved August 12, 2007, from http://informationr.
International Conference on Systems, Man and net/ir/8-1/paper144.html
Cybernetics, (Vol. 3, Issues 10-13, pp. 2143-2149).
O’Brien, D., & Fitzgerald, B. (2006). Mashups,
remixes and copyright law. Internet Law Bulletin,
9(2), 17-19.
This work was previously published in International Journal of Knowledge Management (IJKM), edited by Murray E. Jennex,
pp. 65-76, copyright 2008 by Information Science Reference (an imprint of IGI Global).
278
279
Chapter 1.20
Testing Guidelines for
Developing Quality EAI Projects
S.R.Balasundaram
National Institute of Technology, Tiruchirappalli, India
B.Ramadoss
National Institute of Technology, Tiruchirappalli, India
“The greatest challenge to I/S in every large enter- cess of EAI projects depends on so many param-
prises finding better and simpler ways of making eters, ‘testing’ is the most significant phase that
application systems work together more effectively” can ensure the quality as well as the success of EAI
- Gartner Group projects. Components integrated without testing in
EAI systems may affect the enterprise system as a
whole. This chapter focuses on the testing aspects
AbstrAct related to EAI applications. Especially the signifi-
cance of testing for various types of “Integrations”
The rapidly changing nature of business environ- is discussed in detail.
ments requires organizations to be more flexible
to gain competitive advantages. Organizations are IntroductIon
turning into a new generation of software called
Enterprise Application Integration (EAI) to fully Enterprises across the globe have come to the
integrate business processes. It is an activity that realization that within the increasingly complex
integrates and harmonizes an enterprise’s isolated environment of the business world, there is a funda-
business applications, processes and functions mental need for many varied information platforms
involving real time data. Developing quality EAI to be properly enabled and optimized. To possess
projects is quite a big challenge. Even though suc- the state of the art business processes, enterprises
need to go for devising newer approaches to benefit
DOI: 10.4018/978-1-61520-625-4.ch012 the maximum from the underlying Information
Testing Guidelines for Developing Quality EAI Projects
Technology (IT) (Kumar & Mota, 2009) world. namely the individual components available in
These approaches are strongly based on various the integrated systems, the types of integrations
types of integrations possible within as well as and the environmental factors supporting these
across the enterprises. Majority of the enterprise integrations.
code deals with different types of integrations. Modern enterprises heavily rely on integrations
Integration does not stop with linking various linking systems and business processes using real
fragments of software products. Rather integration time data. Components integrated without testing
can be defined for databases, hardware, network, in EAI systems may affect the enterprise system
people, standards etc. This in turn will result in as a whole. This in turn may result in revenue loss
the following major benefits: and status degradation in the competitive edge.
To overcome this, testing becomes the mandatory
▪ Cost reduction towards expenses incurred process. Testing is an important activity in soft-
in Technologies, Administrations and ware development applications. It is the process
Operations. of determining whether a system has any error or
▪ Improved customer satisfaction not. When testing is essential for simple applica-
▪ Better and faster business decisions. tions, it becomes the most essential activity for
enterprise applications.
Earlier information systems defined point-to- This chapter focuses on the testing aspects
point interfaces from one application to all other related to EAI applications. Especially the signifi-
applications to share information. Defining these cance of testing for various types of “Integrations”
interfaces and managing them were highly dif- is discussed in detail.
ficult as well as challenging. To overcome these
problems the era of Enterprise Resource Planning
(ERP) came into existence. In general, ERP has erP And eAI
focused on the integration of various internal busi-
ness functions to provide “one-system-fits-all” Enterprise Resource Planning (ERP) systems
solution. The implementation of ERP requires a are more popular software applications that have
substantial amount of time and financial commit- emerged to help business managers to implement
ment (Inmon, 2000). As an alternate, Enterprise their business activities. The aim of ERP systems
Application Integration (EAI) automates the inte- is to integrate all data and processes of an organi-
gration process with less effort than that required zation into a unified system (Polson, 2009; Enter-
with ERP. The strength of any EAI totally lies prise, 2009). Typical ERP system will use multiple
on the levels of integration and their supporting components of computer software and hardware
features. At each and every level of integration to achieve the integration. An ERP system may
care must be taken to see whether the integration combine the functionalities of several subsystems
is possible, feasible and durable. Due to techno- of an enterprise such as Inventory Management
logical, physical and conceptual changes, these System (IMS), Marketing Management System
integrations may weaken the entire system. (MMS), Customer Management System (CMS),
Bringing harmony to the entire system by Financial Accounting System (FAS), Supply Chain
linking various entities of the system (entities Management (SCM) System, etc.
-intra as well as inter) is quite a big challenge in ERP systems have certain limitations in defin-
any EAI development scenario. Also, developing ing the complete business processes. Building ef-
successful EAI projects need lot of care to be fective ERP systems is complex and costly. While
taken related to the correctness of various entities, integrating ERP systems with legacy systems,
280
building necessary interfaces is a critical task. applications in an enterprise (Lee et.al. 2003;
At this point, various integration solutions were EAI-Enterprise, 2009). Typically, an enterprise
developed leading to Enterprise Application Inte- has existing legacy applications and databases
gration (EAI). As enterprises need to streamline and wants to continue to use them while adding
their processes to be competitive, there is a need or migrating to a new set of applications that
for integrating different information systems to exploit the internet, e-commerce and other new
make information consistent and easily accessible technologies. EAI is an emerging generation of
(Hasselbring, 2000; Laudon & Laudon, 2003). As integration software used to integrate and harmo-
a better and highly extensible step, industries are nize an enterprise’s isolated business applications,
moving towards the next generation integration processes and functions even other ERPs. It pro-
system called EAI. vides common, sharable business applications,
In the mid-1990s, EAI was introduced as an functions and services within as well as across the
alternative or supplemental technique for ERP. EAI enterprise. It is clear that industries are moving
is a business computing term which refers to the towards a new era of business computing from a
plans, methods and tools aimed at modernizing, system of point-to-point integration applications
consolidating, and coordinating the computer to a system of EAI applications (see Figure.1).
Figure 1. Migration from simple integrated solution to the level of EAI
281
Software products such as SAP, Oracle ERP, for all the legacy systems such that all the
PeopleSoft, JDEdwards, Siebel and Clarify would procurement orders would be routed through
work well individually, but possibly would create that.
“information islands”. When changes in data oc-
cur, such changes have to be reflected manually
in the related domains. This is not an efficient ap- Features of eAI Projects
proach. To cater to long term scalability solutions,
EAI paradigm came into existence (Enterprise EAI has always focused on integrating new soft-
Application, 2009). The following are some of ware modules or applications with its existing
the commonly used EAI example scenarios (see systems. Over the past decade, EAI solutions
Figure.2): provide the infrastructure and mechanisms for
extending and integrating both old and new appli-
i. Figure 2.a, represents manufacturing systems cation systems. They also provide the facility for
which are linked with EAI that provides a the internet access of an enterprise’s applications
way to communicate with external systems and help enterprises to share the information and
for sales tracking, demand forecasting, and applications. On one hand, though EAI offers lot
maintaining pricing details. of advantages to the enterprise, on the other hand
ii. Figure 2.b, EAI Seebeyond eGate software it poses technological and business challenges
is used to integrate the various individual (UML, 2009; Gorton et al. 2003).
legacy systems with the SAP-based purchase EAI solutions vary in their underlying ap-
order system. In this scenario, the SAP-based proaches and adopt different terminologies and
purchase order system acts as a data hub concepts. The diversity of the EAI approaches
Figure 2. Examples of EAI (b) scenario 2
282
has made selection of an EAI solution difficult. testing concerned with ‘integrations’ is the focus
Consequently, the more the EAI approaches, the of this chapter.
more the EAI solutions, the harder it is for an en-
terprise to find the most suitable solution Mosawil
et al. (2006). Lot of research activities are already tyPes oF InteGrAtIons
in the streamline to improve the performance of
EAI projects. Much of the research is focused An enterprise system is based on business process-
towards certain prime areas such as data mining es and data (Kumar & Mota, 2009). The designer
and reverse engineering applicable to EAI systems or developer of EAI systems must understand and
[(Davis & Aiken, 2000; Stonebraker & Hellerstein, define clearly the amount of work to be done, the
2001). Also, sufficient literature is available on duration required to complete the activity and the
areas such as common messaging middleware types of integrations available in EAI systems.
(Medvidovic, 2002; Young et al. 2002). Sauer et Mainly, EAI projects are characterized by the
al. (2003) have discussed more on activities related availability of different types of integrations with
to encapsulation and wrapping of legacy systems them. Mosawi et al. (2006) have analyzed and
in a component-based approach. But, there exists identified the following types of integrations in
very little information about research activities EAI. (see Figure.3):
on testing of EAI projects. In this regard, the
concern of this work is to concentrate on levels • Data-level EAI: This is the most common
of testing possible in EAI projects. Especially, integration approach to implement. At this
the work highlights the identification of key ar- level, backend data stores are integrated to
eas where testing activities have to be involved enable the movement of data between them.
by the developers in EAI projects. More to say, This deals with the information extraction
Figure 3. Types of integrations in EAI
283
from one database, if needed, processing can form part of the integration Mosawil et
that information and updating the same in al. (2006).
another database (Linthicum, 1999; Ruh et • Presentation level EAI: This approach
al. 2000). This approach provides access to allows creating new user interface by re-
a wide range of data and allows data to be mapping the old presentations Mosawil et
reused across other applications (Mosawil al. (2006).
et al. 2006; Vujasinovic et al. 2006). • Process level EAI: It is also called busi-
• Application Interface Level EAI: This ness method as it involves the flow of in-
refers to defining interfaces contained formation and automation of business sys-
within custom or packaged applications tems (Mosawil et al. 2006; Gerrard, 2007).
such as SAP, PeopleSoft or Baan. These It enables the coordination of decisions
interfaces are leveraged to provide access and the management of the dependencies
to business processes and information. between enterprise processes.
Using these interfaces, developers are able • Internal process level EAI: This ap-
to bring many applications together, allow- proach allows tracking the status of busi-
ing them to share business logic and infor- ness process activities internally Mosawil
mation (Mosawil et al. 2006; Linthicum, et al. (2006).
1999). • Cross-enterprise process level EAI: This
• User Interface-level EAI: This approach approach allows sharing processes among
is based on developing a user interface that multiple Business-2-Business (B2B) cor-
imitates end user actions by using screen porate entities. So it is suitable for B2B ap-
scraping or advanced terminal emulation plication integration Mosawil et al. (2006).
(Linthicum, 1999). It is the most primitive
form of EAI that integrates various appli- Problems related to eAI Projects
cations through the user interfaces. Though
not preferable, this approach is the only According to 2008 Gartner research, the global
way of integrating tasks (Kumar & Mota, revenue for the application infrastructure and
2009; Mosawil et al. 2006). middleware software market totaled more than $14
• Method level EAI: This is a more com- billion in 2007, increasing nearly 13 percent from
plicated form of application-level integra- the previous year. Enterprise service buses and
tion and is used less frequently. It involves business process management suites contributed
approaches to share the business logic to this steadfast growth. Focus on EAI project
available within the enterprise. Operations development has considerably increased in vari-
that are common to several applications ous parts of the globe including North America,
are combined to form a single application Western Europe, Japan and Latin America. On
(Linthicum, 1999; Ruh et al. 2000). the other hand, this growth may be slowed down
• Object-level EAI: It encapsulates the busi- due to several issues such as B2B integration
ness logic and data within objects. This in challenges and availability of varieties of tools
turn allows the objects to be linked togeth- and technologies, complex human factors and
er in a plug and play manner to interop- other environmental factors.
erate. This has the highest level of signifi- Though the positive aspects of EAI projects
cance for the enterprise of heterogeneous make them to be acceptable by industries, no
systems and architectures. Legacy systems doubt there exist certain pitfalls or demerits. As
can be wrapped into objects so that they the market changes and grows, enterprises often
284
struggle to keep up with the fast pace of innova- There are various forms of testing for dis-
tions (How the future, 2009). There exists a basic covering different types of faults and effects in
phenomenon that EAI projects need to wait for software. Testing may be done by the end-user to
completion of the other streams of work in order evaluate the usability or human-interaction issues
to proceed. This leads to the elongation of time of the software. Performance testing is often done
lines. Dependencies and long durations make EAI to assess the behavior of software under special
projects as large and risky, making them particular usage scenarios such as when the system is un-
targets of budget cuts in a recession (Thought der heavy stress of data loading. Ultimately, the
works, 2008). Early adapters of EAI struggled software must be functionally correct in order for
with new technologies and they lacked established it to serve its designated purpose and be accepted
processes to define the EAI architecture and man- by the user.
age project lifecycles and ongoing operations. In software development phase, planning stage
Though solutions can be provided for the above is applicable for development and testing activi-
said problems, the testing of EAI systems has not ties. A software project test plan is a document
attained much significance. that describes the objectives, scope, approach,
EAI projects also involve integrations of ERP and focus of a software testing effort. The process
projects. The software vendors have their own of preparing a test plan is a useful way to think
methodologies for testing ERP products, but through the efforts needed to validate the accept-
they focus on process flows through configured ability of a software product. The following are
components. The behavior of ERP suites are the elements of test plan:
typically controlled by configuration parameters.
Software maintenance is not involved in them. As • Establish objectives for each test phase
testing ERP systems is a time consuming and cost • Establish schedules for each test activity
involving process, needless to say the same may • Determine the availability of tools and
be expected in EAI systems also. Data migration, resources
the prime component of ERPs, may pose chal- • Establish the standards and procedures to
lenges in EAI also. be used for planning and conducting the
tests
• Reporting test results
soFtWAre testInG • Set the criteria for test completion
• Set the criteria for the success of each test
Software testing is the most important phase in any
software development projects, so that we know testing of eAI Projects
whether the project is going to be successful or it
will fail before it goes live. So software testing is a EAI methodologies are widely accepted in main
process of verifying and validating that a software stream market. But, testing which can define qual-
application works as expected and it meets the ity assurance is the most neglected component
business and technical requirements that guided of EAI (EAI Quality, 2009). It is due to the fact
its design and development. The main purpose that, EAI testing is often delayed until the end of
of testing is to find the presence of faults before the project. There is too little time available for
the customer finds them. Even if testing does not the developers to focus on complete testing of
reveal any fault it still provides confidence in the systems before the delivery of the products. More-
correctness of the program. over, defining testing strategies for EAI projects
285
are challenging ones and also much expertise is link the entities within the enterprises, internal
needed to define such strategies. For the entire or external or other components of the systems.
life span of EAI systems, not much testing tools Within the enterprise, individual subsystems can
are available. The major hurdle is that testing has be linked. Between enterprises, integrations can be
to be focused on various levels of EAI to provide defined. Business partners, consumers and service
quality systems. providers act as external components which can be
EAI systems are mainly characterized by their integrated with EAI systems. The entire system is
individual enterprises, external components and driven through numerous supporting parameters
the integrations. To provide better deliverables, such as networks, devices, technologies, etc.
the testing scenario of EAI projects should focus
on testing these components namely individual testing Individual enterprises
systems, integrations and the environment (see
Figure.4). Here, Ei represents the individual en- Every enterprise is composed of various sub
terprise. Ai, Bi and Ci are individual subsystems of systems and in turn every sub system has numer-
various enterprises. For example, {A1, A2, A3} may ous components related to data, procedures, user
correspond to {MMS, FMS, CRM}. The colored interfaces, objects etc. Testing must be performed
lines indicate the integrations that are possible to at every stage of defining, building and assembling
Figure 4. Categories of testing EAI projects
286
these individual components. Every component is presentation, and not at the actual inter-connection
tested as separate unit, in isolation, so that when it between applications and data. Special attention
is integrated to the rest of the system, no faults are has to be given for the verification of handshak-
encountered. Same way, every enterprise system ing mechanisms to ensure the proper message
must be tested on the whole before it is integrated delivery in asynchronous system to guarantee
to any other part of the entire application. delivery of message at the presentation. Business
requirements have to be thoroughly understood to
testing on Integrations verify the transactions across the interface.
In general, interface integration requires a lot
Integrations are the key factors defined in EAI of co-ordination and follow ups to be done by
projects. Testing ‘Complex Integration’ involv- the testers. To reduce the failure of EAI projects,
ing many end systems is difficult but identifying application of interface testing is highly recom-
faults at an early stage than at a later stage is less mended.
expensive. Various challenges and possibilities for
designing testing methodologies for EAI projects Method Level
related to integration levels are discussed as fol-
lows (Sauer et al. 2003; Linthicum, 1999; EAI Consider the ‘Update Product Information’
Quality, 2009; Vengayil, 2009; Linthicum, 2001): method. This method can be accessed from vari-
ous applications and there is no need to rewrite
Data Level each method within the respective application.
Since all applications interact with the integrated
An EAI project allows data to be reused across applications through single front end application,
various applications. This limits the ‘real time’ method-level integration requires the integrated
transactional capabilities. EAI solutions depend applications to support a Remote Procedure Call
on asynchronous messages. This in turn poses new (RPC). Testing must be focused on the integrated
problems (concurrency problems and temporal application API which will break the front appli-
conditions). The development/test team must have cation components and the applications that are
complete knowledge about messaging standards coupled with them.
technologies such as XML, Middleware brokers, Verification of failover mechanisms and re-
TCP/IP etc. covery mechanisms are of more important in case
of integration projects. It is almost impossible in
Interface Level a large integration project to orchestrate end-to-
end tests with application test tools. Performance
Application Interface: An EAI solution may be tests defined earlier in the development phase
expected across multiple languages and platforms. may help to avoid the risks associated though
That is, EAI may exist in a heterogeneous environ- this integrations.
ment. Not much standards available for creating
API level integrations across applications with Object Level
varying languages and protocols.
User Interface: This approach is synchronous This object level integration allows the legacy
in its communication style, and requires the origi- system to be wrapped into objects, and participate
nal application to be online. It cannot be scaled, in the object integration. However, this approach
and thus cannot handle more than a few screen is complex and the most difficult approach. Every
interfaces at a given time. It only takes place at the object must be tested with their semantics when
287
they are extended across enterprises. That is, an Miscellaneous

object intended for one enterprise, when it is
extended for use in another application, its basic Test cases involving timers, events and re-pro-
meaning must be retained. Also, interoperabil- ducing such scenarios are complicated. Other
ity must be properly defined while transferring major hurdles in the EAI testing could include
objects from one application to other. To enable components developed in different programming
this, rules related to object level testing must be languages. Difficulties related to the following
made available. items have to be thoroughly examined and solu-
tions must be provided:
Presentation Level
• Test data generation
Presentation level is driven, especially by end • Test case generation
user actions. When the user interacts with the user • Test result comparisons
interface of a component, it will react according • Test bed identifier
to its own user interface behavior. This may result • Base lining
in some state changes which should be known by • Handling timer events
the rest of the component in the same composite • Elaborate error-prone traces
application. So that, they can update their user • Test stub
interfaces accordingly. Testing should be done
to cope up with the state changes. testing on environments
Process Level As a complete product, the EAI projects rely on

so many related factors in and around the project
When the integration solutions streamline devel- environment. These factors may be:
opment and architecture of an existing solution the
complexity, paths and permutations still remain i. Human Related: Various human related is-
and need to be tested. Testing has to be done for sues to be understood and managed while
all of the complex business logic, which will be linking the enterprises.
now under another layer of integration-ware after ii. Device Oriented: The applications may be
the streamlining process. activated through various devices such as
Inter process level: V-model testing may be PDAs, Mobiles, Handheld devices, etc.
suitable which explicitly specifies testing activities The applications should provide flexibility
associated with each phase of the development to take care of technological changes and
cycle. This inter process level requires higher upgradations.
dependency on an availability of end applications iii. Connectivity / Networks: When applications
to perform different levels of testing. across boundaries are connected, the entire
Cross-enterprise process level: A complete un- system may be defined with heterogeneous
derstanding of the business requirement is needed network models. While integrating applica-
in order to verify all the transactions across the tions involving intranets, extranets, internet,
enterprise process. proper testing guidelines have to be framed
to smoothen the information flow.
288
conclusIon Gorton, I., Thurman, D., & Thomson, J. (2003).

Next generation application integration challenges
EAI is the ongoing process of putting an infra- and new approaches. In Proceedings of 27th An-
structure in place. It is not something that we can nual International Computer Software and Appli-
buy nor is a product. It is something we build to cations Conference (COMPSAC), 585-590, IEEE.
increase the business value of the IT environ-
Hasselbring, W. (2000). Information system
ment. Though several benefits can be attained
integration. Communications of the ACM, 43(6),
through EAI projects, successful development
33–38. doi:10.1145/336460.336472
and deployment depends on so many issues. To
achieve quality as well as to define successful How the future. (2009). How the future of the
EAI projects, the designers must focus on testing EAI Market? Retrieved March 2009, from http://
aspects of the systems. In this regard, the chapter toostep.com/idea/how-the-future-of-the-eai-
highlights the importance of testing in three broad market
areas of EAI systems, namely, Individual enter-
Inmon, W. (2000). A brief history of integration.
prise level, Integration level and Environmental
eAI Journal. Retrieved from www.eaijournal.com/
level. Especially testing the integration levels of
applicationintegration/BriefHistory.asp
EAI projects is discussed in detail.
Kumar, S., & Mota, K. (2009). Enterprise ap-
plication integration. Retrieved March 2009,
reFerences from http://www.roseindia.net/eai/enterpriseap-
plicationintegration.shtml
Davis, K. H., & Aiken, P. H. (2000). Data reverse
engineering: A historical survey. In Proceedings Laudon, K. C., & Laudon, J. P. (2003). Manage-
of the Seventh Working Conference on Reverse ment information systems (8th ed.). Upper Saddle
Engineering (WCRE’00), 1095-1350, IEEE. River, NJ: Pearson Education Publications.
EAI-Enterprise. (2009). EAI-Enterprise Applica- Lee, J., Siau, K., & Hong, S. (2003). Enterprise inte-
tion Integration – Solutions from iWay software. gration with ERP and EAI. Communications of the
Retrieved April 2009, from http://www.iwaysoft- ACM, 46(2), 54–60. doi:10.1145/606272.606273
ware.com/eai-enterprise-application-integration. Linthicum, D. (1999). Enterprise Application
html Integration. Reading, MA: Addison-Wesley
Enterprise. (2009). Enterprise resource planning. Publications.
Retrieved March 2009, from http://moneyterms. Linthicum, D. (2001). B2B application integra-
co.uk/erp/ tion. Reading, MA: Addison-Wesley Publication.
Enterprise Application. (2009). Enterprise ap- Medvidovic, N. (2002). On the role of middleware
plication integration techniques. Retrieved in architecture-based software development. In
April 2009, from http://www.cs.ucl.ac.uk/staff/ Proceedings of the 14th International Conference
ucacwxe/lectures/3C05-02-03/aswe21-essay.pdf on Software Engineering and Knowledge Engi-
Gerrard, P. (2007). Test methods and tools for neering (SEKE), 299-306, New York, ACM Press.
ERP implementations. Testing: Academic and
Industrial Conference Practice and Research
Techniques – MUTATION, 40-46.
289
Mosawi1, A. A., Zhao, L., & Macaulay, L. (2006). Young, P., Chaki, N., & Berzins, V. & Luqi. (2003).
A model driven architecture for enterprise applica- Evaluation of middleware architectures in achiev-
tion integration. In Proceedings of the 39th Hawaii ing system interoperability. In Proceedingsof 14th
International Conference on System sciences. IEEE International Workshop on Rapid Systems
Prototyping, 108-116, IEEE.
Polson, J. (2008). Why you should use SAP ERP
System. Retrieved December 2008, from www.
planetarticle.net
Quality, E. A. I. (2003). EAI quality assurance
a service oriented architecture. Zetta Works. API: An Application Programming Interface
Retrieved March 2009, from http://hosteddocs. is a set of functions, procedures, methods or
ittoolbox.com/ER102904.pdf classes that an operating system, library or service
Ruh, W., Maginnis, F., & Brown, W. (2000). provides to support requests made by computer
Enterprise Application Integration. New York: programs.
John Wiley &Sons Inc., Publications. Data Mining: It is sorting through data to
identify patterns and establish relationships.
Sauer, L. D., Clay, R. L., & Armstrong, R. (2000). EAI: Enterprise Application Integration is a
Metacomponent architecture for software in- process of data and application integration tech-
teroperability. In Proceedings of International nologies which focuses on linking transactional
Conference on Software Methods and Tools (SMT), applications together, typically in real time.
75-84, IEEE. Encapsulation: It is the inclusion of one thing
Stonebraker, M., & Hellerstein, J. M. (2001). Con- within another thing so that the included thing is
tent integration for E-Business. SIGMOD Record, not apparent.
30(2), 552–560. doi:10.1145/376284.375739 Enterprise: An environment where more than
one computer is used. These computers are joined
Thought works. (2008). Thought works: Reces- by a network.
sion news Cios can use, adding value in uncertain Enterprise System: A system that supports
Times – the Agile and lean Advantage. Retrieved enterprise-wide or cross-functional requirements,
June 2008. rather than a single department or group within
UML. (2009). UML™ for EAI. Retrieved April the organization.
2009, from http://www.omg.org/docs/ad/01-09- ERP: Enterprise Resource Planning systems
17.pdf are more popular software applications that have
emerged to help business managers to implement
Vengayil, P. (2009). Quality assurance in EAI their business activities. The aim of ERP systems
projects. Wipro Technologies. Retrieved March is to integrate all data and processes of an orga-
2009, from http://hosteddocs.ittoolbox.com/ nization into a unified system
PV111805.pdf Integration Testing: Integration testing is
Vujasinovic, M., Marjanovic, Z., & Bussler, C. the phase of software testing in which individual
(2006). Data level application integration. [BPM software modules are combined and tested as a
workshops, Springer Verlag Berlin Heidelberg.]. group. It follows unit testing and precedes system
LNCS, 3812, 390–395. testing.
Legacy Systems: A computer system that has
been in operation for a long time, and whose func-
290
tions are too essential to be disrupted by upgrading TCP) developed to get data from one network
or integration with another system. device to another.
Message-Oriented Middleware: It is the term Testing: It is the process of determining
for software that connects separate systems in a whether a program has any error or not.
network by carrying and distributing messages Test Bed: An execution environment config-
between them. The messages may contain data, ured for testing. May consist of specific hardware,
software instructions, or both together. operating system, network topology, configura-
Metadata: Metadata is Information about the tion of the product under test, other application
data warehouse or data mart system. Metadata or system software, etc.
encompasses all aspects of the data warehouse Test Case: The specification (usually formal)
or data mart system, including technical, human of a set of test inputs, execution conditions, and
and data resources. expected results, identified for the purpose of
QA: Quality Assurance for short refers to making an evaluation of some particular aspect
planned and systematic production processes that of a Target Test Item.
provide confidence in a product’s suitability for Test Data: The definition (usually formal) of
its intended purpose. a collection of test input values that are consumed
Reverse Engineering: Legally sanctioned during the execution of a test, and expected results
method of copying a technology which begins referenced for comparative purposes.
with an existing product and works backward. Validation: Validation is the process of ensur-
RPC: Remote Procedure Call is a programing that a product or a process conforms to defined
ming interface that allows one program to use the user needs, requirements, and/or specifications
services of another program in a remote machine. under defined operating conditions.
The calling program sends a message and data to Verification: In the context of hardware and
the remote program, which is executed, and results software systems, formal verification is the act of
are passed back to the calling program. proving or disproving the correctness of a system
SCM: Supply Chain Management is the man- with respect to a certain formal specification or
agement of a network of interconnected businesses property, using formal methods.
involved in the ultimate provision of product and XML: It stands for eXtensible Markup Lan-
service packages required by end customers. guage. XML is the World Wide Web Consortium’s
TCP/IP: Transmission Control Protocol / (W3C) recommended standard for creating for-
Internet Protocol, a set of protocols (including mats and sharing data on the Web.
This work was previously published in Enterprise Information Systems and Implementing IT Infrastructures: Challenges and
Issues, edited by S. Parthasarathy, pp. 167-179, copyright 2010 by Information Science Reference (an imprint of IGI Global).
291
292
Chapter 1.21
Technology and Human
Resources Management
in Health Care
Stefane M. Kabene
University of Western Ontario, Canada
Lisa King
Candace J. Gibson
AbstrAct up-to-date patient care information. ICT can also

be used for distance training and education for
Health care has lagged behind most industries those facing geographic isolation and provide a
and businesses in its adoption of information and medium through which continued education can
communication technologies (ICT). Many of the be maintained for both rural and urban health
current information technologies and those to be professionals. However, due to the differences in
deployed and developed over the next few years barriers to ICT use found for each group, such as
(e.g. electronic health records, telehealth applica- computer illiteracy, geographic isolation or poor
tions, elearning technologies, social networking infrastructure, different steps need to be taken in
via Web 2.0) could be of benefit in health care order to ensure the successful implementation and
delivery and improvement of the quality, effi- use of information technologies in both urban and
ciency and effectiveness of health care services. rural communities in developed and developing
The uses of technology in human resources man- regions across the world. I
agement (HRM) can help improve the medical
care that health professionals provide to their
patients. For instance, technology can be used ntroductIon
to maximize communication, collaboration and
support between health professionals separated Human resources management (HRM) includes
by distance, as well as provide immediate and the selection and recruitment of personnel, their
retention (which is affected by compensation,
DOI: 10.4018/978-1-61520-777-0.ch027
Technology and Human Resources Management in Health Care
benefits, quality of work life, and support), their regions; ‘rural developed’ regions are those found
training and development, career progression in Westernized remote locations; ‘urban develop-
and promotion. Human resources planning and ing’ refers to city centers in non-industrialized or
effective human resources management are es- low resource/income countries; and rural areas
sential within a single institution or institutions, of the developing world would be considered to
a local region or even at the national level to meet have an increased level of poverty, poor trans-
organizational objectives, requirements and needs. portation and fewer specialized services and
Within the health care field HRM is focused on organizations to provide health care (Goldsmith
making sure that the right mix of health care pro- et al, 1995). These disparities have been attrib-
viders is in place to meet the health care needs of uted to individual, organizational, and economic
citizens, on adequate recruitment and retention (i.e. issues – the lack of coordinated national health
encouraging more people to enter the health care human resources policies being one (Dussault
field and improving working conditions to retain & Franceschini, 2006). Although not a panacea,
them), and increasingly on new paradigms that ICT presents a tool to both manage and improve
adjust the way in which health professionals are aspects of human resources including satisfaction
educated and practice in interprofessional health and retention of personnel, recruitment, train-
care teams (Kabene et al, 2006). ing, improved working conditions and access to
New information and communication needed information. Telemedicine or telehealth,
technology(ies) (ICT), and the growing use of the utilization of telecommunication technology
computers and the Internet, can address many for delivery of health services at a distance (that
of the information, communication, and training is, for diagnosis, treatment, and patient care, as
issues faced by health professionals and hu- well as training and continuing education), can
man resources managers. HRM itself relies on “provide expert-based health care to understaffed
information systems and information needs. All remote sites and to provide advanced emergency
of the activities in health HR management rely care through modern telecommunication and in-
on the existence of information systems that can formation technologies” (Lin, 1999, p. 28).
provide high-quality data that are timely, accurate, Information technologies encompass any tech-
comparable, accessible and relevant. Priority in- nology which processes and communicates data. It
formation areas for HRM are information on the includes: computers, voice, data and image sensing
demographics of the workforce, their education/ and communications devices, graphics devices,
training, geographical distribution, migration or multi-media storage, et cetera. ICT hardware
non migration-related attrition, and employment or devices include those used for sensing (e.g. bar
practice characteristics (Dussault & Franceschini, code scanners, keyboards, mice), communication
2006; Henderson & Tulloch, 2008; Hongoro & (fax, cellular phones, local or wide area networks
McPake, 2004; Wibulpolprasert & Pengpaibon, (LANs and WANs)), analysis (computers - mi-
2003). The focus of this discussion though relates cros, minis, mainframes), and display (monitors,
to the use of ICT to assist in the training, recruit- printers, voice output, television) and the soft-
ment and retention of health professionals. ware applications (operating systems, end-user
Additionally an unbalanced distribution of applications) needed to provide theses services.
health professionals both within countries (urban In analyzing the HRM literature we have con-
versus rural) and between countries (developed centrated on looking at those areas where informa-
versus developing) has been recognized as a wide- tion technology can reinforce and improve some
spread global problem. In this discussion ‘urban of the resource management issues, e.g. increased
developed’ refers to city centers in Westernized retention of health professionals through improved
293
working conditions by providing health informa- The term “information needs” refers to the
tion where and when it is needed, providing clinical knowledge required for health care profession-
guidance and support, improved training and edu- als “to carry out patient care and professional
cational opportunities (both initial training and in duties” (Dorsch, 2000, p. 347). According to
continuing professional education), and improved Westberg and Miller (1998), information needs
geographic distribution by providing support for in clinical practice fall on two axes: formal to
rural physicians and professionals in low income informal and general to specific. An example of
countries in the developing world. This review of formal information is peer reviewed literature
the effects of these telecommunication services from controlled scientific studies. Examples of
and current status and use among health care pro- informal information would be unwritten, common
fessionals highlights and emphasizes HRM’s use practices followed at a clinic. Knowledge that is
of technology in health care and suggests where widely available and applies to several categories
improvements can be made. of patients is general knowledge; while findings
in a patient’s chart, for example, would be local
or specific knowledge.
tHe IntersectIon oF Ict In terms of access to information resources all
And Hrm In HeAltH cAre health professionals can be divided into the fol-
lowing three groups (Dorsch, 2000): the served,
Ict and Hrm Issues in urban- underserved, and the unserved. Common among
developed regions all groups is that information is underused, all
face barriers to optimal information use, and all
The practice of medicine is an information-based, prefer referring to colleagues or personal library
knowledge intensive business. Experienced physi- collections over bibliographical sources for
cians use about two million bits of information to acquiring information. Moreover, “information
manage their patients (Smith, 1996). It is estimated literacy is the ability to recognize the need for
in the US that about a third of a doctor’s time is information; to find, organize, evaluate and use
spent recording and synthesizing information and such information for effective decision making
about a third of hospital costs are spent on personal or problem solving; and to apply these skills to
and professional communications (Smith, 1996). independent, life-long learning” (Sheppard and
With an increasing number of medical journals Mackintosh, 1998, p. 192).
and articles (for example over 14 million citations For both urban and rural health care profes-
and abstracts to articles in over 5,000 life sciences sionals, the primary reasons for seeking infor-
journals appear in PubMed/Medline with about mation are related to patient health care. In the
10,000 to 20,000 added weekly) (MEDLINE – ordinary course of patient care family doctors ask
Fact Sheet, 2007); over 5,000 systematic reviews a number of questions dealing with diagnosis,
in the Cochrane Collaboration library database treatment and management of diseases. The most
(Cochrane Collaboration – http://www.cochrane. frequently asked questions relate to the causes of
org), and over 20,000 prescription drugs (e.g. symptoms, what tests may be indicated, doses of
Health Canada – Drug Product Database - http:// drugs, and how a disease should be managed or
www.hc-sc.gc.ca/dhp-mps/prodpharma/data- treated (Dorsch, 2000; Ely et al, 1999). Answers
basdon/index-eng.php) it is impossible for the to most of these questions (64%), with the ex-
average physician to keep that information “in ception of drug prescribing, are not immediately
their heads” to effectively diagnose, prescribe pursued and for those that are, doctors spent an
and treat patients. average of less than 2 minutes pursuing an answer.
294
Their preference for resource material to answer 5. automated inventory systems to track and
these questions was readily available print or text manage pharmaceuticals and other medical
sources and human resources (a nearby colleague) and general hospital supplies;
(Dorsch, 2000; Ely et al, 1999). Few questions 6. computerized provider order entry systems
generated either a computer-based search on the for laboratory tests, diagnostic images, drug
Internet or with a CD-ROM resource. Previous prescribing;
studies have shown that most computer applica- 7. decision support systems that will give
tions were too slow or cumbersome to use at the health care providers real-time advice on
bedside or point of care. diagnosis and treatment options based on
During the next decade, we will undoubtedly continuously updated information and best
see the development and deployment of compre- evidence (Anvari, 2007; Westberg & Miller,
hensive electronic health information systems 1999).
incorporating all aspects of in-hospital and out-
of-hospital care. These systems will allow health Use of new electronic record systems will
providers at all levels and across geographic improve the quality of patient care by streamlin-
distances to access information quickly and share ing processes, reducing duplication of orders and
information, improving both patient care outcomes tests, and minimizing the risk of medical errors,
and operational efficiency. Health information such as administration of the wrong drug or dos-
technologies that are being implemented and age, or even the performance of a wrong surgery.
newly emerging technologies that will have an What makes the HR responsibilities of these
impact on health care, health services delivery evolving technologies more difficult is that there
and HRM include: are as yet no hard-and-fast answers to formulate
best-practice guidelines. For example, identi-
1. electronic lifetime health records (EHR) fication cards with radio frequency identifica-
to store data from many sources (e.g., text tion devices for employees allow employers to
and voice notes, medical images, laboratory better maintain a secure workplace, but privacy
values, patient demographics), with avail- advocates say that employers can (and do) abuse
able access from any locale; RFIDs by tracking employee movements at any
2. mobile computing technologies such as or all times, which may modify potential deviant
hand-held computing devices (or personal employee behaviour (Vilamovska et al, 2009).
digital assistants, PDAs), mobile clinical This technology also has the potential to help
assistants designed for both provision and with health-care applications when used to carry
entry of relevant health information at the an employee’s medical records.
point of care (Adatia & Bedard, 2002; 2003); Case studies suggest that security and public
3. picture archiving and communication sys- safety trump personal privacy; such that securing
tems (PACS) that capture, store and provide the workplace, investigating instances of theft or
access to diagnostic images (X-ray films, misconduct, and accounting for employees after
magnetic resonance images, computed to- emergencies, plus providing effective responses
mography scans) from any location; to medical problems are the priorities favoured
4. radio frequency identification (RFID) sys- in designing and operating the systems that use
tems will use radio waves to wirelessly track RFIDs (Vilamovska et al, 2009). Key human re-
hospital patients, and microchips to carry source issues regarding technology and health care
information on medications, laboratory tests, are the dual responsibilities of supervised access
imaging studies and medical devices; to personnel (and personal) data from a standpoint
295
of both database security and medical ethics (i.e. When shifts were open, a nurse manager added
monitored access for quality assurance purposes, the shifts to the website. Qualified staff hospital
outcome assessments and disease surveillance) nurses or other pre-approved independent nurses
(Anvari, 2007). bid on their preferential shifts for the best qualify-
Similarly the benefits of a fully electronic ing rate per hour; perhaps even earning more than
record in improved quality of care – increased sharing fees with an agency. The results showed
adherence to care guidelines, enhanced monitoring that for the St. Peter’s case, everybody benefited.
and surveillance, decreased medication errors and In 2003 the bidding system filled 43,400 of St.
decreased utilization of care may outweigh the Peter’s available shift-hours at an average of $37
risks of breaches of privacy and confidentiality per hour, saving the hospital nearly $750,000 over
particularly when safeguards using the technol- the course of a year (Robinson, 2003). Shift bid-
ogy itself can be built in (e.g. audit trails, levels ding and scheduling systems are now widely used
of authorization and access, consent documenta- to ease staffing shortages and result in decreased
tion) (Chaudry et al, 2006; McGraw et al, 2009). agency use and cost savings, improved employee
The benchmark studies on benefits of ICT exist recruitment and retention, reduced administrative
primarily in academic institutions in urban centres time spent scheduling and improved staffing
that have often implemented internally developed practices that optimize the use of the existing
information systems; whether or not, and when, workforce (Kulma & Springer, 2006).
other institutions with commercially available
software solutions can achieve similar benefits, Ict and Hrm Issues in rural
and at what cost are not clear. developed regions
As technology becomes more pervasive in
urban, developed centres, new found mobility Those health professionals characterized as part
will allow a greater dispersion of medical knowl- of the underserved group are primarily the rural
edge and human capital. HRM will need to adopt health professionals in developed regions. This
technology as a strategy to acquire and retain group faces significant barriers to acquiring
knowledgeable, well trained staff. information, especially when compared to their
Technology can also be used in simple yet urban counterparts (Dorsch, 2000). In this survey
innovative ways to solve human resources prob- of information needs, rural health professionals
lems. One real-life example is the following ranked what they perceived to be major barriers to
story found at St. Peter’s Health Care Services, acquiring information. They included: lack of time,
a hospital in Albany, NY. St. Peter’s typically geographic isolation, lack of a library, technology
filled its nursing requirement shortages by hiring illiteracy, lack of equipment (computers), and cost.
agency-based temporary nurses, each costing the As was found with most health professionals,
hospital approximately $54 per hour. These nurses rural health professionals preferred their col-
were seldom familiar with procedures specific to leagues or personal libraries over bibliographi-
St Peter’s, creating inefficiencies financially and cal sources as information sources; for example
procedurally (Robinson, 2003). To address this 63.1% would discuss a decision with a colleague
problem, human resources management, using before adopting clinical methods, while only 17%
an idea borrowed from Priceline.com and eBay, would search the literature for more information.
put together a team of people from information Moreover, when judging the scientific soundness
technology and patient care services to develop of medical information, 90% compared the find-
the “shift-bidding” system. Two months later, ings to their personal experiences, rather than
the hospital launched “St. Peter’s Jobs Online.” appraising “the findings based on study methodol-
296
ogy and statistical significance” (Dorsch, 2000, provides training opportunities for hospital staff
p. 347). The average age of health professionals and employees, and can enhance educational
in rural areas in this study was 46 years old, with experiences for primary care practitioners via
17 years of work experience (compared to the consultations with specialists and/or attendance
average age of urban health professionals of 38 at virtual grand rounds (Curran, 2006). The use
years old with 9 years of work experience). Of of these technologies can alleviate some of the
those who had been in practice for less than ten isolation felt by rural health professionals and
years, 80.6% owned computers equipped with CD- reduce the costs, travel time and staff absences
ROM drives, however, among those who had been associated with attendance at distant continuing
practicing for more than 30 years, the percentage education programs.
of computer ownership was only 32.4% (Dorsch,
2000). Throughout the 1990s, use of electronic Ict and Hrm Issues in urban
information resources among rural health profes- developing regions
sionals remained constant. For instance, 40% used
MEDLINE, 25.6% used CD-ROMs for literature Effective human resource practice is crucial for
searching and 44.9% for accessing medical texts. improving the efficiency of the health care sys-
More than half (52%) also used the CD-ROM for tem in developing countries. Because developing
entertainment (Dorsch, 2000). nations have the heaviest burden of disease, it is
In the field of health care, continuing education even more important to use human resources man-
is essential and necessary for all health profes- agement to increase the overall quality of health
sionals in order to ensure that quality health care care and to balance health workforce deployment
is being provided and that access to the most up- with where it is most needed (i.e. usually in rural
to-date and best medical evidence for practice and communities). It is often more difficult for people
care is available. Since health professionals prefer in developing countries and low income nations
their colleagues as an information source above all to provide effective and efficient health care
else, a better means of facilitating communication than those in industrialized countries due to the
among all health care professionals is needed. failure or inability to provide appropriate social
Due to the barriers specifically cited by rural infrastructure for health care due to limited finan-
health professionals, some technologies may be cial resources and budgets, as well as adequate
more useful than others. Those most useful are funding for training or education of health care
audioconferencing (even simple teleconferenc- workers. For example, of the approximate 8.8
ing), videoconferencing, and CD-ROMs (compact million new tuberculosis (TB) infections in 2005,
disc-read only memory) (Curran, 2006; Sheppard 7.4 million (84%) were concentrated in Asia and
& Mackintosh, 1998). The Internet provides sub-Saharan Africa. Nigeria has the highest TB
networks with access to medical databases and burden in Africa, fifth highest in the world, while
online communities of practice, as well as elec- Kyrgyzstan has the second highest tuberculosis
tronic mail (e-mail) (Coma del Coral et al, 2005; burden in Central Asia, after Kazakhstan (Awof-
Curran, 2006). Studies have also shown that, eso et al, 2008). The main human resource issues
although awareness of telemedicine’s applications affecting effective TB control are insufficient
and capabilities among health professionals in quality, quantity and distribution of health workers
rural areas is minimal, the attitudes towards its (Awofeso et al, 2008).
adoption are favourable (Lin, 1999). The deliv- From 2004 to 2007, “government annual
ery of tele-education programs via ICTs enables budgets for tuberculosis control in Nigeria and
the timely dissemination of new developments, Kyrgyzstan averaged US$ 14 million and US$
297
1 million respectively, barely enough for staff better social infrastructure. One important aspect
salaries, let alone training” (Awofeso et al, 2008, that technology can do to enhance the effectiveness
p. 2). Training/retraining programs for tuberculosis of HRM is to provide a broader scope of training.
workers may be used as an incentive to influence Virtual universities, networks of institutions and
distribution patterns; either by preferentially fund- professional associations, international standards
ing applicants from high TB prevalence regions of certification and distance learning give people
or by developing and funding distance learning in developing countries more channels to access
and on-site training programs for workers in these higher education and training. An example of
regions. These can include high tech solutions such a distance learning approach for tuberculosis
through Internet or web-based programs as well control and training of doctors and nurses was
as low tech solutions such as correspondence developed jointly by the International Council of
courses and the use of radio or TV narrow casts Nurses, the International Hospital Federation, and
(Zarocostas, 2008). the World Medical Association in 2006 (Awofeso
According to Henderson and Tulloch (2008), et al, 2008). This provides efficient and high qual-
the main causes of health care labour shortages ity tuberculosis training for health care workers
in low income countries include a lack of effec- internationally at a reduced cost.
tive planning, limited health budgets, limited E-health innovations such as electronic health
employment opportunities, low salaries, poor records, telehealth, health knowledge manage-
working conditions, weak support and supervi- ment, health information systems, and clinical
sion, and limited opportunities for professional databases have brought significant change to
development. HRM in health care in developing countries.
The lack of effective human resource practices First, the globalization of communication and the
also result in recruitment and retention problems, advances in information technology offer unprec-
or even brain drain, which make the shortages edented opportunities to improve public health
of local health care workers even worse. The worldwide. Videoconferences, telecommunica-
high demand for doctors and nurses in Western tion, and professional institution networks give
countries attract health care professionals from health care professionals in developing countries
developing countries to migrate with the promise the opportunity to discuss problems with those
of better working conditions, higher remunera- in industrialized nations, and can also create a
tion, and additional work opportunities for family medical network within their own countries. In
members. From 1985 to 1997, in Ghana, it was Thailand, for example, “the launch of the coun-
estimated that the cumulative average annual try’s first communications satellite allowed for
emigration rate was about 14%, which meant the implementation of a nationwide telemedicine
that half of medical graduates emigrated within network by the Ministry of Public Health, cur-
4.5 years, mainly to the United Kingdom and the rently linking 19 hospitals with health facilities
United States. During the same period, 25% of all over the country” (Dussault & Franceschini,
Thai doctors immigrated to the United States, and 2006, p. 10).
most of them never returned (Wibulpolprasert & Information and communication technology
Pengpaibon, 2003). can also help to address the shortage of health care
Although technology innovation may not bring professionals through international volunteering
as significant change to low income countries as programs. These programs, which are supported
to developed countries, the health care system in by large corporations, can connect with each
developing countries does benefit from this, es- other and contact governments in order to send
pecially in urban areas because of the comparably volunteer health care professionals from devel-
298
oped countries to the places where they are most very low (only 38.6% of the population can read
needed. For example, the Pfizer Global Health or write) and little of the national budget is spent
Fellows program sent 72 Pfizer employees to on education (only 2.9%). Although training fa-
work with organizations in 19 countries such as cilities to improve literacy have been attempted
Uganda, Kenya, Ghana, South Africa and India in rural Pakistan, they have largely failed due to
from 2003 to 2005, with a goal of promoting a lack of perception of the populations’ need or a
better health by “improving the service delivery clear strategy of what benefits would be gained.
capacity of local partners in poor countries” (Vian Within the rural community agriculture forms
et al, 2007, p. 2). Many of these fellows imparted the basis of the local economy. An innovative
skills or enhanced operations of non-governmental model proposing the introduction of rural kiosk
organizations in HIV/AIDS and other health pro- machines (RKM) within rural community centres
grams providing training to clinical and research and schools has been proposed to provide text,
personnel; strengthening laboratory, pharmacy, audio and video material on agricultural crops and
financial control, and human resources manage- to provide training to rural farmers in their use..
ment systems; and helped to expand organization Machines will be connected via wireless connec-
networks. Local staff also reported the program tions to a central department for regular updates
changed their work habits and attitudes (Vian et of the information. The kiosks will serve at least
al, 2007). Although transferring volunteer health two purposes, one to provide valuable information
care professionals to low income countries can- to enhance the growth and sustainability of crops
not completely solve the shortage problem, it and two to help raise the literacy levels in rural
does relieve, to some extent in the short term, the communities (Sattar, 2007). This kiosk technol-
human resource burden in developing countries. ogy could be easily adapted to provide health
information to rural communities.
Ict and Hrm Issues in rural Within rural communities shifting of tasks to
developing regions health workers with less training may provide
needed preventive health services, basic treatment
As we’ve seen over and over again, human resourc- services, and serve as a liaison between the com-
es are a crucial part of the health care industry and munity and more highly skilled workers in short
the assurance of the delivery of quality health care supply. Well designed programmes using such
with improved health outcomes; however health community health workers with proper support
professionals remain in critically short supply in and supervision have been successful and may
developing countries, particularly in rural regions even reduce the number of medical consultations
(Hongoro & McPake, 2004). One strategy to ad- and hospitalizations (Dussault & Franceschini,
dress this shortage of human resources is to make 2006). However some professional bodies may
use of the capacity within the developed world by themselves prove an impediment to this potential
the promotion of distance learning opportunities solution of task shifting. It was noted that in rural
or provision of training staff through networks Uganda many rural midwives only supervised one
or twinning of medical schools with developing birth per day, yet there was a very high need for
countries. These efforts may be hampered how- those midwives to be cross-trained as nurses to
ever by the lack of interest or funding or cultural do the regular nursing tasks involved in primary
sensitivity in the developed world. care (Hongoro &McPake, 2004). Although the
General literacy levels within a country may larger problem of providing training to these rural
be a major hurdle in recruitment and training of and often remote areas remained, there was also a
any rural workers. Literacy rates in Pakistan are level of protectionism against such cross-training
299
from professional nursing organizations in Uganda In Afghanistan, coordination of health human

(Hongoro & McPake, 2004). resources was transformed by an outstanding
The World Health Organization indicated director of Human Resources in the Ministry of
that poor human resources information systems Public Health (Schiffbauer et al, 2008). A central
hinder the proper deployment of health workers database containing information on about 24,500
to where they are needed and reliable numbers of health workers was established to identify the
properly trained workers (Awofeso et al, 2006). numbers, types and locations of health workers
Incomplete reporting of where TB cases and on the and determine their level of competency. Training
distribution of health care workers in general are is being provided to refresh the skills of health
hampers the ability to send trained health workers workers and train new workers including female
to areas of high prevalence of TB where they are doctors, nurses and midwives that are needed
most needed. Also in Nigeria where training for in rural communities (Schiffbauer et al, 2008).
front-line workers was well provided, the finan- Similarly centralized HR programs in South Africa
cial and travel incentives to be trained removed and southern Sudan have used similar strategies;
these rural workers away from their duties for a development of a comprehensive HR policy, es-
significant period of time. The adverse impact tablishment of HR units in the ministries of health
of training on availability of workers could be and training of HR managers, nationwide health
minimized by developing and funding distance HR information systems, expanded recruitment,
learning and on-site training programs. training, testing and certification of health work-
Trouble filling nursing postings as well as a miers, and revision of pre-service training curricula
gration away from rural, developing areas has also for doctors and nurses (Schiffbauer et al, 2008).
been a large problem in South Africa particularly Once human resources were moved to a
in addressing the need for large scale antiretroviral national level in Afghanistan it allowed for a
treatment programs among the large number of HIV/ consistent training basis and also allowed for a co-
AIDS cases (van Rensberg, 2008). Introduction of ordination amongst other governmental agencies.
new positions in an ART program resulted in an Across a number of countries in east and southern
increased recruitment of nurses, but often at the ex- Africa the development of HRM systems and HR
pense of other existing facilities or regions – no new information systems have improved health care
nurses were recruited, nurses simply moved from workers satisfaction and retention by being able to
one facility to another or one department to another. better manage resources and personnel and provide
Supporting health care employee’s with appropriate incentives to maintain staffing needs (Dambisya,
information systems to track HIV patients and drugs 2007). Such incentives can have a great impact,
is also critical, yet often lacking in rural, develop- for instance in Uganda such changes resulted in
ing areas due to technological limitations (Fraser et a considerable reduction in turnover and growth
al,2004).Asimple, web based medical record system in patient services by 50%.
linking remote areas in rural Haiti was used to track
clinical outcomes, laboratory tests, and drug sup-
plies. The use of open source software and satellite summAry: solutIons And
and wireless technologies linked through a central recommendAtIons
location kept costs low (Fraser et al, 2004). Health
workers were also able to conduct consultations by While some of the benefits of information and
contacting more experienced colleagues, and share communication technologies have been realized
images and efficiently manage patient information in the health care field, the full impact across
and test results. many disciplines will be more fully realized as it
300
is applied in a more systematic form, rather than Differences in information needs and prefer-
in the current patchy and uneven implementation. ences do not differ greatly from rural to urban
To date, technological advances such as electronic health professionals, rather, it is more dependent
health records are typically instituted by larger on the type of practice whether solo or group
institutions in urban centres with significant re- practice or family physician versus specialist
sources capable of aligning funding with available (Dorsch, 2000). However, there is a difference in
technological expertise. However over time this the frequency of information use in that rural health
technology will facilitate the decentralization of, professionals were found to search for answers to
and access to, expert medical equipment and staff clinical questions and use information sources less
(e.g. telerobotic surgery, telementoring or distance frequently than their urban counterparts (Dorsch,
education). Other changes, such as personnel 2000). Rural health professionals face more sig-
and shift management, as well as core medical nificant barriers for meeting their information
practices (i.e. evidence-based practice, health needs, such as accessibility issues (i.e. the lack of
care teams and interprofessional collaboration) libraries, technological illiteracy, and the lack of
are facilitated by technology. equipment). Demographic barriers may also play a
Additionally, as technology advances, the part; for instance, rural health professionals tended
types of medical care provided are likely to to be older and had been in practice longer. When
change drastically over time, such as minimally compared to urban health professionals, those in
invasive surgical interventions, gene therapy and rural areas on average had lower computer use
personalized medicine (Anravi, 2006). This will and spent less time pursuing clinical questions
have an impact on the shift of medical health care and were, therefore, less likely to use information.
professionals from one sub-specialty to another The nature of the health care profession requires
and new roles for health professionals on the health that all health professionals remain up to date on
care team. Also, this will undoubtedly impact the current methods and research findings. Therefore,
type of health care coverage HR professionals will it is crucial that rural health professionals are able
offer for their employees, and may even evolve to bypass such barriers in order to maintain their
further to define specific benefit coverage for quest for continued education. Technology in the
certain high-impact employees. form of audioconferencing (both by conventional
As the joining of technology and health care telephone and voice over IP), videoconferencing,
moves forward, patient transfers will be replaced CD-ROMs, and Web-based resources can provide
by the transfer of medical expertise from one a link to urban resources and colleagues.
health center to another, e.g. from urban centres Audioconferencing is an ideal tool for rural
to rural communities, and across even greater health professionals because it provides a simple
distances across countries from the developed and convenient means of communication between
world to the developing world. Along with this groups or individuals and its interactive nature
comes the potential accessibility of private patient provides a means to get instant information, sup-
medical information anywhere in the world. The port, and feedback (Curran, 2006; Shepherd &
implications of the information benefit versus Mackintosh, 1998). Videoconferencing provides
privacy duality are huge. Medical records perhaps the opportunity to add a visual link as well as the
represent the most personal and private informa- audio link for simultaneous communication be-
tion regarding patients or employees. As such, the tween two or more sites. It can be used to deliver
responsibility and care of these records require lectures, show practical techniques, or demonstrate
priorities of security from both a technical and clinical procedures; hence making it a great tool
ethical perspective. for continued education (Curran, 2006). However,
301
rural health professionals need to have access to In addition, e-mail allows health professionals to
a videoconferencing site or be able to access via send and receive documents quickly, efficiently,
their own computers, either of which may not be and in a cost effective way. Secure networks and/
available. Installation of computers and broadband or encryption can be used to ensure privacy and
transmission of videos can also be a financial confidentiality of transmitted health information.
burden, especially for the small practices or small E-mail lists are also beneficial to rural health
community hospitals found in rural areas. professionals because they allow one to build a
CD-ROMs are disc packages that contain larger network of colleagues and professionals
information usually found in medical texts, but for information sources and feedback. Networks,
can provide far greater interactivity than the usual such as UniNet, allow for continued medical
medical text since it can contain still and animated education; communication tools, such as video
graphics, as well as video sections. Interactive or text conferencing, and reduced costs (such as
features can also provide users with question and long distance phone calls or purchasing costs for
review sections, as well as immediate feedback acquiring books and other materials).
regarding their performance. CD-ROM packages Because developing countries often have lim-
are especially useful for rural health professionals ited budgets to invest in education and training
because they provide information that can be easily of health care professionals, distance education
received at great distances (packages can even be is a feasible option due to its comparably lower
delivered by regular mail service). One drawback cost and equivalent or higher quality. Students
is the data on CD-ROMs is not regularly updated, and trainees can access advanced knowledge
which means health professionals must continue and skills from Western countries without going
to purchase new packages in order to stay up to abroad or even leaving the community. The use of
date. The Internet and the Web browser interface information technology to provide online learning
facilitates the creation of virtual networks which can generate a larger pool of qualified health care
enable a group of users who share a common aspect workers and assist in alleviating the problem of
of knowledge (such as health care) to gather and worker shortages. However, poor infrastructure in
post relevant information or hold discussions on low income countries may still limit the applica-
specific topics. They are “universal, free of lan- tion of technology, particularly, as in the West,
guage or geographical boundaries and [are] open outside of urban centres.
to all interests” (Coma del Corral et al, 2005, p. 2). Although the advances in information and
These networks provide rural health profession- communication technology can greatly increase
als with a means to maintain communication and the efficiency of HRM in health care in developing
collaboration with their colleagues by use of chat countries, problems still exist. Brain drain is one
rooms, text conferencing, e-mail lists, and access of the leading causes of health care professional
to medical databases (e.g. MEDLINE, PubMed, shortages. While Internet-based distance studies
or online publications). provide higher education and training to people
Research shows that the use of such networks in developing countries, these individuals tend
has achieved positive changes in medical care in to leave their home country after they become
rural areas. Studies have shown a decrease in the professionals in order to find better working and
total average of hospital stays due to increased living conditions. It is difficult for governments in
access to information on patient care (Coma del low income countries to retain health care work-
Corral et al, 2005). E-mail provides easy and im- ers by providing financial incentives or increased
mediate contact with colleagues and can reduce remuneration due to their limited financial bud-
time spent waiting for missed or returned calls. gets. As a result, solving the problem of health
302
care worker shortages is a long term process. system can be used to predict and manage levels of
Developing countries need to focus on improving recruitment, training, delivery and production needs
infrastructure that allows for information sharing, (Dambisya, 2004). Through the use of such tracking
training and development through technology in and monitoring systems, appropriate training can
order to retain their human capital. be given to workers while being able to maintain
Distance learning abilities such as wireless staffing needs. A strong commitment to rural areas
technology is an opportunity that should not be in training is vital as it shows not only workers, but
overlooked by human resource professionals in the population that they do not need to migrate away
order to meet training needs. It allows remote from rural areas to urban centers to get the quality
access to human capital and creates opportunities training they need. Monetary incentives remain a
for remote training where there once may have popular way of retaining and recruiting health care
been little recourse but to remove the worker to workers in rural developing areas, however other
a centralized location. In rural areas that have incentives can be vitally important. This includes
limited health workers to start with, removing housing, transportation, security and promotion
workers for training for significant amounts of prospects. All of these can then be used to retain
time could mean a significant reduction in services and recruit workers in rural, developing areas. These
and care for the population they serve. Wireless non-financial incentives create an incentive for health
technology such as wireless local area networks care workers to not only stay in the country, but to
and mobile technologies requires only a modest stay in rural areas, as most developing countries are
investment at a central area, and little in the way unable to compete on wages alone.
of remote access. Moving human resources management to a
Ensuring information is focused on the needs state level and no longer spread between various
of a rural population allows not only the infor- regions in the country and developing a com-
mation but the technology to be successfully prehensive national HRM policy was a first step
transferred and sustained. This is critical for a to a more efficient system in Afghanistan. This
long-term commitment in recruitment and train- set the stage for a database to track training and
ing. Use of the Internet for distance learning or other government agency initiatives; a centralized
even use of cell phones can facilitate learning agency working with all levels of government has
yet keep the worker on the front lines as well as a better chance of fulfilling its HR obligations.
minimize costs. In extremely remote locations
use of technology is not always feasible. In such
cases radio communications or even mail can be tHe Future: trAInInG
used to substitute for those learning opportunities And educAtIon
(Awofeso, 2008). Television is another low cost
opportunity of training workers in rural environ- Information and communication technologies
ments. In South Africa the use of televised com- continue to expand and improve. New dimensions
munications allowed the department of health to of interactivity offered through social networking
facilitate some of its training from a centralized and increased connectivity will have an influence
location without having to transport workers to a on human resources and HRM in health care as
training facility (van Rensburg, 2008). well. Once again these technologies will first be
A proper human resource information system rolled out in urban centres and offer improved ways
that collects data on the training quality and support of training and education of health professionals
given to workers is another key success factor in rural, and these urban centres will be able to provide
developing areas. A human resource information these tools to rural and remote regions.
303
Within medical schools adoption of computer definitions to illustrative images and interactive
technology has been initially slow. In most cases student exercises. At the same time it provides
introduced to assist in administrative functions an introduction to certain informatics practices
medical schools are increasingly turning to including basic security and use of an EMR for
computer-assisted learning in one form or another patient care (Speedie & Niewoehner, 2003).
(Gordon et al, 1999; Riva, 2000, Ward et al, 2001). More advanced virtual patients, e.g. cyberPa-
The spectrum of computer usage runs from simple tient by CyberActive Technology Ltd (Qayumi &
use of email as a communication tool between Qayumi, 1999), provides not only virtual patient
students and teachers to posting links to helpful cases but the opportunity for the learner to con-
web sites, posting course notes, posting interac- duct a physical examination with auscultation
tive quizzes, animations that supplement lecture and palpation modalities, order laboratory tests
material or provide diagrammatic understanding of (blood tests, EKGs, X-rays, etc), offer a diag-
more complex mechanisms, to the use of computer- nosis, plan treatment and if necessary simulate
based simulations or virtual reality. Information a surgical procedure. Feedback can be provided
and communication technology is now seen as an immediately by the ‘chief surgeon’ or physician,
integral part of the medical school enterprise and or at the end of the case. Patients can be followed
often expected by its students as part of the mode up for months or years after their treatment and
of content delivery. What differentiates medicine management; thus students can revisit patients
though from other forms of learning is the interac- and appreciate the long-term follow-up that oc-
tion with patients and the need to acquire practical curs with ‘real’ patients. Student responses can
skills (physical examination, e.g. assessment of be tracked making this virtual patient simulation
heart and lung sounds, palpation; procedures, suitable for testing purposes.
e.g. injections, inserting IV, suturing, surgical Computational advances in natural language
techniques). Can these tasks be supplemented or processing, 3-D modeling and database reposi-
addressed by computer-based tools? tories have led to the development of even more
sophisticated virtual patients that can respond
with emotions and answer questions with natural
leArnInG WItH vIrtuAl PAtIents language (Hubal et al, 2000). Physical signs are
again available in terms of heart and lung sounds,
The University of Minnesota’s Medical School etc. The virtual standardized patient provides
Virtual Clinic with 70 virtual patients is set up feedback, tracking and recording of responses.
as a web-accessible electronic medical record
with tabs that include health history, lab results,
medications ordered, images (e.g. X-rays), prog- medIcAl sImulAtIons
ress notes and links to teaching items. The virtual
clinic is used by first year medical students, and One more step of complexity introduces a ‘simu-
is also to be used in the IVIMEDS virtual pa- lated’ patient or mannequin linked to a multimedia
tient panel (Voelker, 2003). Students ‘attend’ the computer system to more closely approximate the
weekly clinic by logging into the virtual clinic and bedside learning experience and to allow students
see a group of simulated patients representing a to practice in a stress-free environment in situations
variety of conditions and cultural backgrounds where wrong decisions may literally mean life or
and follow them over time. The underlying tools death. One of the most fully developed simula-
facilitate a variety of educational links for any item tors is ‘Harvey’ a cardiology patient simulator
of information from simple popup windows for developed at the University of Miami Centre for
304
Research in Medical Education (CRME) (Gordon pursuing rotations at remote sites” (Jacobs et al,
et al, 1999; http://www.crme.med.miami.edu/ 2003; Project TOUCH (Telehealth Outreach for
harvey_findings.html). Harvey is programmed to Unified Community Health – http://hsc.unm.edu/
present with lifelike vital signs including blood touch). This multi-year collaboration between the
pressure, arterial and venous pulses, heart and University of New Mexico and the University of
lung sounds (with accompanying multimedia Hawaii uses case scenarios as virtual models to
instruction), access to laboratory data, pathology, make the learning of critical concepts relevant
and treatment choices. Harvey has been used by and more life-like. Students are fully immersed
medical students, nursing students, primary care in a representational virtual environment and can
physicians, cardiologists, EMTs, firemen and other see fellow students or team members and interact
health professionals. as if they were physically present even when
Successful simulations/simulators now exist separated by significant distances. The “Flatland’
for anesthesia, emergency medicine, and surgery. environment allows real-time exploration and
In the latter it may be far more preferable to de- manipulation of 3-D objects and images within
velop skills in a simulation environment for a new the virtual world. The application is displayed
or complex medical procedure rather than in the using the Access Grid – a broadband Internet2
traditional apprenticeship model on real patients videoconferencing, multicasting network that
(Gorman et al, 2000). Surgical training has been allows real-time distribution of the learning en-
extended with the marriage of virtual reality or 3-D vironments across academic medical centers and
modeling into the simulation exercise. Manipula- distant training centers.
tions are performed through haptic interfaces that Second Life (http://secondlife.com) is a popu-
not only convey a rich three-dimensional depic- lar virtual world platform that is being used for
tion of the tissue, but can also simulate touch and medical and health education. On the “Healthinfo
pressure. Examples include laparoscopic camera island” (Second Life Medical and Consumer
and cholecystectomy simulators, a VR-based Health Libraries), for instance, librarians have
needle-driving simulator for suturing technique, a provided extensive consumer health information
Web-based surgical training system for abdominal services that users can wander through in a virtual
aneurysm stent graft deployment and a Web-based environment (Boulos et al, 2007).
CardioOp system that allows for the composition
of multimedia fragments that can be reused to suit
the needs of individual learners in cardiac surgery communIty oF leArners/
(Gorman et al, 2000). communItIes oF PrActIce
“Future physicians will be able to rehearse an The Internet can offer a variety of modes for
operation on a projectable palpable hologram establishing communities of learners via discus-
derived from patient-specific data, and deliver the sion boards, chat rooms, virtual seminars and
data set of that operation with robotic assistance I.P. videoconferencing. Peer-to-peer learning is
the next day” Gorman et al, 2000, p. 353 enhanced and encouraged in these settings and
faculty can also act as mentors for students at a
Another example weds all of these elements distance. Virtual advisors were linked to students
into the learning environment (i.e. simulation, an from North American and international medical
immersive virtual reality environment, Internet schools with experienced emergency medicine
technology) to “overcome geographic barriers for faculty mentors in which encounters relied on
delivery of tutorial sessions to medical students electronic or voice correspondence depending
305
on their geographical location and preferences. expertise to teach medical students at a distance.
Feedback was positive for both advisors and Their aims are to apply innovative thinking and
advisees (Coates et al, 2004). new learning technologies (e.g. e-learning via
Canada’s first new medical school in over three the Internet, CD-ROM, virtual reality), new ap-
decades, the Northern Ontario School of Medicine, proaches to curriculum planning and instructional
is a collaborative effort between Laurentian Uni- design (e.g. curriculum mapping, outcomes-based
versity in Sudbury, ON and Lakehead University education, electronic study guides, peer-to-peer
in Thunder Bay, ON. It uses a distributed learning learning, a bank of reusable learning objects),
network with advanced information technology and the development of a flexible curriculum that
support to facilitate small group learning for will meet the needs of different students in many
students seeing patients in rural, northern, under- different geographic locations to a new medical
serviced and Aboriginal communities (Rourke, school open to students from around the world
2002). Components of the technology include (Harden & Hart, 2002). Two central features of
an e-learning (virtual learning) environment, the model are the curriculum map and reusable
online collaboration and telehealth (using the learning objects. The curriculum map identifies
existing NORTH network infrastructure), a medi- what, where, when and how students can learn
cal school information portal (on a high-speed and in essence they will be able to follow their
Intranet – the ORION network that connects the own map through the curriculum. On each node of
two universities), e-library (health information the map learning resource material related to that
resources centre), administrative infrastructure node is identified. This material is contained in a
(that include student/faculty/staff access and IT learning repository which contains small, discrete
support services) and IT research support. The self-contained learning objects. Benefits of this
goal is to provide faculty and students with secure, program are seen to be an increased accessibil-
reliable access to curriculum, cohorts, coaches, ity of medical education globally and improved
and faculty regardless of geographic location. medical education that is more responsive to
community needs.
The use of digital technologies through the
InternAtIonAl vIrtuAl Internet and world wide web interface continue
medIcAl scHool (IvImeds) to influence and affect the delivery of health
care and health services and with regard to HRM
This represents a ‘revolutionary’ approach to can provide the venue for training and educa-
medical education leaving the notion of a physi- tion of health workers, access to the medical
cal medical school associated with a teaching literature and knowledge databases providing
hospital far behind (Harden, 2000; Harden & best evidence for clinical care and clinical care
Hart, 2002; http://www.ivimeds.org/). Based in guidelines, continuing education opportunities
Dundee, Scotland and formally launched in 2003 and invaluable links to colleagues and mentors
IVIMEDS builds on the successful experiments that will improve the retention of health profes-
with virtual degrees and virtual universities sionals in remote and rural regions. Considerable
(Pritchard, 2003). An international group of 30 progress and use of these technologies has been
to 40 medical schools (from the US, UK, Europe, made, as we have discussed, and in the next
Australasia) is exploring the feasibility of estab- generation the so-called Web 2.0 technologies
lishing a virtual medical school with a broadly- will have an even greater impact on these activi-
based consortium of dozens of medical schools ties with their emphasis on social connection
that will contribute shared learning resources and and collaboration.
306
Web 2.0 technologies take the original Web little actual use of these technologies. Medical
based resources beyond the simple provision of students had both greater familiarity and use,
text, images and video and add a dimension of especially of instant messaging, media sharing
interactivity and social networking that enhance and social bookmarking. There was interest in
professional, patient and consumer interactions the use of Web 2.0 technologies for education
(Boulos & Wheeler, 2007). More common ap- by all groups but both practitioners and students
plications in health and medical education include wanted more training in their use. Other barriers
the use of wikis, web information sites that are for use in education were different learning pref-
editable by anyone; blogs or web logs or online erences, concern about quality of resources, lack
journals that can offer text or multimedia–based of time and difficulties with ICT access (Sandars
discussions in one-way or two-way environments; & Schroter, 2007).
and podcasts or audio files that can be downloaded Since these are interactive tools and devel-
to a user’s computer or audio device and listened oped collectively by a community of interested
to anywhere, anytime (Boulos et al, 2006; Boulos participants the resulting quality can vary greatly.
& Wheeler, 2007). Media sharing sites such as Although some proponents profess a “darwikian”
You Tube contain a number of video presentations theory of survival of the fittest or best, that is not
of educational material available for widespread always the case (for example, good content can
use and the social networking sites, MySpace be replaced by poor content or by misinformation,
and FaceBook, have become areas where patients sites can be hijacked by groups of users with selec-
can find support from fellow patients, students tive points of view) (Boulos & Wheeler, 2007).
can share work and learning, and researchers In some cases those perils can be lessened by the
and health professionals can find like-minded presence of a moderator or access restricted to
individuals to share information and establish only certain individuals, but to a certain extent
connections across the world. Examples of each that defeats the purpose of the original idea of
of these technologies in health education exist, social networking and building free and open
(e.g. Ganfyd a collaborative medical knowledge collaborative networks.
database that anyone can read and medical pro- Many of these web tools require quite high
fessionals can edit – http://ganfyd.org; Clinical bandwidth and high connectivity speeds, infra-
Cases and Images blog from the Cleveland Clinic structure that may be readily available in urban
- http://clinicalcases.org/; Johns Hopkins Medical centres, but may still not be available to those in
Podcasts for health professionals and the pub- remote regions or in low income countries. ICT
lic - http://www.hopkinsmedicine.org/mediaII/ use and penetration continues to expand across
Podcasts/index.html; medical school pathology the globe with the greatest increases over the past
lectures on YouTube – http://www.youtube.com). several years seen in mobile cellular technologies
It is still early days in the use of these tech- and networks. In the developing world mobile
nologies in training and education, and research phones had reached an estimated 49.5% penetra-
is still needed into how these technologies best fit tion rate by the end of 2008 (from almost zero ten
into undergraduate, postgraduate and continuing years ago) (ITU, 2009, p. 1). Fixed Internet access
education of health professionals (Boulos et al, in developing countries is still limited and where
2006; Sandars & Schroter, 2007). A recent survey available is often slow and expensive. Although
of 3,000 medical students and 3,000 medical there is still a digital divide the latest indicators of
practitioners in the UK found that although there ICT use and prevalence indicate that it is slightly
was familiarity with these Web 2.0 technologies closing between countries with high and low ICT
for both personal and educational use, there was levels (which are in turn correlated to high (Europe
307
and North America) and low income countries and bioinformatics will mean the addition of new
(central and southern Asia, sub-Saharan Africa)) health professionals to the health care team with
(ITU, 2009, p. 1-2). One encouraging sign for the expertise in the use of these new technologies
coming decade is the increase in, and adoption of, and their integration into a fully electronic health
national eHealth policies that provide a framework information system (Anvari, 2006). Technology, in
for developing and using ICT in the health sec- some instances, will also lead to the disappearance
tor. Although industrialized countries still show of jobs (e.g. medical transcriptionists, radiology
the highest rates of adoption of eHealth policies, film technicians, health records clerks).
African regions projected a doubling of adoption Shifting of care from the hospital and health
of policies and ICT development by the end of facilities to the community and home will also
2008 (WHO, 2006, p. 18-20, 31).. Accompanying mean an increased need for community and
the introduction of policies and infrastructure is home health care workers and expanded roles for
the recognition of the need for building capac- family and community-based physicians, nurse
ity, i.e. the need to train the health workforce in practitioners, and even patients in controlling and
use of the eHealth applications being deployed. monitoring their own conditions and health. In
ICT training of both health science students and many countries there is a lack of national health
continuing education for health professionals is human resources policies and planning to address
relatively stable across all geographic regions with these current and future needs.
roughly two thirds of countries reporting they
have introduced some form of university level
and continuing education in ICT (WHO, 2006, conclusIon
p. 46-47). The need to build human resources
capacity in health informatics and eHealth was Technology in health care overcomes barriers for
reported by responding countries and often cited health professionals in all regions. It minimizes
as a barrier to full eHealth implementation (WHO, distance and accessibility issues, providing a
2006, p. 19). medium whereby information is easily and im-
mediately attained. Technology furthermore
maintains communication between health pro-
HumAn resources: neW roles fessionals and academic sources/centres which
allows for continued education and training. In
The introduction of new technologies and changes addition, it enhances the ability to recruit and
in practice will also have an impact on the types select qualified personnel from all regions for
of health professionals needed in the workforce optimal deployment of human resources by the
in the future. As already mentioned there will be use of HR information systems. Computer and
a need for increased training in the use of ICT for Internet access to health professionals in rural
all health professionals and an increased need for regions have provided a better means of access
health informatics specialists to develop, maintain to critical medical information, distance training
and use these technologies and manage the mas- and continuing education and have improved
sive amounts of health data that will be available medical care services to patients. Although initial
across shared data repositories. set up costs may be substantial, such investments
Multidisciplinary and interprofessional health indicate long term HRM benefits. In developing
care teams will be supported by improved technol- areas, governments – specifically those in urban
ogies. The introduction of robotics and minimally and rural developing regions – need to improve
invasive surgical techniques, pharmacogenetics, technological infrastructures in order to provide
308
better telecommunication services. In rural devel- Boulos, M. N. K., & Wheeler, S. (2007). The
oping areas, centralized management will ensure emerging Web 2.0 social software: an enabling
consistent information is being provided to health suite of sociable technologies in health and
professionals, and will provide well developed health care education. Health Information and
HRM strategies in all areas of the government Libraries Journal, 24, 2–23. doi:10.1111/j.1471-
pertaining to health care management. 1842.2007.00701.x
Chaudry, B., Wang, J., Wu, S., Maglione, M.,
Mojica, W., & Roth, E. (2006). Systematic Re-
reFerences
view: Impact of health information technology
Adatia, F., & Bedard, P. L. (2003). Palm reading: on quality, efficiency, and costs of medical care.
2. Handheld software for physicians. Canadian Annals of Internal Medicine, 144, 742–752.
Medical Association Journal, 168, 727–734. Clements, A. (2008). It’s Good to Get a Second
Adatia, F. A., & Bedard, P. L. (2002). Palm read- Opinion. Human Resources, 66. Retrieved No-
ing: 1. Handheld hardware and operating systems. vember 9, 2008 from http://proquest.umi.com/pq
Canadian Medical Association Journal, 167, dweb?did=1582613671&sid=7&Fmt=3&clientI
775–780. d=77774&RQT=309&VName=PQD
Anvari, M. (2007). Impact of Information Technol- Coates, W. C., Ankel, F., Birnbaum, A., Kosiak, D.,
ogy on Human Resources in Healthcare. Health- Broderick, K. B., & Thomas, S. (2004). The virtual
care Quarterly, 10, 84-88. Retrieved November 9, advisor program: linking students to mentors via
2008 from http://www.longwoods.com/product. the world wide web. Academic Emergency Medi-
php?productid=19320&cat=513&page=1 cine, 11, 253–255. doi:10.1111/j.1553-2712.2004.
tb02205.x
Awofeso, N., Dalhatu, A., & Schelokova, I.
(2008). Training of Front-line Health Workers for Coma del Corral, M. J., Luquín, P. A., Guevara, J.
Tuberculosis Control: Lessons from Nigeria and C., Movilla, A. O., Torres, G. T., & Garcia, J. L.
Kyrgyzstan. Human Resources for Health, 6(20). (2005). Utility of a thematic network in primary
Retrieved October 28, 2008 from http://www. health care: a controlled interventional study in a
human-resources-health.com/content/6/1/20 rural area. Human Resource for Health, 3, 4-11.
Retrieved November 9, 2008 from http://www.
Boulos, M. N., Marimba, I., & Wheeler, S. (2006). human-resources-health.com/content/3/1/4
Wikis, blogs and podcasts: a new generation of
Web-based tools for virtual collaborative clinical Curran, V. R. (2006). Tele-education. Jour-
practice and education. BMC Medical Education, nal of Telemedicine and Telecare, 12, 57–63.
6, 41–49. doi:10.1186/1472-6920-6-41 doi:10.1258/135763306776084400
Boulos, M. N. K., Hetherington, L., & Wheeler, Dambisya, Y. A. (2007). Review of Non-financial
S. (2007). Second Life: an overview of the Incentives for Health Care Worker Retention in
potential of 3-D virtual worlds in medical and East and Southern Africa. Regional Network for
health education. Health Information and Librar- Equity in Health in East and Southern Africa, 1-72.
ies Journal, 24, 233–245. doi:10.1111/j.1471- Retrieved October 30, 2008 from http://www.equi-
1842.2007.00733.x netafrica.org/bibl/docs/DIS44HRdambisya.pdf
309
Dorsch, J. L. (2000). Information needs of rural Harden, R. M. (2000). Evolution or revolution

health professionals: a review of the literature. and the future of medical education: replacing
Bull Med Libr Assoc, 88(4): 346-354. Retrieved the oak tree. Medical Teacher, 22, 435–442.
November 9, 2008 from http://www.pubmedcen- doi:10.1080/01421590050110669
tral.nih.gov/articlerender.fcgi?tool=pubmed&pu
Harden, R. M., & Hart, I. R. (2002). An inter-
bmedid=11055302
national virtual medical school (IVIMEDS): the
Dussault, G., & Franceschini, M. C. (2006). Not future for medical education? Medical Teacher,
Enough There, Too Many Here: Understanding 24, 261–267. doi:10.1080/01421590220141008
Geographical Imbalances in the Distribution of
Heeks, R., Jang, S., Ko, K., & Lee, H. (2008).
the Health Workforce. Human Resources for
Analysing South Korea’s ICT for Development
Health, 4, 12-28. Retrieved November 18, 2008
Aid Programme The Electronic Journal of In-
from http://www.human-resources-health.com/
formation Systems in Developing Countries, 35,
content/4/1/12
1-15. Retrieved October 28, 2008 from http://
Ely, J. W., Osheroff, J. A., Ebell, M. H., Bergus, www.ejisdc.org/ojs2/index.php/ejisdc/article/
G. R., Levy, B. T., Chambliss, M. L., & Evans, viewFile/496/252
E. R. (1999). Analysis of questions asked by fam-
Henderson, L. N., & Tulloch, J. (2008). Incentives
ily doctors regarding patient care. BMJ (Clinical
for Retaining and Motivating Health Workers in
Research Ed.), 319, 358–361.
Pacific and Asian Countries. Human Resources
Fraser, H. S. F., Jazayeri, D., Nevil, P., Karacao- for Health, 6, 18-37. Retrieved November 18,
glu, Y., Farmer, P. E., Lyon, E., et al. (2004) An 2008 from http://www.human-resources-health.
information system and medical record to support com/content/6/1/18
HIV treatment in rural Haiti. BMJ, 329, 1142-1146.
Hongoro, C., & McPake, B. (2004). How to
Retrieved October 30, 2008 from http://www.bmj.
Bridge the Gap In Human Resources for Health.
com/cgi/content/full/329/7475/1142.
Lancet, 364, 1451–1456. doi:10.1016/S0140-
Goldsmith, H., Manderscheid, W., & Merwin, E. 6736(04)17229-2
(1995). Human Resource Issues in Rural Men-
International Telecommunication Union (ITU).
tal health Services. Community Mental Health
(2009). Measuring the Information Society: The
Journal, 31, 525-537. Retrieved October 28,
ICT Development Index, 2009 Edition. Geneva,
2008 from http://www.springerlink.com/content/
Switzerland. Retrieved April 21, 2008 from http://
p68r402563qp0755/fulltext.pdf
www.itu.int/ITU-D/ict/publications/idi/2009/
Gordon, M. S., Issenberg, S. B., Mayer, J. W., & material/IDI2009_w5.pdf
Felner, J. N. (1999). Developments in the use of
Jacobs, J., Caudell, T., Wilks, D., Keep, M. F.,
simulators and multimedia computer systems in
Mitchell, S., & Buchanan, H. (2003). Integration
medical education. Medical Teacher, 21, 32–36.
of advanced technologies to enhance problem-
doi:10.1080/01421599980002
based learning over distance: Project TOUCH. The
Gorman, P. J., Meier, A. H., Rawn, C., & Krum- Anatomical Record, 270B, 16–22. doi:10.1002/
mel, T. M. (2000). The future of medical education ar.b.10003
is no longer blood and guts, it is bits and bytes.
American Journal of Surgery, 180, 353–356.
doi:10.1016/S0002-9610(00)00514-6
310
Kabene, S. M., Orchard, C., Howard, J. M., So- Sandars, J., & Schroter, S. (2007). Web 2.0 tech-
riano, M. A., & Leduc, R. (2006). The importance nologies for undergraduate and postgraduate
of human resources management in health care: a medical education: an online survey. Postgradu-
global context. Human Resources for Health, 4, ate Medical Journal, 83, 759–762. doi:10.1136/
20–36. doi:10.1186/1478-4491-4-20 pgmj.2007.063123
Kulma, M., & Springer, B. (2006). Easing the Sattar, K. (2007). A Sustainable Model for Use
bottom-line impact of staffing shortages: A case of ICTs in Rural Pakistan. International Journal
study in shift bidding. Healthcare Financial of Education and Development using Informa-
Management, 60, 92–97. tion and Communication Technology, 3, 89-97.
Retrieved October 30, 2008 from http://proquest.
Lin, J. C. (1999) Applying telecommunication
umi.com/pqdweb?did=1409299671&Fmt=4&cli
technology to health-care delivery. Engineering
entId=11263&RQT=309&VName=PQD
in Medicine and Biology Magazine, IEEE, 28-31.
Retrieved November 9, 2008 from http://ieeex- Schiffbauer, J., O’Brien, J. B., Timmons, B. K.,
plore.ieee.org/xpls/abs_all.jsp?arnumber=775486 & Kiarie, W. N. (2008). The Role of Leadership
in HRH Development in Challenging Public
McGraw, D., Dempsey, J. X., Harris, L., & Gold-
Health Settings. Human Resources for Health, 6,
man, J. (2009). Privacy As An Enabler, Not An
23. Retrieved October 30, 2008 from http://www.
Impediment: Building Trust Into Health Infor-
human-resources-health.com/content/pdf/1478-
mation Exchange. Health Affairs, 28, 416–427.
4491-6-23.pdf
doi:10.1377/hlthaff.28.2.416
Sheppard, L., & Mackintosh, S. (1998). Technol-
Pritchard, L. (2003). Great Scots! Medical
ogy in education: what is appropriate for rural
Education, 37, 493–494. doi:10.1046/j.1365-
and remote allied health professionals? Aust J
2923.2003.01529.x
Rural Health, 6, 189-193. Retrieved November 9,
Qayami, A. K., & Qayami, T. (1999). Com- 2008 from http://www3.interscience.wiley.com/
puter-assisted learning: cyperPatient™ - A journal/119830096/abstract
step in the future of surgical education. Jour-
Smith, R. (1996). What clinical information do
nal of Investigative Surgery, 12, 307–317.
doctors need? BMJ (Clinical Research Ed.), 313,
doi:10.1080/089419399272296
1062–1068.
Riva, G. (2000). From telehealth to e-health:
Van Rensburg, D., Steyn, F., Schneider, H., &
Internet and distributed virtual reality in health
Loffstadt, L. (2008). Human Resource Develop-
care. Cyberpsychology & Behavior, 3, 989–998.
ment and Antiretroviral Treatment in Free State
doi:10.1089/109493100452255
Province, South Africa. Human Resources for
Robinson, K. S. (2003). Online Bidding Fills Health, 6, 15. Retrieved October 28, 2008 from
Nursing Jobs. HRMagazine. Alexandria, 44. http://www.human-resources-health.com /con-
tent/6/1/15
Rourke, J. T. B. (2002). Building the new Northern
Ontario Rural Medical School. The Australian
Journal of Rural Health, 10, 112–116.
311
Vian, T., Richards, S. C., McCoy, K., Connelly, Westberg, E. E., & Miller, R. A. (1998). The Basis
P., & Feeley, F. (2007). Public-Private Partner- for Using the Internet to Support the Information
ships to Build Human Capacity in Low Income Needs of Primary Care. Journal of the American
Countries: Findings from the Pfizer Program. Medical Informatics Association, 6, 6–25.
Human Resources for Health, 5, 8-19. Retrieved
Wibulpolprasert, S., & Pengpaibon, P. (2003).
November 18, 2008 from http://www.human-
Integrated Strategies to Tackle the Inequitable
resources-health.com/content/5/1/8
Distribution of doctors in Thailand: Four Decades
Vilamovska, A. M., & Hatziandreu, E. Schindler, of Experience. Human Resources for Health, 1, 12.
H.R., van Oranje-Nassau, C., de Vries, H., & Retrieved November 18, 2008 from http://www.
Krapels, J. (2009). Study on the requirements human-resources-health.com/content/1/1/12
and options for RFID application in healthcare:
World Health Organization (WHO). (2006)
Identifying areas for Radio Frequency Identifica-
Building Foundations for eHealth: Progress of
tion deployment in healthcare delivery: A review
Member States: Report of the Global Observa-
of relevant literature. Rand Europe, European
tory for eHealth. Geneva, Switzerland; WHO.
Commission. Retrieved May 4, 2009 from http://
Retrieved November 9, 2008 from http://www.
www.rand.org/pubs/technical_reports/TR608/
who.int/ehealth/resources/bf_full.pdf
Voelker, R. (2003). Virtual patients help medical
Zarocostas, J. (2008). World Medical Association
students link basic science with clinical care.
scales up training for multidrug resistant tuberculo-
Journal of the American Medical Association,
sis to fight epidemic. BMJ (Clinical Research Ed.),
290, 1700–1701. doi:10.1001/jama.290.13.1700
336, 1155. doi:10.1136/bmj.39584.662130.DB
Ward, J. P. T., Gordon, J., Field, M. J., & Lehmann,
H. P. (2001). Communication and information
technology in medical education. Lancet, 357,
792–796. doi:10.1016/S0140-6736(00)04173-8
This work was previously published in Ubiquitous Health and Medical Informatics: The Ubiquity 2.0 Trend and Beyond,
edited by Sabah Mohammed and Jinan Fiaidhi, pp. 574-595, copyright 2010 by Information Science Reference (an imprint
of IGI Global).
312
Section II
Development and Design
Methodologies
This section provides in-depth coverage of conceptual architectures, frameworks and methodologies related to
the design and implementation of enterprise information systems. Throughout these contributions, research fun-
damentals in the discipline are presented and discussed. From broad examinations to specific discussions on
particular frameworks and infrastructures, the research found within this section spans the discipline while also
offering detailed, specific discussions. Basic designs, as well as abstract developments, are explained within these
chapters, and frameworks for designing successful decision support systems, integrating new technologies, and
developing and implementing efficient processes are included.
314
Chapter 2.1
Enterprise Information Systems:
Aligning and Integrating Strategy,
Technology, Organization and People
Paul T. Kidd
Cheshire Henbury, UK
The implementation of Enterprise Information One of the expected benefits of Enterprise Informa-
Systems is a difficult task, even for large compa- tion Systems is increased responsiveness, flexibility
nies. It can be even more so for Small and Medium and agility. Yet sometimes these benefits can also be
Size Enterprises (SMEs), for most are seriously delivered through non-technical means, namely by
constrained in terms of time, money and skills. The an appropriate organizational design, based on the
key to successful implementation lies in achieving right choice of organizational design parameters:
an alignment between strategy, technology, organi- differentiation, integration, decision structure, and
zation and people, and also in achieving commit- formalization. And people are also crucial as well,
ment to the new technologies. An implementation for they need the right skills and motivation to de-
method called HiSTOP, which stands for High liver the sought after benefits. But the story does
Integration of Strategy, Technology, Organization not end with these elements, for strategy must also
and People, is described. This method provides a be considered, because different business strategies
means of adjusting all four elements so that each require, in general, different forms of responsive-
is appropriate and also so that all four elements fit ness, flexibility and agility. Therefore the key issues
together. Although the method was initially designed are aligning and integrating these four elements,
with SMEs in mind, the method is also suitable for and dealing with the interactions between them, so
larger companies, hence the chapter considers both that, for example, the needs of the organizational
types of enterprises. The method places emphasis design are reflected in the technology design, and
on development of internal implementation com- vice versa.
petencies. The key foundational principles of the Achieving this alignment and integration is a
method are discussed along with some key findings very difficult thing to bring about, even for big
from early trials. companies with large budgets, capable of buying-in
external skills and expertise to support the whole
DOI: 10.4018/978-1-60566-892-5.ch015
process of selection, design, and implementation. Armed with this new version of HITOP, which
For Small and Medium Size Enterprises (SMEs) is called HiSTOP (High Integration of Strategy,
the challenges are even more daunting, for most Technology, Organization and People), the
are seriously constrained in terms of time, money chapter will then consider the key components,
and skills. Moreover, SMEs can also be micro which constitute several steps that SMEs and
versions of large enterprises in that they also bring large enterprises can take to improve the imple-
with them a set of internal organizational politics, mentation of Enterprise Information Systems.
conflicts, outdated practices, change resistance, These considerations will involve addressing the
etc., that is to say, the features and problems that process by which the design and implementation
are so often found in large enterprises. work is undertaken, as well as the analysis of the
SMEs therefore provide a challenging en- interrelationships between Strategy, Technol-
vironment with respect to the implementation ogy, Organization, and People. Using HiSTOP,
of Enterprise Information Systems in terms of it becomes possible to begin to consider design
aligning and integrating Strategy, Technology, choices in terms of all four dimensions of the
Organization and People. What therefore can be problem.
done to help such businesses? Answering this The conclusions provide insights into critical
question is the focus of this chapter. success factors with respect to acceptance of the
However, before addressing the above it is method. These factors were identified at an early
necessary to define the theories and concepts upon stage in the development work and were therefore
which the chapter will be built. After providing a used to shape the method to ensure that it would
brief overview of the main challenges, a social not be seen as being overcomplicated or too dif-
shaping of technology approach is presented. This ficult to use. Insights from experiences of using the
method is rooted in sociotechnical theory. The method are also provided. In particular the finding
method, called HITOP (High Integration of Tech- that different modes of use are possible, since it
nology, Organization and People) was developed is not necessary to follow the whole method, and
in the United States in the late 1980s, in the context different parts can be used as required according
of implementation of advanced manufacturing to the inclinations of a particular business and its
technologies within traditional hierarchical mass specific interests and constraints.
production environments operated by large corpo-
rations. The HITOP method provides a means of
considering the organization and people require- bAckGround
ments of computer-based manufacturing technolo-
gies. This method is briefly reviewed, highlighting Failures in the implementation of Information
its underlying philosophy and assumptions, along and Communications Technologies (ICT) have
with its strengths, weaknesses and limitations. been documented in the literature for decades
Following this, a new method is discussed, (e.g. see Kidd (1994), pp 55-56, for an overview
which is an improved, generalized and extended of failures relating to manufacturing ICT). Failure
version of the HITOP method, and which is based here is used in the sense that outcomes fall short
on a broader set of needs and issues, including of expectations (Bignall & Fortune, 1984; Kidd,
those of SMEs, and also addressing the require- 1994 (p 192)). Some key reasons for failure in
ments to align and integrate Strategy, Technology, relation to implementation are:
Organization and People. The improved version
of HITOP is however also relevant to large en- • Technology oriented rather than business
terprises. and market focused investment plans;
315
• Inappropriate or non-existent business here is that, whatever the motivation for change,
strategy; there must be a sound business-driven justification
• Failure to link business and Information and for the implementation. Enterprise Information
Communications Technology strategies; Systems implementation should be concerned
• Inadequate assessment of the costs and ben- with achieving a return on investment, and this
efits of investments in new technologies; implies identifying business objectives.
• Neglect of organization and people issues. Defining and prioritizing business objectives is
extremely important. There is universal agreement
These problems were documented in Kidd in the literature about this–if business objectives
(1994), primarily in the context of advanced manu- are not clearly stated and prioritized at the very
facturing technologies and manufacturing related beginning, before embarking upon any sort of
ICT, but there is a pattern as the problems tend to software selection, then the implementation runs
reoccur across time and across different types of the risk of encountering serious problems. With-
ICT. For example, in the world of enterprise-wide out clear and prioritized business objectives it is
systems such as Enterprise Resource Planning unlikely any significant benefits will be achieved.
(ERP) and Enterprise Information Systems, gaps This in effect corresponds to establishing
between corporate expectations and results are strategic fit, meaning that an enterprise should
commonplace. For such systems there is now a achieve alignment between its business envi-
well-documented catalogue of implementation ronment and internal resources, such that these
issues and difficulties, along with prescribed criti- resources support the strategic goals. As Miles
cal implementation factors (e.g. see Davenport, & Snow (1984) have noted, truly outstanding
1998; Holland & Light, 1999; Strong & Volkoff, performance is achieved by companies that
2004; Santhanam et. al. (2006); Guang-hui et. al. achieve a tight strategic fit, both externally with
(2006); Jing & Xun Qiu (2007)). the business environment and internally among
Whilst there are many dimensions and is- strategy, structure and process. And while achiev-
sues to be considered in the implementation of ing strategic fit is crucial to excellence, it is also
enterprise-wide software systems, a number of extremely difficult to bring about.
critical points are worth mentioning. Establishing and prioritizing business objec-
The first critical point is that it is necessary for tives is therefore essential for managing the process
firms implementing enterprise systems to make the of selecting Enterprise Information Systems soft-
strategic choices needed to configure the systems ware and for determining the detailed functional-
and processes. This reduces to being very clear ity required. For example, a prioritized business
about business objectives. objective might be to share order information
There can be many reasons for change–tech- electronically across the business. If this is not
nical or business or both. Existing ICT systems clearly specified up-front, then it is not possible
may be old, costly to maintain, inflexible, and to evaluate packages correctly to find one that
suffering lots of downtime. Or perhaps the need can support this requirement. Moreover, training
for change stems from a broader business per- objectives cannot be specified to ensure that staff
spective, involving factors such as speed of order can use this system functionality effectively.
fulfillment, inventory reduction, improvements in However, it is not enough to just specify ge-
co-ordination and control, and so on. These are neric objectives, for example, improved customer
but a few of the potential benefits of Enterprise service. Such generic statements do not provide
Information Systems, and there are many more sufficient information for decision making and
(e.g. see Shang & Seddon (2002)). The key point there is a danger that ICT specialists will begin
316
to make decisions about what functionality is this bridging exercise between ICT and non-ICT
required. As a result, Enterprise Information people are rare, and would normally be brought
Systems priorities will still be out of line with in for specific projects.
business objectives. What is needed is to be very Finally, the third key point is the special nature
specific, for example, improve order processing of Enterprise Information Systems. These are not
and fulfillment by accepting customer orders from functional or department based systems. They
any location into one system, assigning shipment provide access to information right across the
dates to available products in real time, scheduling enterprise. This has job design and organizational
future shipment dates for products not in stock, implications. For example, when people elsewhere
and checking order status at any time. have to take decisions using information generated
A second key point is that of control over in another part of a business, then these people
the implementation process. Implementing an need to understand what the information means
Enterprise Information System can be risky, is (which has training implications). But the business
potentially expensive, and involves a lot of time also needs to decide if jobs and organizational
and effort. Sought after benefits, no matter how structures need to be redesigned to bring people
clearly linked to business objectives, will not using the data closer together, for example to
be achieved if the system is not implemented work in teams, or to be co-located, or to define
correctly, and there are plenty of pitfalls and jobs around the use of particular types of data.
lots of scope for making mistakes. If a company Without such considerations, benefits such as
solely relies on vendors or external consultants speed of response and flexibility may not be fully
for success, then they are likely to experience achieved simply because there is no alignment
a significant shortfall between expectation and between the features of the technology and the
achievement. For example, it is necessary to avoid organizational design.
a circumstance where inadequacies in the Enter- Moreover, the organizational structure itself
prise Information System software are discovered may be outdated and unresponsive, and in need
during the implementation. If this does happen, of modernization. For example, it may be too hi-
then implementation teams will find themselves erarchical, too functional, and be based on narrow
working in a reactive, trouble-shooting mode departmental perspectives that hinder effective
rather than as part of a strategic plan. Consequently team working and cross-functional operations.
costs will increase, schedules will slip and people Such organizations tend to be inappropriate to
will become disillusioned. This is often the cause modern business environments, and often need to
of major Enterprise Information Systems imple- be redesigned to make for more responsiveness
mentations failing to meet business objectives. and flexibility.
However, it can also be a mistake to follow an Consequently, implementing Enterprise In-
entirely in-house approach. The problem of getting formation Systems cannot be done in isolation
business objectives and Enterprise Information from organizational considerations, and some
Systems into full alignment is a significant one, form of organizational alignment, both with the
and will probably require external help. Also, time technology and with the business environment
needs to be spent explaining the business benefits should be undertaken if the expected business
of Enterprise Information Systems to non-ICT benefits are to be fully achieved. Yet, even though
managers. Only after this has taken place, can organizational issues are a critical area, the matter
managers appreciate the role that the Enterprise is often missing from implementation methods,
Information System will play in meeting busi- whether these methods be directed at SMEs or
ness objectives. The skills required to undertake large enterprises. For example, the implementa-
317
tion method described by Blackwell et. al. (2006), social sub-system to match the technology, or a
though focused on SMEs, does not provide any mixture of both.
specific means for SMEs to undertake organiza- The Sociotechnical School of thought has been
tional redesign or alignment of the organization articulated in the form of principles, (Cherns,
with strategy and technology. Moreover, even 1976, 1987) that embody the values and key fea-
academic research on the topic of organizational tures of sociotechnical design. These principles,
issues, while strong an analysis of the issues (e.g. of which there are 11, are: Compatibility; Mini-
see Ward et. al. (2005)), generally does not provide mum Critical Specification; Variance Control;
a practical implementation methodology aimed The Multifunctional Principle–Organism vs.
at delivering a means of handling organizational Mechanism; Boundary Location; Information
design. Furthermore, implementation methods Flow; Support Congruence; Design and Human
that claim to be integrated (e.g. see Leem & Kim Values; Incompletion; Power and Authority; and
(2002)), are in fact only integrated in the narrow Transitional Organization.
sense of bringing together a subset of the primary Among the above there is a sub-set of principles
elements, mainly business objectives, economic that are primarily organizational in nature. These
justification, software evaluation and selection, are: TheMultifunctional Principle–Organism vs.
and implementation. Organizational design and the Mechanism; Boundary Location; Information
business of aligning all the components (strategy, Flow; Support Congruence; and Power and
technology, organization and people) are just left Authority.
out of the process. The Multifunctional Principle–Organism vs.
In summary therefore, it can be stated that ad- Mechanism refers to traditional organizations
dressing all three critical areas discussed above is which are often based on a high level of special-
not easy, and requires a process that can: (i) achieve ization and fragmentation of work, which reduces
the necessary alignments; (ii) ensure that the cor- flexibility. When a complex array of responses is
rect decisions are taken; (iii) deliver specifications required, it becomes easier to achieve this variety
that are appropriate, and; (iv) provide a means of if the system elements are capable of undertak-
addressing organization and people issues. ing or performing several functions. Boundary
Location is a principle that relates to a tendency
in traditional hierarchical organizations to or-
mAIn Focus oF tHe ArtIcle ganize work around fragmented functions. This
often leads to barriers that impede the sharing of
existing theories, methods data, information, knowledge and experience.
and concepts Boundaries therefore should be designed around
a complete flow of information, or knowledge,
Key among the existing body of knowledge or materials, to enable the sharing of all relevant
relating to implementation approaches that seek data, information, knowledge and experience. The
to achieve an alignment between technology, Information Flow principle addresses the provi-
organization and people is sociotechnical design. sion of information at the place where decisions
The Sociotechnical School’s central tenant and actions will be taken based on the information.
is that surrounding technology, which can be Support Congruence relates to the design of re-
regarded as a sub-system, there is also a social sub- ward systems, performance measurement systems,
system. These two sub-systems can be designed to etc., and their alignment with the behaviors that
be compatible, either by changing the technology are sought from people. For example, individual
to match the social sub-system, or modifying the reward for individual effort, is not appropriate if
318
team behavior is required. Power and Authority nature, it impacts technology as well. It implies
is concerned with responsibilities for tasks, and that what has to be done needs to be defined, but
making available the resources that are needed how it should be done should be left open. In
to fulfill these responsibilities, which involves terms of features and functions of technology,
giving people the power and authority to secure the technology should not be over determined,
these resources. but should leave room for different approaches. It
There is also another sub-set of principles that implies a degree of flexibility and openness in the
largely relate to the process by which technology is technologies. Turning now to Variance Control,
designed. These are: The Compatibility Principle; this is a principle that, as its name suggests, is
The Incompletion Principle; and The Transitional focused on handling variances, these being events
Organization Principle. that are unexpected or unprogrammed. Variances
The Compatibility Principle states that the that cannot be eliminated should be controlled as
process by which technology is designed needs to near to the point of origin of the variance as pos-
be compatible with the objectives being pursued, sible. Some of these variances may be critical, in
implying that technologies designed without the that they have an important affect on results. It is
involvement of users, would not be compatible important to control variances at source, because
with the aim of developing a participatory form not to do so often introduces time delays. Next on
of work organization where employees are in- the list of principles is that of Design and Human
volved in internal decision making. Incompletion Values. This is concerned with quality of working
addresses the fact that when workplace systems life. In the context of the working environment it
are designed, the design is in fact never finished. manifests itself in issues such as stress, motiva-
As soon implementation is completed, its conse- tion, personal development, etc. This principle
quences become more evident, possibly indicating has both a social sub-system dimension and a
the need for a redesign. The Transitional Orga- technology sub-system dimension, in that both
nization principle addresses two quite distinct can be designed to reduce stress, and to enhance
problems when creating new organizations: one is motivation and personal development.
the design and start-up of new (greenfield) work- While these principles are extremely useful
places, the other relates to existing (brownfield) from the perspective of organizational design
workplaces. The second is much more difficult and technology redesign, there are a number of
than the first. In both situations the design team, limitations with respect to Enterprise Information
and the processes it uses, are potentially a tool to Systems. Specifically there is no explicit consider-
support the start-up and any required transitions. ation of, or an attempt to consider, business strat-
What remain from Cherns’ set of 11 socio- egy or the alignment between business objectives
technical design principles, is a sub-set that is and specific organizational characteristics. Nor
significantly technology oriented, although the is there any easy way to link specific features of
principles also have organizational implications. Enterprise Information Systems technology with
The principles in question are: Minimum Criti- organizational requirements. To some extent vari-
cal Specification; Variance Control; Design and ance analysis can contribute to this, but variance
Human Values. analysis does not specifically address technical
The principle of Minimum Critical Specifica- features and consider the organizational implica-
tion states that only what is absolutely necessary tions of these. Moreover, the sociotechnical ap-
should be specified, and no more than this, and proach is highly specialized, and the competencies
that this applies to all aspects of the system: tasks, to undertaken this approach are not commonly
jobs, roles, etc. Whilst this is organizational in found in businesses, both small and large.
319
One method, related to the sociotechnical ap- • The determination of requirements and op-
proach, built upon the principles described above, tions for pay, promotion and organizational
and one that tackles the issues of competencies and structure.
technical features, is HITOP, which stands for the
High Integration of Technology, Organisation and The analysis thus provides a direct and
PeopleHigh Integration of Technology, Organiza- ordered consideration of critical technology,
tion and People (Majchrzak et. al., 1991). HITOP organization and people factors, and helps to
is a method that was designed to assist with the identify those factors which require in-depth at-
implementation of advanced manufacturing and tention. The analysis also provides an expanded
information technologies. The HITOP process al- insight into the total organizational and people
lows an analysis to be undertaken of organizational impacts of specific technologies, going well
and job-design implications of critical features beyond skills and training. Identification of
of proposed technologies. It then helps with the people and organizational cost drivers in tech-
identification of key task and skill requirements nology implementation is also another result
so that human resources can be properly planned of the analysis.
and developed to meet operational needs. HITOP allows alternative organizational
HITOP comes in the form of a workbook (Ma- designs and different ways for managing people
jchrzak et. al., 1991), that is in effect, an easy to to be specified, given specific technology plans.
read analysis manual, providing step-by-step guid- HITOP also provides guidance in determining the
ance, rationales for analysis, blank analysis forms, appropriate time for implementing technology
and worked examples. It covers a wide range of plans, and helps to identify those equipment and
issues and is based on a six-stage methodology. system choices that are likely to create the greatest
The first stage of the methodology involves mak- number of people and organizational problems, so
ing an assessment of organizational change:readiness that the implementation team is better prepared
forreadiness for change, which is followed, in the to deal with these problems.
second stage, by an assessment of the technology By performing HITOP analysis, guidance is
that is proposed, to identify its critical features. provided through an iterative, systems based pro-
The third step is an analysis of the essential task cess in which all the critical features of the orga-
requirements, which leads to an assessment of the nization, people and technology environment are
skill requirements. The fifth step is concerned with systematically assessed and all implementable op-
determining how people should be rewarded. The tions are identified. This enables the consequences
final step deals with designing the organizational of major decisions to be understood before those
changes that need to be achieved given the technol- decisions are implemented. As a result, surprises
ogy and people requirements, which leads to the downstream can be reduced and needed changes
generation of a specific implementation plan. The to the technology, the organization, or the people
HITOP design tool therefore provides: involved, can be identified.
Typical benefits of using HITOP include re-
• An assessment of organizational readiness duced implementation time, improved acceptance
for change; of technical changes, and better fit between chosen
• A definition of the critical technical fea- technologies, organizational designs, and people
tures of advanced technologies; skills. All these lead to lower implementation
• The determination of essential job require- costs and the faster delivery of benefits sought
ments, job design options, skills, training from the new technologies (flexibility, cost re-
and selection requirements; ductions, etc.).
320
While this approach has many strong features, the HITOP approach has been strengthened, for
and is very good at identifying the organizational example by introducing strategic considerations
and people implications of specific features of the into the process. Also the method has been gen-
technology (called critical technical features), it eralized so that it is no longer just appropriate to
does have a number of limitations, which primar- manufacturing technologies.
ily reflect the context in which it was developed. The revised method that is being developed
Firstly, like the sociotechnical approach, HI- is based on a number of foundational principles.
TOP does not explicitly address strategy. In effect These reflect the circumstances often found in
this is taken as given, along with technical features, SMEs, but the foundational principles are also
and strategy remains invisible during the process. relevant to the needs of larger companies.
Moreover the focus of the HITOP reference The first of these foundational principles relates
manual tends to be on one person undertaking the to the basic approach or underlying philosophy
analysis necessary to figure out the organizational with respect to the balance between the use of
and people implications of a given technology. external and internal expertise. Any company
This is not to say that the method cannot be used engaged in technology implementation, whether
by a multifunctional, multiskilled team, just that this be the implementation of Enterprise Informa-
this team-based approach to implementation is tion Systems, or other technologies, can adopt one
not emphasized or supported in the manual. The of four basic approaches:
method was also developed in the context of large
organizations seeking to implement new technolo- • passive implementation;
gies in existing, rather hierarchical organizational • consultancy-centered implementation;
designs, and as a result the reference manual tends • sub-contracting based implementation;
focus on the implementation of technology, and • company-centered implementation.
tweaking the organization to make this work, and
the language used in the reference manual is rather A passive implementation approach is one
technical in nature. Consequently, it is not the most where no-one is really taking the implementa-
SME-friendly method. In fact, many businesses, tion seriously. Reliance on external consultancy
both large and small, would struggle to understand expertise is small and the development and use
the HITOP jargon used in the manual. of in-house know-how and capabilities is low. A
consultancy-centered implementation is one where
The HiSTOP Method there is a heavy reliance on external consultants
and the development and use of in-house know-
Whatever the limitations of HITOP it is neverthe- how and capabilities is minimal or non-existent.
less a power tool and provides a good starting posi- This is a traditional approach that relies on the
tion for further development. Such development use of external experts. A sub-contracting based
of the method has been undertaken with a view to implementation is one where a lot of external ex-
addressing its weaknesses and providing an even pertise is used, but the process also involves a high
more powerful implementation support tool, and level of development and utilization of in-house
one that also deals with the specific needs of SMEs. know-how and capabilities. Consultants are used
Specifically the focus has been upon making the as sub-contractors to undertake specific tasks and
method more user-friendly, so that it can be used to provide extra manpower. A company-centered
by people who are not experts in organizational implementation involved a high degree of develop-
design or technology implementation, thus making ment and utilization of in-house know-how and
it more suitable for use by SMEs. Additionally, capabilities. Consultants are used in a selective
321
way, for example, to help set-up change processes that no-one understood, it would exclude just
and to provide facilitation or to provide specialist about everybody and few would believe in it or the
technical expertise. results that emerge. For these reasons, HiSTOP is
Practical experience suggests that firms being developed to be as simple as possible, but
implementing advanced ICT such as Enterprise without compromising the validity and the power
Information Systems are much more successful of the original HITOP method. One of the chal-
and do things faster if they adopt an approach lenges has therefore been developing a method
that involves a significant development and use that involves a lot of conceptual complexity, but
of internal expertise. In effect this involves com- at the same time, is one that is easy to use.
panies doing their own Enterprise Information The third foundational principle is that the
Systems implementations with external experts method should be a group-based process. Group
supporting the process The method entails training processes are considered to be fundamental to
and developing teams of employees to analyze successful implementation. The implementation
the technology, formulate new organizational ar- of technologies such as Enterprise Information
rangements, plan the implementation and manage Systems require cross organizational working
the changes. and understandings. As such, an approach based
A company-centered method has many ad- upon a small number of experts working in isola-
vantages over the other styles and is particularly tion tends to create an elitist atmosphere. Hiding
effective in terms of: behind questionnaires, audits, process mappings,
etc., experts working in this way rarely establish
• enabling culture change; a working dialogue and establish ownership of
• involving those affected by change and change. When implementing new technologies it is
achieving ownership of the proposed necessary to understand and value the process used
technologies; to implement the technology. Enterprise Informa-
• breaking down organizational barriers and tion Systems implementation is a real opportunity
improving cross-functional communica- to transform organizations–people, culture, at-
tions and understandings; titudes and so on. Moreover, the implementation
• identifying issues that need to be resolved process can be used to establish an environment
to make effective use of the technologies; where it becomes the norm to make better use of
• tapping into local knowledge about exist- the creativity and knowledge of all the people in
ing processes. the company. Using a group process it is possible
to achieve several things:
The second foundational principle adopted for
the development of the HiSTOP method relates • initiate real teaming and cross-functional
to the issue of complexity versus simplicity. It working;
is very easy to create complex and sophisticated • tap into knowledge about what is wrong–
implementation methods that are intellectually people already have a good idea about
pleasing, but this is not very helpful in an industrial problems;
setting. It is highly probable that few people within • open up a vast resource of new ideas;
a business would be able to understand a complex • start to unfreeze the status quo and to gain
method, and even fewer would be willing to use emotional commitment to change.
it. It would be an expert’s tool and as such would
be incompatible with the aim of providing support The fourth foundational principle is that the
for non-experts. It would be full of terminology implementation process should be a dynamic
322
one. This means that the implementation team method should make formal links between features
should not spend too much time looking at the of the technology and features of the organization,
existing situation. Instead they should focus on human resource infrastructure and management
what needs to be achieved and then figure out how practices required to support the technology and
to get there and what problems stand in the way. to make effective use of it.
Many consultancy methods begin with a detailed The HiSTOP method will now be described.
analysis of the as-is situation (getting people to It consists of four phases:
fill-in questionnaires, undertaking detailed process
mapping etc). This takes a lot of effort, is pretty • Strategic Visioning and Assessment;
dull and is often not required in such detail. It • Analysis of the Technology;
leads to a lot of information that makes for good • Organizational Design;
consultancy reports, but its value in helping with • Implementation Planing and Execution
the design a new situation is questionable.
Normally it is not necessary to spend a lot of The first phase, Strategic Visioning and Assess-
time and money analyzing and diagnosing work ment, is the point where the strategic dimension is
processes that are about to abandoned or changed considered. At a basic level this involves: establish-
significantly. It is certainly necessary to understand ing and training an implementation group; verify-
the current situation, but the analysis that goes on ing the project benefits and links with strategy;
in most companies goes way beyond understand- and informing people in the organization about
ing. The aim should be to: the project, its business purpose, what is involved
and the process of implementation. This phase can
• use the group’s collective knowledge to also however be used as an opportunity to review
develop an understanding of the work pro- and revise strategy. Importantly also, this phase
cesses that are being used; provides an opportunity to consider the existing
• establish what in the existing work pro- organization design, and to ask questions about
cesses is broken and what needs to be fixed; its appropriateness to the strategy and to consider
• get everybody to reach a common un- what organizational changes may in themselves
derstanding of the as is situation, what is be beneficial, apart from those that might be
wrong and how the new technology will needed to achieve alignment with the Enterprise
help improve the work processes. Information Systems technology.
The second phase, Analysis of the Technology,
Without this common understanding people involves: establishing a common understanding
will not agree about what needs to be done to of the Enterprise Information Systems technology
use technologies such as Enterprise Information and the work process that are affected by the new
Systems, effectively. technologies; identifying major tasks involved in
The fifth and final foundational principle is that the work processes; analyzing variances and their
the implementation method should be structured, impacts; defining the features of the Enterprise
systemic and systematic. This is an important re- Information Systems technology; and establish-
quirement since structure provides a step by step ing organizational needs. This phase leads to a
roadmap to follow and activities to undertake. requirement specification in terms of technology,
Systemic means that the method should address organization and people.
the design of the whole system–technology, The third phase, Organizational Design, in-
organization, human resource infrastructure and volves: establishing organizational design con-
management practices. Systematic means that the straints and evaluation criteria; generating job
323
and organizational design options; and evaluation • considering how the information aris-
of options and final selection of new jobs and an ing from the technology can be exploited
organizational design. This stage leads to an orga- throughout the firm.
nizational design that is aligned with both strategy
and the chosen Enterprise Information Systems The implementation of ICTs such as Enterprise
technologies. The organizational structure is Information Systems should aim to identify and
designed based on some identified requirements, deal with technology, organization and human
taking account of organizational design issues such resource characteristics that are mutually interde-
as differentiation, integration, decision structure, pendent. Technical features such as: information
and formalization (Majchrzak, 1992). integration; reliability; flexibility; often generate
The fourth and final phase, Implementation specific needs for communication, or coordination,
Planing and Execution, involves: establishing or problem solving, or decision making, or all four.
what resources are available and who will do what; It is important to recognize that ICT, and this
creating an implementation schedule; and defin- is especially so for Enterprise Information Sys-
ing post-implementation follow-up actions. Once tems, has a dual capability to both automate and
these elements have been defined the enterprise inform (Zuboff, 1988). The informing aspects of
is ready to move into actual implementation, but the technology are often overlooked or not given
will be doing so with the involvement and com- serious attention. Information generated can be
mitment of the different people and functions useful for several purposes:
affected by the implementation project.
Each of the above phases is broken down into • for continuous improvement activities;
smaller steps, and along the process there are tools • to support activities undertaken by people
that support the steps. in other parts of the company;
An important part of the four-phase method, • to build up information about customers,
specifically designed into the method, is the con- process usage, etc.
sideration of information and its use, something
that is crucial to Enterprise Information Systems. It is necessary when analyzing technologies
In many companies technology, organization and such as Enterprise Information Systems to ensure
human resources have been planned and managed that the informing capabilities of the technology
independently of each other. For example, an are being fully exploited and if necessary to
engineer might prepare detailed plans to acquire change the technical specifications or planned
new technology without simultaneously consid- use of the technology or both to ensure that this
ering human resource factors or organizational is achieved.
issues. A result of this independent, asynchronous A useful approach when analyzing Enterprise
planning is failure of the technology to live up to Information Systems is to consider issues of place
its potential. One objectives of the method is to and time. Features of the Enterprise Information
analyze the technology to establish what should be System may affect the place where activities can
done to make the technology work as planned by: or need to be done, or where people can or need
to be located, or where information generated can
• by identifying the requirements and prob- be exploited, or all three. Moreover, features of
lems of the technology; the Enterprise Information System may affect the
• developing an understanding of the organi- time needed to undertake activities, or the time to
zational and human resource consequence transfer generated information, or the time required
of the technology; to respond to events, or all three.
324
By looking at issues of place and time it is Future trends

possible to begin to understand the way that En-
terprise Information Systems may no longer fit Information and communications technologies
with taken for granted assumptions about who are becoming ubiquitous, both in society at large
does what, organizational demarcations, the time and in business. Future visions of the information
needed to do activities, the place where tasks can society (for example Kidd, 2007) foresee a world
be done, and so on. where ICTs are even more crucial and central to
Importantly however, features of the Enter- business operations that they have been hitherto.
prise Information System must also be examined In addition, not only will the scope of technol-
from the perspective of organizational and people ogy applications expand, but rapid technological
requirements, and these should then be linked to change will also result in the need to undertake
the organizational design. This is done by ex- rapid updating and replacement of installed busi-
amining critical technical features, for example ness supporting technologies such as Enterprise
information integration and then linking these to Information Systems. The net result of this will
organizational requirements in the form of the need be that implementation of ICTs within enterprises
for communications between people and groups, will shift from one-off separate projects to an
co-ordination among employees, and decision ongoing process of implementation.
making, specifically what decisions and where Such a change will require enterprises to
these are best undertaken (for example locally improve their implementation capabilities and
or remotely). become agile (or adaptable) in the face of con-
Once the need for communications, co-ordi- tinuing structural changes (Kidd, 2008); structural
nation, and decision making has been established, changes that manifesting themselves in the form
these specific needs have to be addressed by of technological innovations and developments.
redesigning jobs and the organization. This can Enterprises will therefore need to improve their
be broken down into five major design issues: systems design and implementation performance.
But this performance should be not be judged
• differentiation–breaking tasks down into just against internal criteria (for example, time
jobs and grouping these into organizational to implement), but also against their competitors’
units so that work is done efficiently; implementation performance. As long as an en-
• integration–deciding what coordination terprise’s implementation performance is below
mechanisms are needed to ensure effective that of its competitors, then they should judge
working between jobs and organizational their performance as unsatisfactory. Basically, if
units; some other enterprise can do it better, that is to
• decision structure–deciding who takes de- say, for example, implement an Enterprise Infor-
cision about what; mation Systems faster, better and with less cost
• formalization–deciding what aspects of the and disruption, then such a competitor will have
work need to be defined through formal a capability of deriving a competitive advantage
procedures; from their technology implementation processes.
• management framework–developing a It is likely therefore that at some point in
management approach that supports the the future, faced with a circumstance where an
new jobs and organization. enterprise needs to improve its technology imple-
mentation processes, it will need to take steps that
325
will achieve a rapid improvement in performance. • an increased capability to avoid post-im-

Once this has been achieved it will also need to plementation problems leading to reduced
maintain its lead, and the way it should do this is costs over the life-cycle of the investment;
through a process of continuous improvement. Its • improved capability to match ICT with
competitors’ implementation performance should business needs and to make more effective
not be viewed as a static target but as a moving one. use of the technologies.
Such a vision foresees technology implementa-
tion performance as a competitive issue. This will Achieving improved Enterprise Information
place greater reliance on either developing internal Systems implementation capabilities is a chal-
competencies through methods such as HiSTOP, lenging task in any organization, even more so
or the complete outsourcing of Enterprise Infor- in an SME, which generally has limited time,
mation Systems, their implementation and their money and expertise to undertake implementa-
management. In effect companies will have to take tion projects. The development of the HiSTOP
a strategic decision, either to keep implementation method was initiated with the focus on SMEs,
in-house and use this as a competitive tool, or to taking into account the three basic constraints of
outsource Enterprise Information Systems, and time, money and expertise. The implementation
compete on other terms. For those who adopt the method is relatively simple, is based on a group-
first approach there is a need to undertake research based approach, enables a dynamic process of
into the means by which technology implemen- communication and consensus building, and
tation processes can be used for competitive provides a structured, systemic and systematic
advantage and also be continually improved and approach. Importantly efforts have been made to
refined to reflect changing circumstances. The make the method easy to use by:
work on the development of HiSTOP provides a
starting point for such an advanced perspective on • avoiding abstract language and concepts
the role of technology implementation as a tool wherever possible;
for improving competitive performance. • making things visual by using wall charts,
This is an issue that affects both large and small diagrams and pictures, making use of
enterprise, thus HiSTOP, while its development shape, patterns, graphs, and colors.
was initially motivated by the needs of SMEs, is
in fact relevant to both types of enterprises. Initially it was thought that a method might
be possible whereby no external assistance would
be needed. However, discussions with businesses
conclusIon and preliminary trials have indicated this goal to
be idealistic, and some form of external support is
Far too many firms underestimate the complexi- unavoidable in most circumstances. However, it
ties and difficulties of implementing ICTs such also emerged from the above that the method does
as Enterprise Information Systems. Yet mastering not need to be used in its entirety, and companies
technology implementation can bring significant can pick from the method steps and tools and incor-
benefits such as: porate these into their own internal methods. This
was something that was not foreseen at the outset
• a reduction in the time taken to implement, of the work, but with hindsight seems sensible.
enabling the sought after investment ben-
efits to be achieved more quickly;
326
reFerences Kidd, P. T. (Ed.). (2007). European visions for

the knowledge age: A quest for new horizons in
Bignall, V., & Fortune, J. (1984). Understand- the information society. Macclesfield: Cheshire
ing system failures. Manchester: Manchester Henbury Publications.
University Press.
Kidd, P. T. (2008). Agile holonic network orga-
Blackwell, P., Shehab, E. M., & Kay, J. M. nizations. In G.D. Putnik & M.M. Cunha (Eds.),
(2006). An effective decision-support frame- Encyclopedia of networked and virtual organiza-
work for implementing enterprise information tions (pp. 35-42). Hershey, PA: IGI Global.
systems within SMEs. International Journal
of Production Research, 44(17), 3533–3552. Leem, C. S., & Kim, S. (2002). Introduction to
doi:10.1080/00207540500525270 an integrated methodology for development and
implementation of enterprise information systems.
Cherns, A. (1976). Principles of sociotechni- Journal of Systems and Software, 60, 249–261.
cal design. Human Relations, 29(8), 783–792. doi:10.1016/S0164-1212(01)00096-6
doi:10.1177/001872677602900806
Majchrzak, A. (1992). Management of techno-
Cherns, A. (1987). Principles of sociotechnical de- logical and organizational change. In G. Salvendy
sign revisited. Human Relations, 40(3), 153–162. (Ed.), Handbook of industrial engineering (2nd ed.,
doi:10.1177/001872678704000303 pp. 767-798). New York: John Wiley & Sons Inc.
Clegg, C. W. (1984). The derivation of job design. Majchrzak, A., Fleischer, M., Roithman, D., &
Journal of Occupational Behaviour, 5, 131–146. Mokray, J. (1991). Reference manual for per-
doi:10.1002/job.4030050205 forming the HITOP analysis. Ann Arbour, MI:
Davenport, T. H. (1998). Putting the enterprise into Industrial Technology Institute.
the enterprise system. Harvard Business Review, Miles, R. E., & Snow, C. C. (1984). Fit, failure
(July-August): 121–131. and the hall of fame. California Management
Guang-hui, C., Chun-qing, L. & Sai Yun-xiu, S. Review, (Spring): 10–28.
(2006). Critical success factors for ERP life cycle Santhanam, R., Sasidharan, S., Brass, D., & Sam-
implementation. In A. Tjoa, L. Min Xu, & S. bamurthy, V. (2006). The influence of knowledge
Chaudhry (Eds.), Research and Practical Issues transfers on the implementation of enterprise
of Enterprise Information Systems (pp. 553-562). information systems. In A. Tjoa, L. Min Xu, & S.
Boston: Springer. Chaudhry (Eds.), Research and practical issues
Holland, C. P., & Light, B. (1999). A critical success of enterprise information systems (pp. 579-581).
factors model for ERP implementation. IEEE Soft- Boston: Springer.
ware, (May/June): 30–36. doi:10.1109/52.765784 Shang, S., & Seddon, P. B. (2002). Assessing and
Jing, R., & Xun Qiu, X. (2007, June). A study on managing the benefits of enterprise systems: The
critical success factors in ERP systems imple- business manager’s perspective. Information Sys-
mentation. Paper presented at the International tems Journal, 12, 271–299. doi:10.1046/j.1365-
Conference on Service Systems and Service Man- 2575.2002.00132.x
agement, Chengdu, China. Strong, D. M., & Volkoff, O. (2004). A roadmap
Kidd, P. T. (1994). Agile manufacturing: Forging for enterprise system implementation. Computer,
new frontiers. Wokingham: Addison-Wesley. (June): 22–29. doi:10.1109/MC.2004.3
327
Ward, J., Hemingway, C., & Daniel, E. (2005). HiSTOP: High Integration of Strategy, Tech-
A framework for addressing the organisational nology, Organization and People
issues of enterprise systems implementation. The Integration: Bringing together enterprise
Journal of Strategic Information Systems, 14(2), subsystems to create an enterprise system fit
97–119. doi:10.1016/j.jsis.2005.04.005 for purpose. These subsystems include strategy,
technology, organization, and people.
Zuboff, S. (1988). In the age of the smart ma-
Strategic Fit: Aligning strategy, technology,
chine: The future of work and power. New York:
organization, and people with the needs of the
Basic Books.
business environment.
Strategically Driven Implementation: An
implementation of Enterprise Information Sys-
key terms And deFInItIons tems driven by strategy, as opposed to a technology
driven implementation where the business strategy
Alignment: The process of achieving stra- is not the primary driver for the implementation.
tegic fit between an enterprise and its business Structured, Systemic and Systematic
environment and arranging resources internally Implementation Method: An implementation
to support that alignment. method that provides a step by step implementa-
Business Objectives for Enterprise Informa- tion roadmap (structured), addressing the design
tion Systems: The specific business objectives of the whole system–technology, organization,
that should be made explicit prior to Enterprise human resource infrastructure and management
Information Systems selection, and which should practices–(systemic), and making formal links
be fully supported by the Enterprise Information between features of the technology and features
System. of the organization, human resource infrastructure
Critical Technical Features: Features of and management practices required to support
technology that have implications for organisation the technology and to make effective use of it
design, specifically in terms of communications, (systematic).
co-ordination, and decision-making.
HITOP: High Integration of Technology,
Organization and People
This work was previously published in Enterprise Information Systems for Business Integration in SMEs: Technological,
Organizational, and Social Dimensions, edited by Maria Manuela Cruz-Cunha, pp. 271-285, copyright 2010 by Business Sci-
ence Reference (an imprint of IGI Global).
328
329
Chapter 2.2
An Adaptive E-Commerce
Architecture for Enterprise
Information Exchange
Youcef Aklouf
University of Science and Technology, Algeria
Habiba Drias
University of Science and Technology, Algeria
AbstrAct modifying their Information System. The case

study also indicates that our system is developed
This article contributes to the design of a generic as a Service Oriented Architecture.
framework for providing a new way to exchange
information between enterprises. This concept is
a well addressed in the context of B2B standards. IntroductIon
Many organizations are increasingly searching
for adopting these standards to automate data Over the last 20 years, e-commerce (EC) has been
exchange. But the limit of such models resides recognized as an efficient tool to handle complex
in the fact that the content of exchange is defined exchange and transactions between companies.
in several formats which make their use difficult. EC is becoming ever more important in develop-
To overcome this difficulty, we have explored the ing new activities and new models especially for
possibility to integrate new models for describ- business-to-business (B2B) interactions.
ing content involved in B2B transaction which B2B is concerned with all activities involved in
represent a key issue. Our finding establishes the obtaining information about products and services
feasibility of integrating product models described and managing these information flows between
by ontology with e-commerce standards especially organisations (Peat & Webber, 1997). The B2B
at the business process level. This article presents architectures are difficult to conceptualize be-
a descriptive model allowing partners to exchange cause they handle several scenarios as a Business
information with other organisations without Process (BP) and several contents with different
Copyright © 2011, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited.
An Adaptive E-Commerce Architecture for Enterprise Information Exchange
formats as product catalogs defined separately by architecture used for IT and semi-conductors
different organisations. content. The BP model of RosettaNet named PIP
EC systems are well-conceived to address (Partner Interface Process, RosettaNet, 2001c) can
these different issues, even when the multiples be used with other catalogues. ebXML (ebXML,
sources of information will cause heterogene- 2001a) is a horizontal model defined without any
ity of exchange (Tellmann & Meadche, 2003). relation to any product catalogues, it just describes
To ensure and exchange competitively between some specifications for BP, core component,
partners it will be necessary to have a new view registry, and all pieces required to implement a
about how we describe all concepts involved in global B2B system.
such interactions. As a result, a B2B system is built using the
The main concepts related to EC models are existing standards cited above by overcoming their
a BP concept and a content (payload or useful limits and sharing their different advantages. Also,
data) concept. These two pieces of information are the system will ensure interoperability, which is a
defined separately by each standard in a specific key issue in order to share several product catalogs
format. This article proposes an adaptive archi- with the same architecture. This is the main rea-
tecture that gathers product catalogues standards son why we propose in this article to use product
with EC standards in order to automate exchange ontology1 for characterizing product catalogs and
and purchasing operation. trying to define BP ontology for managing any
A number of standards try to define a global kind of transactions. Shared ontologies play a
and a generic architecture to consider large in- crucial role for supporting an adaptive architec-
dustry sectors and areas. The old one is the UN ture to save time and money and to increase its
initiative: Electronic Data Interchange for Admin- efficiency through B2B transactions.
istration, Commerce, and Transport (EDIFACT) This work aims to define an adaptive in-
(United Nations, 1999). The shortcoming of this frastructure using ontologies for the electronic
generation of approaches like EDIFACT is in the commerce, especially in the B2B framework. The
fact that they require a significant programming exchange of data, information, and knowledge are
effort from organizations to be able to use these the key issues. Two kinds of ontologies can be used
standards. Thus, the cost will is higher for this to improve exchange in such architecture. One
reason. EDIFACT has just been used by small addresses products and services characterization,
number of companies and was not allowed for whereas the other addresses business processes.
small organizations. To overcome this limitation We demonstrate the capability of integration of
and to close this gap in order to reduce cost and to these two kinds of ontologies in a unique and
improve quality of interaction and communication global B2B-layered architecture. The idea is to
between partners, new standards have appeared use several business process ontologies conjointly
in the same time with the development of the with several products ontology in order to ensure
Internet infrastructure. For example, there are reliability and feasibility of such exchange in
some e-commerce portals for online purchasing different ways.
and ordering product from online catalogues— In the first part of this article, we argue that
harbinger.net, mysap.com, verticalNet.com—in ontology for both BP and product is needed to
which transaction content specifications can be automate B2B exchange. In the second part,
rendered easily by standard browsers. In addition the proposed architecture and its three parts are
to XML and EDI initiatives (Westarp, Weitzel, presented. The third part describes the differ-
Buxmann, & König, 1999), a proprietary standard ent components of the system with an overview
(de facto) exists like RosettaNet which is a B2B of objectives and functionalities for each part.
330
Finally, the last part shows a case study of the a high operation, a cost project and it is a complex
developed platform using PLIB ontology and the process. According to this, different organiza-
Web services paradigm. tions and industries leaders must work conjointly
in order to drive B2B standards, to define both
content ontology and collaboration in its differ-
ontoloGIes For e-commerce ent implementations steps in order to reach an
ArcHItectures agreement about the shared ontologies.
Therefore, on top of the structure of the ex-
In spite of the promise that e-commerce holds, changed message, two kinds of knowledge need
most implementing organisations have not re- to be modeled and shared across all the partners
ceived the full benefits from their B2B invest- involved in B2B transactions: business process
ments. This is because they use their proprietary knowledge and product and service knowledge.
solutions (backend systems) for modelling BP Such shared conceptualizations are usually called
and for describing product catalogs. In addition, ontologies. Thus, two kinds of standard ontologies
this information doesn’t provide an agreement have to be developed: ontology of products and
among all participants. For this reason, a B2B services on one hand, and ontologies of business
e-commerce exchange requires some consensual processes on the other hand.
decisions about all data shared and all services With this motivation in mind, our objective
used in transactions. As suggested by several is to define a generic architecture allowing the
authors (BIC, 2001; Trastour, Preist, & Colemann, use of any BP interfacing it with any product
2003), it is necessary to have a meta-model that catalogs content. This model could be applied in
states the high-level architectural elements without several areas.
getting into details of their implementations. Such
a model will remain relatively stable over time,
even if technologies, standards, and implemen- tHe ProPosed ArcHItecture
tation details evolve. It also was allowed to split
the work and to select a best-in-class solution This section proposes an exchange model repre-
for each facet of the global system in order to be sented by a layered architecture (Aklouf, Pierra,
adaptive, to ensure scalability and to provide a Ait Ameur, & Drias, 2005) gathering the vari-
highly integrated organization. Ontology is being ous classes of information necessary to ensure a
used to define this global view as an abstraction completely automatic exchange between partners.
level sharing product catalogs and BP descriptions. This exchange model is inspired from several
Thus, product data and BP will be shared and have infrastructures proposed in the literature such as
a same description for all participants in order to ebXML (ebXML, 2001), RosettaNet (RosettaNet,
promote exchange between them. Ontology deals 2001a) OAGI (Randy, 2001), and so on. The model
with interoperability of business contents and of with its three layers is also called an integration
message exchanges among business systems of model within the company processes. This model
different enterprises. makes it possible to integrate processes remotely
If it is clear that ontologies are necessary to between several companies.
be integrated in e-commerce architectures (Mc- These processes can be industrial or logistic.
Guinness, 1999; Jasper, & Uschold, 1999), many Integration can require the installation of a work-
issues and challenges have to be addressed very flow system between organizations (Trastour et
carefully. The implementation of such system al., 2003). This business model has as main objec-
technology is also a key management issue. It is tives, the reduction of the coordination costs as
331
well as the reduction of durations of exchange of Figure 1 illustrates the architecture of the
goods and services. model with these three components.
The most categories of information (classes) This architecture allows exchange between two
identified by the model are summarized in the or several partners represented by the systems as
following: A, B ... Z, via interfaces between layers. These
interfaces provide services making it possible to
1. Information that allow a system to locate and deliver the necessary information to the adjacent
discover automatically a partner, which pro- layer. The three levels represent the three parts of
vides the service required by the consumer the system and can work conjointly to increase the
(like a specific search engine for companies automaticity of the B2B exchange model.
and services); The partner must give some technical solu-
2. Information about the supported business tions describing various measurements taken
process, gathering different steps to be fol- and standards chosen to adapt the company local
lowed during collaboration. This is defined system to support a communication with others
by several languages and protocols; partners. This requires on one hand, the exten-
3. Information describing products and ser- sion of the internal information system by tools
vices used in the exchange (products ontolo- belonging to these various levels and, in addition,
gies describe this category of information). the integration of these tools at the local level,
allowing each company to have a uniform global
Figure 1. The layered infrastructure representing the exchange model
332
system in accordance with the suggested B2B have an effective and a reliable exchange system
model (Aklouf & Drias, 2007). in which the integration of the companies is done
As stated previously, to carry out exchanges with lower cost.
between companies using this model, it is neces- The main objectives of this architecture are:
sary to develop the following points.
• The description of an infrastructure that pro-
• First, use the discovering layer (1) and have poses an intelligent module for discovering
the possibility to locate in an automatic way, and localizing partners who propose services
the company having the capability to answer and eBusiness catalogues: the most known
the requests with the customer requirements standards are UDDI (Dogac, Cingil, Laleci,
whatever their nature might be; & Kabak, 2002) and the ebXML registry.
• Then, once the localization of the partner is • The proposition of a business process ontol-
accomplished, a collaboration agreement is ogy model based on the existing standards
established by the layer business process (2). such as PIP of RosettaNet (RosettaNet,
This layer provides all steps to be followed 2001c), ebXML BPSS (ebXML, 2005),
and executed during the exchange; BPMN (Business Process Management No-
• Finally, the exchange can start with the use tation), BPEL (Business Process Execution
of a format of contents accepted by both Language), and so on.
actors (3). This supposes that a preliminary • And finally, the integration of the exist-
consensus on the nature and the representa- ing industrial catalogues of components
tion of the product catalogs of data exchanged describing objects or services from several
was established previously. industrial sectors. Among them we find,
RNTD, RNBD of RosettaNet (RosettaNet,
objectives of the Proposed 2001b), PLIB (ISO 10303-11, 1994), and so
Architecture on.
The objective of the suggested model is to set Our principal focus is on the definition of an
up an adaptive architecture which can be used open architecture allowing the integration of dif-
as a horizontal or a vertical system. A vertical ferent technologies and knowledge coming from
standard is a system which is specific to some heterogeneous communities. This architecture
kind of activities or some particular products. requires the modeling of business processes adapt-
For instance, RosettaNet is a standard specific to able to the needs and the requirements dictated
semiconductors and electronic information com- by the specificity of these exchanges.
merce. A horizontal standard is a general system
which defines exchange protocols and informa- Architecture Functionalities
tion formats without referencing any product or
service. The standard ebXML is an example of This model presents a set of useful functionalities:
a model which proposes generic and standard-
ized services for the most industry branches and 1. The possibility of adding new functionalities
which can be adapted to particular fields and to the system as a product catalogue model
contexts. This model provides also a collabora- or as a separately developed dictionary;
tive environment allowing industrial managers, 2. The possibility of managing applications
the consortia and the developers of standards, to and data locally or remotely;
work conjointly or in collaboration in order to
333
3. The factorization of a set of knowledge as calization might be, could publish its services and
standards and business rules useful for the find and localize in a dynamic way the required
various parts of the system; services. This vision is not yet materialized. There
4. The possibility of modifying the topology are, however, some concrete uses of standards
of the model, following the adoption of new such as UDDI and ebXML. Currently, it is about
standards or the implementation of a new a limited use within a restricted and controlled
tool (ensure evolution of the system – scal- framework: Intranets.
ability-); The fact of indexing the various services of-
5. The flexibility of the model accepting the fered by the same organization is mainly important
adhesion of new partners without modifying if this organization wishes that its services will
the architecture; be used and shared. The importance of having
6. the possibility offered by the loosely coupled a central directory is to guarantee discovery of
aspect of the model, which offers the pos- the services provided by organizations if they are
sibility to take into account competences published (it is like a specialized research engine
and tools proposed by partners in order to for organizations profiles and services). It is exactly
be used by the defined architecture (Aklouf the same case for a Web page which will not be
et al., 2003). visited if it is not referred to in a search engine.
In other words, this component makes it pos-
In the next section we describe separately and sible to determine how company information and
in more details each part of the system. services must be organized in order to allow the
community which divides this directory to have
discovering layer access. This directory or repository must provide a
general schema and each company must publish its
The first task which must be realized by the profile like its services according to this schema.
exchange system consists in finding the partners The repository must support the following
with whom to collaborate. This part of the system principles bases:
must provide exactly the required service and
the contact information about the supplier of • The access to the directory and the suggested
services. To satisfy this requirement, a repository schema is free, without control and limit;
(directory) containing business documents, data, • The access to the data directory is done by
meta-data and necessary mechanisms to research contextual research or through a hierarchical
and discover partners, must be developed. This keys organization;
module must be carried out with the collabora- • The management of the repository is ensured
tion of all partners involved on the development by one or more organizations;
of a technical specification making it possible to • The possibility of having broker space for
publish and retrieve information, the companies’ receiving submissions before the publishing
profiles and the provided services. and validating the data is offered. This space
makes it possible to correct and modify
objectives of the discovering information with some errors.
component
directory Functionalities
The initial goal of this module is to provide a
shared and universal place where each company, The directory must provide to the various com-
whatever its size, its sphere of activity and its lo- panies a set of functions and methods to ensure
the following functionalities:
334
1. Exchange messages and business docu- the information system. For this purpose, the
ments; integration of the various applications must be
2. Manage business relationships between implemented correctly.
companies; However, the objective is the piloting of the
3. Communicate data in a standard format; architecture of the exchange model by business
4. Define and save the business process; processes. This part of the model is fundamen-
5. Retrieve the ontology model defining the tal because it is the basis of autonomy and the
contents; agility given to the company in the evolution of
6. Download and install tools for manag- its information system. The assembly and the
ing the various existing ontologies. dismantling of the business functions are made
available to the users from the system without
business Process ontology layer having to call upon the usual supervision of the
technical managers of the information system.
Once the exchange partner is located, it is neces- The goal desired by including various tools
sary to agree on the various steps of the exchange and business process models in this layer is to
scenario from the business process. A business provide a generic and a flexible architecture.
process represents an ordered set of interactions. The characteristic of this multilevel model is in
Each interaction represents an atomic exchange the integrality and the homogeneity of the tools
of messages between partners (Tellmann & and methods which are provided to create the
Meadche, 2003). The emergence of business necessary business processes. These tools will
process modeling standards makes it possible to be discussed in the case study section.
create a rationalization of the analysis methods of
the processes and the creation of a community of business Process ontology
knowledge shared by all exchange actors. Functionalities
The description of the business processes must
be based on formalism and a clear method, which The functionalities of business process ontology
allow a real sharing of this knowledge by the vari- are described as follows:
ous actors (business or computer science). This
formalism must integrate a typology of process, a • Formalize collaborations and the dialogues
strategy of rigorous decomposition which allows between partners based on rules defined by
a homogenization and an optimal visibility of all the process model used;
levels of responsibility. A set of basic rules and • Increase the effectiveness and the quality
ways of making the tools, adapted to the contexts of the exchanges between partners;
of the company—for example, tools like UML, • Ensure interoperability between the vari-
XML (Peltz & Murray, 2004)—make it possible ous processes which reference the various
to answer effectively the customer’s requirements contents;
involved in such exchange. • Ensure the coordination based on the shared
resources. This is related to the scheduling
objectives of the bP ontology and the control of the shared resources (a
component resource is an informational entity handled
by the business process);
The role and all tasks of the company are rep- • Manage the choreography of the business
resented by a set of processes. These processes activities.
are in close connection with the components of
335
The goal of Web services is to achieve a uni- pany, and the other is shared by all the actors. This
versal interoperability between applications using last is called ontology in our case. Indeed, the
the Web standards (Randy & Hall, 2001). This is essential objective of the integration of ontologies
why they are basically integrated in the descrip- in the exchange model is to ensure a reliable and a
tion of business processes. Indeed, the automation consensual sharing and managing of data during
and the management of business processes were exchanges. This is provided by ontologies since
historically a difficult challenge for the companies they allow a mature product characterization by
because of the inflexibility of their data-processing ensuring a unique identification of the concepts
infrastructure. The Service Oriented Architecture and their attributes. This identification ensures
(SOA) provides this flexibility. It gives to the a non-ambiguous concept referenced by various
processes the necessary abstraction level for both partner’s applications. Thus, the model integrating
the definition and the execution of the process. ontologies will be characterized by the possibil-
The advent of SOA makes integration disappear ity of separating the contents from the container
as a distinct activity. (process) allowing a separate development on
each level in a flexible way.
content ontology
content ontology Functionalities
Once the partner is localized and the exchange
process well established and defined, it remains The layer which manages the contents specifies
to have a consensus on the data (catalogues of the structure and the semantics of the data as well
products) which will be exchanged. It should be as the refinement constraints and the composition
also guaranteed that the information used by the of the properties such as cardinality. Two types
various systems during exchange is interpreted of ontologies can be used:
without ambiguity.
The management of contents in general cov- 1. The business oriented ontology has as a role
ers, in the data-processing field, the set of steps the identification of the properties related
governing the life cycle of the documents available to the company business and is indepen-
on the information system: from their creation dent from any product or component. This
and validation to the management of the publica- information is necessary to achieve the
tion rules. Contents indicate any information and collaboration and the transactions between
knowledge of a company intended to be exchanged partners.
between several systems. More precisely, contents 2. The product oriented ontology represents
are all document, component, data which can useful information (payload) defining the
be involved as a part of the useful information object of exchange or the products catalogues
(Pierra, 2000). on which the transactions will occur. It is,
in general, a set of classes, properties and
objectives of the content ontology values domains (instances of objects).
The contents management in companies and The utility of this ontological level in the
organizations became a major requirement these exchange model is to provide a common knowl-
last years. From the architecture point of view, a edge sharing of the products used by the actors
content management system intended to be used of exchange and to make sure that the interpre-
by several heterogeneous infrastructures, requires tation of the data will be done in a correct and
two sub-systems, the first is backend to the com- non-ambiguous way.
336
The idea to separate the business contents The following sections describe in more detail
from the useful contents is not completely new. the several parts of the architecture.
RosettaNet uses this principle by proposing two
separated dictionaries, one representing the busi- ebxml registry
ness ontology (RosettaNet Business Dictionary
RNBD) and the other representing the products on- The ebXML registry is central to the ebXML
tology (RosettaNet Technical Dictionary RNTD) architecture (Kappel & Kramler, 2003). The
(RosettaNet, 2001b). Let us note that PLIB (Jasper registry manages and maintains the shared in-
& Uschold, 1999) can be used to define separately formation as objects in a repository. Repositories
these two types of ontology either separately or provide trading partners with the shared business
in the same dictionary. semantics, such as BP models, core components,
Once the three defined levels for the model and messages, trading partner agreements, schemas,
the functionalities and the objectives of each part and other objects that enable data interchange
are established, the integration of these different between companies. The ebXML registry is an
modules describes the global exchange model. interface for accessing and discovering shared
The following sections describe the technical business semantics. In this section, we explain
choices adopted and the proposals made for each the registry usages, the business semantic model,
level of the model. and the registry functionality including registry
classification, registry client/server communica-
tions, searching for registry objects, and managing
cAse study registry objects.
Our registry implementation is based on in-
Our work proposes the implementation and the formation in the primary ebXML registry refer-
realization of a B2B platform based on the three ence documents, including the “ebXML Registry
layers described in the previous section. The next Service Specification (ebXML, 2002b)” and the
step is to give answers to the three previously “ebXML Registry Information Model (ebXML,
outlined questions to design a practical solution 2002a).” In a marketplace populated by computer
implementing these different levels. companies with proprietary hardware, operating
First of all, in the discovering layer of our systems, databases, and applications, ebXML
system, gives business users and IT groups control over
their lives. The ebXML registry is not bound to a
1. An ebXML registry (ebXML, 2002b) is database product or a single hardware vendor. It is
developed for this level, in the business designed to interoperate on all kinds of comput-
process layer, ers. An ebXML registry serves as the index and
2. A Web service, based on the PIP2A9 Roset- application gateway for a repository to the outside
taNet BP for product information queries, world. It contains the API that governs how par-
is implemented, and finally, ties interact with the repository. The registry can
3. The content layer uses a PLIB model (ISO also be viewed as an API to the database of items
13584-42, 2003) with its various tools that supports e-business with ebXML. Items in
(PLIBEditor and PLIBBrowser) to search the repository are created, updated, or deleted
and to retrieve the product information through requests made to the registry.
content. Once this part of system is developed (Takher-
oubt & Boudjemai, 2004), the second task is to
define a model to describe business process ontol-
337
ogy in both conceptual and technical ways. Web RosettaNet aims to align the BP of supply chain
services are a new paradigm which is used to de- partners. This goal is achieved by the creation
scribe technical business processes (Speck, 2002). of Partner Interface Processes or PIPs. Each PIP
In the next section we give a short overview about defines how two specific processes, running in
this technology and a comparison with ebXML two different partners’ organizations, will be stan-
BPSS and a RosettaNet PIP in order to provide dardized and interfaced across the entire supply
a new model to be used by our proposed model. chain. PIP includes all business logic, message
flow, and message contents to enable alignment
Web service for the bP ontology of the two processes. RosettaNet defines more
layer than one hundred PIPs. The purpose of each PIP
is to provide a common business/data model and
Web services are modular, self-describing applica- documents, enabling system developers to imple-
tions that can be published and located anywhere ment RosettaNet e-Business Interfaces.
on the Web or on any local network. The provider The PIP studied in this article is the PIP2A9
and the consumer of the XML Web service do (Query Technical Product Information) (Roset-
not have to worry about the operating system, taNet, 2001c). The technical product information
the language environment, or the component is the category of information that describes the
model used to create or to access the XML Web behavioural, electrical, physical, and other charac-
service, as they are based on ubiquitous and teristics of products. There are numerous classes
open Internet standards, such as XML, HTTP of customers within the supply chain that need to
or SMTP. An initiative from Microsoft and IBM be able to access product technical information.
to describe the messages between clients and the These include distributors, information providers
Web server, WSDL (Web Service Description (such as Web portal companies, other commer-
Language) (Brydon, Murray, Ramachandran, cial information aggregators, and end-customer
Singh, Streans, & Violleau, 2004), describes and information system owners), engineering, design
defines Web services. It helps the user to set up a engineering, manufacturing, and test engineering.
system using a service ranging from connection As mentioned previously, our work proposes
details to message specification. a definition of a B2B architecture followed by
A WSDL document defines services as a set an implementation of this platform based on the
of network endpoints (ports) that is associated three technologies shown in the above sections.
with a specific binding. This binding maps a The next step uses these standards conjointly to
specific protocol to a port-type composed of one provide a secure, a reliable and an interoperable
or more operations. In turn, these operations are architecture. Each partner providing a service
composed of a set of abstract messages, represent- must affect a URL to its BP. This URL is stored
ing the data. The pieces of data in a message are in a registry (UDDI or ebXML registry). In our
defined by types. case an ebXML registry is developed for this
purpose (Glushko, Tenenbaum, & Meltzer, 1999).
rosettanet business Process The BP will be discovered and retrieved as a set
or a unique Web service. The activities that may
This section presents the mechanisms and the be undertaken during this step are:
approach used to design the Web service using
the BP from RosettaNet named PIP2A9. First, the 1. The use of a RosettaNet BP (in our example,
PIP2A9 role and tasks are introduced, then some the PIP2A9 is a BP for product technical
details about the matching between PIP2A9 and information query );
Web service are outlined.
338
2. The integration of PIP2A9 in the ebXML cess is defined by a set of documents exchanged
BP model; on collaborations. Collaborations represent PIPs,
3. The development of a Web service based on and the business documents define the contents.
the resulting BP. This model will be implemented using the Web
Services technology (see Web service section)
Integrating PIP2A9 in the ebxml according to the transformation realized as shown
bPss model the figure 3.
Figure 3 shows the description of the BP (part
The next task will be the integration of the re- A) as a Web Service via WSDL document (part
sulting PIP2A9 in the ebXML BP. This task will B). A PIP represents a document exchange as a
be achieved without any error if the mapping request/ response messages. Each communication
between RosettaNet PIP and the corresponding is described by a document which defines content
part of the ebXML BP is given correctly. A PIP for both request and response.
corresponds to a Binary collaboration or exactly
to a business transaction in the BPSS specifica- PlIb ontology in the content layer
tion. A business transaction in ebXML is defined
by a business transaction activity with document The third and the last level of our architecture is
flows exchange based in general on a request and the content layer. As shown previously, this layer
a response documents. Therefore, each PIP in the defines data and products information for which
RosettaNet model will be integrated in the BPSS the use of PLIB ontology model is proposed. PLIB,
model as a business transaction activity. Figure 2 the Parts Library standardisation initiative, was
defines an excerpt of a BPSS with a PIP example launched at the ISO level in 1990. Its goal is to
document. develop a computer-interpretable representation
Figure 3 shows how the integration is realized of parts library data to enable a full digital infor-
using ebXML and RosettaNet standards in the mation exchange between component suppliers
same process model. An ebXML Business Pro- and users. A PLIB data dictionary (ISO 10330-
Figure 2. An integrated ebXML and PIP business process

Content (Instances)
ebXML Business Process
Business Document
Business Document
Description of exchanged Documents
Business Document
Description of partner’s roles ..

.
Description of transactions Business Document
Description of collaborations
Collaboration 1 RosettaNet (PIP)
Collaboration 2 RosettaNet (PIP)
……..
Collaboration n RosettaNet (PIP)
339
Figure 3. PIP transformation on WSDL document

Exchange with PIP WSDL document
Participant A Participant B 3 <message name= PIPRequest,……>
Req <Part name= PIPRequest,…HREF=..//Req.>
PIPRequest ……
1 <Part name= Attribute,….>
……
…….
………………….
<message name=PIPResponse,…HREF=..//Rep.>
Signal
<Part name= Rep,….>
Rep <Part name= Attribute,….>
…… 2
……
……. PIPResponse Transformation 4
Signal
(a) (b)
22, 1997; Shenck & Wilson, 1994) is based on sible to create and publish ontology (Shenck, &
the object oriented concepts (Coad & Yourdon, Wilson, 1994; Westarp et al., 1999).
1992): components are gathered in parts families PLIBEditor is a tool which allows users to
that are represented by classes. This set of classes define and to represent graphically and simulta-
is organised in a simple hierarchy of classes on neously a defined ontology and its instances. It
which factorisation/inheritance applies. Such is an application containing two parts on its user
classes are then, precisely described textually, interface, the left part (frame) allows the definition
with technical drawings. of classes, properties and relations between them,
Finally, each class is associated with a set of and the right part defines the ontology population
technical properties, also precisely described (instances). In another way, PLIBEditor defines in
(domain of values, possible measurement unit…). the same application, data and metadata of ontol-
A basic idea of the definition of a PLIB dictionary ogy. PLIBBrowser has the same role as PLIBEditor
is that properties and classes shall be defined si- but it has a Web oriented presentation.
multaneously: applicable properties allow defining
precisely a parts family, and conversely, a parts PlIb characteristics
family determines the meaning of the property
in its particular context. This section presents the three main characteris-
The modelling formalism used in PLIB is the tics that motivate the use of the PLIB ontology/
EXPRESS language (ISO 10330-11, 1994; ISO dictionary model in our infrastructure.
10330-22, 1997). The ontology model of PLIB is
formally defined and a number of tools have been Separation of Definitions and
developed to create, validate, manage or exchange Identifications
ontologies. They are available on the PLIB server
(http://www.plib.ensma.fr) at LISI/ENSMA. The Business partners or software involved in a B2B
basis of all these tools is a representation of on- transaction must have some common understand-
tology in a processable exchange format. It can ing of the various pieces of information they use
be exchanged in various automatically generated during an exchange. Therefore, descriptions and
formats: XML document (SimPLIB DTD), associ- definitions of each piece of exchanged information
ated possibly with XSL page (SimPLIBVIEWER), must be shared between the parties involved in the
EXPRESS physical file or DHTML document exchange. Otherwise, they cannot understand each
(PLIBBrowser). Finally PLIBEditor makes pos- other. PLIB ontologies allow partners to share and
340
use the same concepts without transferring their Axis uses WSDL2Java tool which generates a
description each time an exchange is performed. set of skeleton classes and interfaces. The latter
In order to allow this, PLIB separates objects and/ is used as containers for the developed Java code
or concepts identification from their description. of our specific Web service.
One can imagine what description that is local to Once the classes are increased with a PIP2A9
each partner may be, because it was exchanged specific codes (Brydon et al., 2004), a set of Java
earlier. Then, identifications need to be exchanged. Server Pages (JSPs) are developed to test the
service. The general schema of the developed
Universal Identification platform is shown in figure 4.
Since we use in our platform, the new database
One important characteristic of PLIB consists in model (OBDB) cited above, and in order to manage
providing a unique global identifier named BSU data, a new query language is required. The same
(Basic Semantic Unit) for each concept (class, team at LISI laboratory is currently developing a
property, etc.) of the ontology (Pierra, 2000). The new query language. It is named Ontology Query
BSU gives a unique worldwide identification for Language (OntoQL) (Jean, Pierra, Ait-Ameur,
each concept, class, and property, defined in the 2005) and a preliminary version runs on several
ontology. As a result, referencing these concepts OBDB implementations. Among these applica-
during exchange is unambiguous and easy. It is tions of OntoQL, we find the usage we propose in
not necessary to get the whole concept or object; our platform. Moreover, in our architecture, when
its BSU is enough to reference it. the customer tries to select products, he must cre-
ate a query to retrieve the required product. This
Orthogonality query is received and transmitted by our Web
service. The Web service we developed supports
Orthogonality results from universal identifica- this kind of query. To simplify the query creation
tion. Indeed, this concept leads to a total indepen- for the consumer, PLIBEditor has been extended
dence between the two upper level layers. One in order to support a user friendly querying user
may use any confusion of any business protocol interface. This extension allows a user to inter-
(PIPs from RosettaNet or ebXML BP) for the BP actively create the request.
ontology layer, together with any PLIB ontology For more details and comprehension, the sce-
(IEC 61360-4 for electronics components, ISO nario on the client side is described as follows.
13584-511 for fasteners, or ISO 13399 for cutting An interface creates the skeleton of the service
tools) for the products ontology layer. Thus, each implemented using the J2EE platform. It repre-
business process of the Business Process layer can sents a set of classes implementing the interface
use any dictionary of the product ontology layer of the Web service functions. This interface is
in its payload (McGuinness, 1999). responsible of displaying the results in a specific
format according to the specific desire of the cus-
tomer. In our case, PLIBEditor makes it possible
tecHnIcAl InFormAtIon to either create the OntoQL query or display the
results when returned from the server side. Notice
Based on the WSDL (Peltz & Murray, 2004) that the returned content is specific to the OBDB
document, several Web services will be designed model: a set of files representing the products
using the Jbuilder development environment in a PLIB defined format (EXPRESS format)
with the Enterprise version and Axis Box Tool (Schenck & Wilson, 1994). For other formats,
for managing all Web services (Heather, 2001). the customer either creates a specific interface or
341
Figure 4. The developed platform
uses its own proprietary existing applications for kinds of ontologies simultaneously in a business
data visualization (Hunter & Grawford, 2002). transaction enables a secure, automatic, and reli-
able B2B exchange.
Based on these two kinds of ontologies and on
conclusIon the standards mentioned above, a new architecture
is proposed. The use of ontology to describe con-
We conclude that many organisations are increas- tent provides a uniform representation of data in
ingly searching for mechanisms and methods to a common and shared format. Moreover, we have
achieve exchange of product catalogs and services shown how B2B architecture can be abstracted and
with minimal cost and without modifying their represented by an adaptive infrastructure based
local systems. Several architectures and standards mainly on three layers. These three parts define
try to accomplish this task both in horizontal and the major concepts required for exchange between
vertical situations. E-commerce consortia like companies. Furthermore, the article has shown
ebXML and RosettaNet are developing standards the need for having a common shared ontology
and initiatives in order to allow interchange of for Business Process used between partners. We
business information among information systems. have developed a Web service platform as ker-
Ontologies aim at providing a shared computer- nel for BP which was integrated to the proposed
sensible representation of domain knowledge, architecture in order to allow independence and
allowing exchange of semantic information across scalability.
heterogeneous systems. Our case study presents the possibility to in-
The article shows that in most exchange archi- tegrate in the same architecture several parts of
tectures, two categories of ontologies are used: several standards simultaneously. For example,
ontologies of products and services and ontolo- we have used ebXML and RosettaNet to design
gies of business processes. Involving these two the BP ontology, PLIB dictionary to describe the
342
content ontology and in the last part, the ebXML Dogac, A., Cingil, I., Laleci, G. B., & Kabak,
registry as a directory to localize and retrieve I., (2002). Improving the functionality of UDDI
services and data of partners. For the future, we registries through Web service semantics. Pro-
suggest developing a new collaboration model ceedings of 3rd VLDB Workshop on Technologies
(Khelifa, Aklouf, & Drias, 2008), to be added E-Services (TES-02), Hong Kong, China, August.
to the architecture based on the Grid Services
ebXML Registry Information Model Schema
technology and also studying the possibility to
(2002). Retrieved from http://www.ebxml.org/
integrate other ontology formalisms like OWL-S
specs/ebRIM.pdf
and DAML-S to our platform.
ebXML Registry Services Specification Version
2.1 (2002). Retrieved from http://www.ebxml.
reFerences org/specs/ebRS.pdf
ebXML. (2001). ebXML technical architecture
Aklouf Y. & Drias, H. (2007). Business process specification. May 11. Retrieved from http://
and Web services for a B2B exchange platform. www.ebxml.org
International Review on Computers and Software
ebXML. (2005). Business process team. Business
(I.RE.CO.S). Prize Worthy Prize.
Process Specification Schema v2.0 February.
Aklouf, Y., Pierra, G., Ait Ameur, Y., & Drias,
Gartner Group. (2000). OAGI: Fostering standards
H. (2003). PLIB ontology for B2B electronic
evolution or revolution?. A Gartner Advisory
commerce. 10th ISPE International Conference
Research Note, December 14.
on concurrent engineering: Research and Appli-
cations. Carlton Madeira Hotel, Madeira Island Glushko, R. J. Tenenbaum, J. M., & Meltzer, B.
– Portugal, July 26 – 30. (1999). An XML framework for agent-based e-
commerce. Communications of the ACM, 42(3).
Aklouf, Y., Pierra, G., Ait Ameur, Y., & Drias,
H. (2005). PLIB ontology: A mature solution Heather, K. (2001). Web Service Conceptual Ar-
for products characterization in B2B electronic chitecture 1.0, May 2001, IBM Software Group.
commerce. Special Issue: E-Business Standards.
Hunter, J. & Grawford, W. (2002). Servlet java
International Journal of IT Standards and
guide du programmeur, Edition O’REILLY.
Standardization Research, 3(2). IRMA& IGP
publishing. ISO 10303-11. (1994). Industrial automation sys-
tems: and integration — Product data representa-
BIC. (2001). XML convergence workgroup:
tion and exchange – Part 11: Description Methods:
High-Level Conceptual Model for B2B Integra-
The EXPRESS language reference manual.
tion, Business Internet Consortium, Version: 1.0,
October 5. ISO 10303-22. (1997). Industrial automation
systems and integration. – Product data repre-
Brydon, S., Murray, G., Ramachandran, V., Singh,
sentation and exchange– Part 22: Implementation
I., Streans, B., & Violleau, T. (2004). Designing
methods: Standard Data Access Interface.
Web services with the J2EE™ 1.4 Platform:
JAX-RPC, SOAP, and XML Technologies’. Sun ISO 13584-42. (2003). Industrial automation sys-
Microsystems, January tems and integration. Parts library– Methodology
for structuring Parts Families, ISO, Geneva.
Coad, P. & Yourdon, E. (1992). Object-oriented
analysis. , Englewood Cliffs, N.J.: Prentice Hall.
343
Jasper, R. & Uschold, M. (1999). A framework for RosettaNet. (2001). RosettaNet architecture con-
understanding and classifying ontology applica- ceptual model, July.
tions. In B. Gaines, R. Cremer, & M. Musen (Eds.)
RosettaNet. (2001). RosettaNet Implementation
Proceedings 12th Int. Workshop on Knowledge
Framework: Core Specification Version: Validated
Acquisition, Modeling, and Management KAW’99
02.00.00, July 13.
(16-21 October 1999, Banff, Alberta, Canada),
volume I, pages 4–9–1— 4–9–20, Calgary. Uni- RosettaNet. (2001). Specification: PIP Specifica-
versity of Calgary, SRDG Publications. tion Cluster 2: Product Information, Segment A:
Preparation for Distribution, PIP2A9: Query Tech-
Jean S., Pierra G., Ait-Ameur Y. (2005). OntoQL:
nical Product Information, Validated 01.01.01,
An exploitation language for OBDBs VLDB PhD
November 1.
Workshop, 29 Août 2005.
Schenck, D. & Wilson, P. (1994). Information
Kappel, G. & Kramler, G. (2003). Comparing
modelling the EXPRESS way. Oxford University
WSDL-based and ebXML based approaches for
Press.
B2B protocol specification. Martin Bernauer,
Business Informatics Group. Vienna University Spek, A.W.A. (2002). Designing Web services in a
of Technology, Austria. business context. Master Thesis, University of Van
Tilburg. Center of Applied Research, September.
Khelifa, L., Aklouf, Y., & Drias, H. (2008). Busi-
ness process collaboration using Web services Takheroubt, M., & Boudjemai, R. (2004). Etude
resource framework. To appear in Proceedings et Mise en œuvre du registre ebXML dans une
of International Conference on Global Business plate de commerce B2B. Master Thesis. USTHB
Innovation and Development (GBID), Rio de Algeria.
Janeiro, Brazil, January 16-19.
Tellmann, R., & Meadche, A. (2003). Analysis
McGuinness, D. L. (1999). Ontologies for elec- of B2B standard and systems. SWWS, Semantic
tronic commerce. Proceedings of the AAAI ’99 Web Enabled Web Services.
Artificial Intelligence for Electronic Commerce
Trastour, D., Preist, C., & Colemann, D.(2003).
Workshop, Orlando, Florida, July.
Using semantic Web technology to enhance cur-
Peat, B. & Webber, D. (1997). XML/EDI - The rent business-to-business integration approaches.
E-business framework, August 1997, Proceeding of the Seventh IEEE International
Entreprise Distributed Object Computing Con-
Peltz, C. & Murray, J. (2004). Using XML sche-
ference, (EDOC’03).
mas effectively in WSDL design. Proceedings
of the Software Development Conference and Westarp, F. V., Weitzel, T., Buxmann, P., & König,
Expo, March. W.(1999). The status quo and the future of EDI -
Results of an empirical study. In Proceedings of
Pierra, G. (2000). Représentation et échange de
the European Conference on Information Systems
données techniques, Mec. Ind., 1, 397-400.
(ECIS’99).
Pierra G., Dehainsala H., Ait-Ameur Y., & Bel-
latreche L. (2005). Base de Données a Base On-
endnote
tologique: Principes et mise en oeuvre. Appeared
in Ingénierie des Systèmes d’Information (ISI). 1
Ontology defines a shared conceptualiza-
tion. In the case of e-business, it provides a
Randy, E. H. (2001). W3C Web service position
conceptualization of the e-business concepts
paper from Intel. W3C Workshop, April 12.
344
on which business partners (i.e., companies,

traders, vendors) have to agree to understand
each other. By specializing and instantiating
concepts and relations of the ontology for a
particular case, a specific business process
may be derived in a precise and structured
way.
This work was previously published in Always-On Enterprise Information Systems for Business Continuance: Technologies for
Reliable and Scalable Operations, edited by Nijaz Bajgoric, pp. 277-293, copyright 2010 by Information Science Reference
345
346
Chapter 2.3
A Structured Approach
to Developing a Business
Case for New Enterprise
Information Systems
Francisco Chia Cua
Otago Polytechnic, New Zealand
Tony C. Garrett
Korea University, Republic of Korea
IntroductIon the agenda. After the agenda-setting stage is the

examination of the available options. Attributes of
The term business case is used to describe both competing options are matched together, enabling
a process and a document. A business case ex- attitude formation in favour or against a particular
ploits an initiative. Exploiting the initiative from option. This results in the creation of a shortlist
awareness to implementation encompasses a of two or three options. A decision is generally
process, referred to in the diffusion of innovation reached at this point. The decision is, therefore,
parlance, as the innovation-decision process. The part of the matching stage. However, this is not
development of a business case concerns this always true in an organisational setting. There
innovation-decision process. The individuals is a third stage after the matching stage. It is the
or the decision-making units pass through the decision (aka, business case) stage. Organisations
innovation-decision process, gaining knowledge generally demand rigour in making the decision.
of a new idea, forming an attitude toward it, and A business case document embodies the rigour in
deciding whether to adopt or reject it (Rogers, the business case development. Consequently, the
2003, p 20). Gaining the knowledge triggers the decision stage culminates with a completed busi-
awareness or enforces it. Then, it leads to setting ness case document and the decision that results
from it: to adopt or reject the innovation. The three
DOI: 10.4018/978-1-60566-026-4.ch574 stages, agenda setting, matching, and decision
A Structured Approach to Developing a Business Case for New Enterprise Information Systems
stages, compose the initiation phase. If the decision big picture of business case development in the
favours adoption, then the implementation phase article, The Role of Business Case Development
proceeds. In the context of implementing the new in the Diffusion of Innovations Theory for EISs
enterprise information systems, the stages in the (hereinafter referred to as The Role of Business
implementation phase consists of pre-production, Case Development...). The structure, suggested
production, post-production (that is, maintenance), in Table 2, delineates the context to the new en-
and confirmation stages. In summary, the busi- terprise information systems. However, prior to
ness case development is a means, and its end is that, certain issues must be addressed:
a business case document.
A complete business case is a formal written • What is the purpose of a business case?
argument and a detailed “point by point” analysis • What should a business case document
(Cannon, 2006, p 4; Carruth, 2001). It purports to contain?
justify the adoption or rejection of investing and • How should the business case be struc-
thereby, implementing the new enterprise informa- tured? How should it be written?
tion systems. The analysis takes into consideration
the stakeholders (Ministry of Health, March 2005),
especially the decision-makers and the end-users. tHe PurPose oF A
Consequently, a business case document is formal, busIness cAse
detailed, and complex.
Using the parlance of “diffusion of innova- A good business case must have a purpose that
tions” (DOI) theory (Rogers, 1962, 2003), a busi- is clear, specific, and relevant to the organisation
ness case document is a communication tool used and the upper managers. The Role of Business
to diffuse the new enterprise information systems, Case Development... mentions “growth and sus-
and to justify their adoption and implementation. tainability” as a strategic goal. That term is too
Diffusion refers to the process by which the execu- broad to be useful in a business case. A similar
tive sponsor, who owns the innovation-decision ambiguous construct is sustainable competitive
process of the new enterprise information systems, advantage (Hammer, 1996; Monczka, Carter,
communicates to the upper managers to get their Petersen, & McDowell, 2006, p 213). A single
approval of the project and funding. detailed statement is far more relevant to the
Diffusion via a business case document for organization than several broad statements. An
a technological product, such as an enterprise actual case study reveals one objective of selecting
information system, must be directed at a single and implementing a proven, up-to-date enterprise
target audience to be effective. The upper managers financial system. The chosen systems must have
represent a chasm that needs to be bridged (Moore, the capacity to meet likely future financial-related
1991). A completed business case document, requirements and growth. This vague objective can
containing relevant information for the managers, be made clearer by citing sustainable competitive
can serve as that bridge. advantages that are strategically valuable to the
In addition to its relevance, a business case organisation, taking into consideration certain
must also be responsible and credible. Therefore, guiding principles (Table 1).
the business case must bring relevance, reputa- Enterprise information systems are the en-
tion, and responsibility (the 3 Rs) into a number abling technologies that foster sustainable com-
of issues and challenges during its development. petitive advantages under certain guiding prin-
This chapter proposes a business case structure, ciples. Table 1 helps to develop the business case
with the 3Rs underlying it. It continues from the backward from the purpose, and includes ALL
347
Table 1. Guiding principles and sustainable competitive advantages
Guiding principles Sustainable competitive advantages

• E-business supply chain • Seamless integration
• Economic value-added focus • Best performance: Quality, price, delivery, technology, cycle time
• Globalisation (velocity, responsiveness, service), safety
• Satisfaction of the needs of customers • Enhanced EVA (increase revenue by broadening the offerings
• Total value management and improving customer value, reduce internal and external cost
• Value/supply chain integration, productivity, and collaboration structure, reduce assets, and improve asset utilization)
(operational excellence and process redesign) • Enhanced EBIT, ROI, cash flows
• Perceived highest customer value
• Revenue generation
• Time-to-market/breakeven
Adopted from Carruth (2001); Hammer (1996); Ministry of Health (2005), Monczka et al. (2006)
Table 2. Suggested structure of a business case
Section Caption Remarks

E Title Page
E Table of Contents
E Executive Summary Prepare a sharp and compelling summary.
E 1 INTRODUCTION Brief, purpose, scope, limitations
E Terms of Reference or Background State the brief (that is, who asks to do what?). What do the readers need to
know? If the background is crucial, then there can be a separate Back-
ground section.
E Purpose of the Report Briefly explain the purpose of the report.
Gaps Analysis Describe the expected consequences (the strategic vision in refer to Section
4). Describe the assessment of the present state (refer to Section 4). De-
scribe the gaps between the present state and the desired future state.
E The Proposed Solutions and Its Highlight the strategic value, the selling point, of the solution. Describe the
Strategic Value relevant issues and their action plans (related to Section 7, the proposed
project). Some questions to ponder are: Is the solution reactive or anticipa-
tory (Figure 3 in The Role of Business Case Development...)? How does
the proposed solution FIT into the big picture? If the solution is reactive,
describe the urgency of the radical change. If the solution is anticipatory,
explain the critical strategic area (Table 1 and strategic vision in Section 4
and Section 6). Why does the solution matter?
E Project Ownership (and Consultant) Corporate power dictates the level of detail this section requires (assumed).
Who is the executive sponsor? How supportive is the executive sponsor to
the business case? Who is the expert the executive sponsor has consulted?
E Scope and Delimitation Delineate the scope of the problem.
O Methodology Briefly describe the methods or methodology of Section 3.
E Structure of the Business Case Briefly outline how the business case is presented.
Document
2 CONCLUSIONS and recommenda-
tions
E 2.1 Conclusions How does the executive sponsor evaluate the Business Case (eg, cost-
benefit analysis)? How logical or intuitive is the evaluation? Based on the
financial and nonfinancial analysis, summarise briefly how the business
case will impact positively and negatively on the organisation? What are
the critical success factors? The critical failure factors? How are the options
considered and documented?
continues on following page
348
Table 2. continued
O 2.2 Recommendations Which option has the executive sponsor chosen and why? The Recom-
mendations section (or the Proposed Project section) may attempt to answer
further questions such as: How does the executive sponsor propose to rally
the support, involvement, and usage? How does the executive sponsor
ensure success?
O 3 METHODS What are the contingency plans? How does the executive sponsor come
about the plan? How reliable is the evidence used to develop the Business
Case?
O 4 Facts and Assumptions This section may form part of the Introduction section.
O Strategic Vision Identify the strategic vision, the guiding principles, and sustainable com-
petitive advantages (Table 1). Assess the key strengths and weaknesses.
This may be incorporated under the future state of the Introduction section.
O Needs Analysis Describe briefly the needs and relevant issues. This section may be incorpo-
rated under the Gaps Analysis of the Introduction section.
O Options Analysis Describe briefly the options considered in this business case. This section
may be incorporated under the Proposed Solutions of the Introduction sec-
tion.
5 Financial Analysis How does the EXECUTIVE SPONSOR explore the sensitivity of key as-
sumptions in the analysis?
E Total Cost of Ownership (TCO) Combine the net present value, discounted cash flows, and total cost of
ownership in the financial analysis. Determine which option enhances the
value to the organisation and not which option has the lowest TCO. Make
sure to scrutinise all “hidden costs.”
E Business Risks How does the EXECUTIVE SPONSOR identify, assess, and resolve the
risks involved?
E 6 NonFinancial ANALYSIS What external factors drive the Business Case? How will the competitive
advantage, macro environments, and stakeholders (competitors, customers,
and vendors) impact the organisation? With what of key assumptions in the
analysis?
E Sustainable Competitive Describe in detail the innovation, its scope, its goals (the sustainable
Advantage(s) competitive advantages in Table 1), and the expected consequences. How
does the innovation fit into, for example, the seamless alignment or the
organisation’s strategy? How does the innovation enhance the competitive
advantage? How does it create the differentiation of the organisation from
its competitors? What is the “compelling reason” why the innovation must
be put into action? What value does the innovation provide?
E Macro environments
E Stakeholder Analysis St Gallen Management Model suggests two approaches. One is the “stra-
tegic stakeholder value.” The other is the “ethically critical stakeholder
value.” The former approach assumes that a balanced consideration of
the long-term interests of all stakeholders is the best way to maximise
shareholder value. The latter approach evaluates all potential stakeholders
equally toward “ethically justifiable legitimacy.” (Rüegg-Stürm, 2005)
E 7 The Proposed Project
Project Plan How does the project plan look like? What are the key milestones? How
abstract or detailed is the project plan? How does the project plan mitigate
the risks of occurrence of undesirable consequences (refer to Appendix A
section)?
The innovation-decision process How did the executive sponsor conduct the initiation phase? What is the
change strategy? How will the executive sponsor implement the new enter-
prise information systems? Reiterate the change strategy, costs, and risks?
349
Table 2. continued
Project Participants How detailed has the Business Case been documented with regards to the
resourcing requirements (that is, the capacity and capability requirements)?
Specifically, what are the skills, experience, and time commitment required
of the project team? What are the skills, experience, and time commit-
ment required within the business after the completion of the innovation-
decision process? How does the executive sponsor propose to manage the
risks related to the resourcing requirements? How does the Business Case
take into account the additional resourcing required through and after the
implementation-decision phase?
E Appendix A UNDESIRABLE CONSEQUENC- Refer to Project Plan section above.
ES TO AVOID
E Appendix B REFERENCE SITES
Legend: E = Essential, O = Optional
the planning activities, resources, and metrics that extended social systems. The systems empower
are critical to the goal (Ministry of Health, 2005, the internal people in the organisation to provide
p 9). Recent research indicates that most success- the best performance with shared information
ful organisations have a crystal clear notion of systems. The systems help to identify and develop
the organisation’s strategy, and how deploying centres of excellence with resources and expertise
information technologies can help actualise that prior to outsourcing. The systems also maximise
strategy. return on information systems investment across
the organisation, maximise the exploitation of op-
A business case Is a portunities, and minimise the risk associated with
two-sided coin the new implementation. All these technological
advantages are about the innovation. However,
The purpose of a business case is to reflect the that is not the purpose of the business case, which
rigour of planning relative to the level of invest- is a reflection of the rigour on the innovation-
ment being undertaken (Ministry of Health, 2005, decision process.
p 3). This is one side of the coin. The other side
relates to justifying the innovation discussed.
On one side, there are sustainable competitive WHAt sHould A busIness
advantages or other reasons to justify the imple- cAse document contAIn?
mentation of the innovation. On the other side is
the purpose of the business case. Justifying the the scope and content of
implementation and matching the best fit pertain business cases vary
to the what and the why questions. Thinking about
the innovation-decision process and substantiating Not only must the business case contain the jus-
the process with the necessary rigour concern the tification of the innovation (what and why) and
how (and why) questions. In the context of the the rigour of the process (how and why), it must
new enterprise information systems, the specific be relevant to the stage of the innovation-decision
sustainable competitive advantage represents process at which it is written. A business case
the primary goals. The enterprise information written at the awareness stage (Stage 1) or match-
systems enable seamless integration (Table 1) of ing stage (Stage 2) differs in scope and content
information across the whole organisation and its with a business case written at the decision stage
350
(Stage 3). For example, the Ministry of Health quence of strategically investing or not investing
(2005) of New Zealand has issued guidelines the chosen option (Weill & Ross, 2004).
for the investment in information technology. Its
business case development is composed of three managing risk and defining
stages equivalent to the three stages in the initia- undesirable consequences
tion phase: the strategic analysis, options analy-
sis, and completed business case. Respectively, Aside from the in-depth quality analysis, the busi-
the outputs are the needs and options analysis, ness case must identify important assumptions
both of which are part of the completed business (also referred in Figure 1 of The Role of Business
case document. Each output is a specifically dif- Case Development...). Risk-taking attitude, sus-
ferentiated business case. The needs analysis is tainable competitive advantages, organisational
a business case. So is the options analysis. The innovativeness, politics, cash flows, and total
needs analysis and options analysis represent a cost of ownership are examples of relevant as-
section in the completed business case document. sumptions.
The completed business case inevitably contains Risk is foremost to these assumptions. The
sections that correspond to the three stages of the implementation of new enterprise information
business case development. systems involves risk. How does the organisation
react to risk? The answer can be found in the or-
Analysis, depth, and Quality ganisations past actions.Risk-seeking, innovative
organisations exploit innovative opportunities
The keys to a successful business case are in-depth which have potential for big returns. Risk-averse
analysis and quality. A Risk-adverse organisation organisations are very late adopters of technology,
with an analytical bent will usually spend consider- and are less innovative. To risk-averse organisa-
able time and effort in fully understanding all the tions, the business case must mention similar
aspects and implications of any significant invest- implementations undertaken by competitors or
ment (Carruth, 2001, p 10). Too much analysis other organisations (refer to Appendix A section
is not cost effective. The content of the business in Table 2).
case must limit and highlight few but crucial Managing risk is simply gaining more power
aspects that impact the organisation. Implement- over the uncertainty brought about by the inno-
ing new information systems affect the operation vation-decision process. Borge (2001) suggests
and well-being of the organisation concerned. certain techniques to managing risk:
The question is: Which of these “effects” give
the greatest impacts to the organisation? Thus, • Being aware of the risks by defining, at the
there is a need to limit and highlight the critical start, the possible undesirable outcomes;
factors. The focus is, therefore, on the quality of • Knowing that taking deliberate action can
the information and the depth of analysis. Cannon increase the odds of desirable consequence
(2006, pp. 2-4, 194-252) suggests 29 analytical and decrease the chance of unexpected or
tools, such as balanced scorecards, cost-benefit undesirable consequences;
analysis, critical success factor, life-time cost • Weighing risks vs. benefits
analysis, risk analysis, sensitivity analysis, and
SWOT analysis. The amount of analysis carried Defining all possible undesirable consequences
out is irrelevant. In a good business case, the is easily overlooked or taken for granted because,
analysis must reflect the right depth of a critical as Borge (2001) puts it, there is no universal defi-
factor that impacts the organisation as a conse- nition of a bad outcome. Like the “new idea” to
351
exploit, the “undesirable consequence” to avoid will be accountable for diffusing its usage in the
depends on the perspective of the people involved. organisation (Weill & Ross, 2004; refer to project
Therefore, the undesirable consequence must be ownership in the Introduction section of Table 2).
explicit in the business case. As mandated by the
Sarbanes-Oxley Act, the executive sponsor and nPv, dcF, and tco
the project team members should be able to see
the threat early, and understand them before they The net present value is an old measure that most
become clear and present dangers (Green, 2004). accountants and financial professionals rely heav-
Here is a last word on the risky business of ily on valuation to business decision making.
managing risk. The rigour of developing a business One related valuation method is the discounted
case is NOT minimising the risk. It is balancing cash flows (DCF) method. It estimates the current
risk with opportunity to create the best overall market dollar value of the new enterprise informa-
value (for details, refer to the concept of “value tion systems. Based on expected future cash flows
at risk” by Borge). and discount rates, the DCF value drives capital
budgeting and therefore, the innovation-decision
other Important Assumptions process. For organisations with positive cash
flows, the cash outflows for future periods can
The attitude to risk is not the only relevant as- be estimated with a certain degree of reliability.
sumption. There are other assumptions that need Another important concept is the total cost of
to be addressed and identified. For example, how ownership, or TCO for short. The traditional notion
do new enterprise information systems foster of cost is the money paid in exchange for tangible
seamless alignment in the organisation? How does and intangible projects or services, where costs
the proposed implementation fit into the organi- can be either direct or indirect. Before making the
sation’s strategy? Table 1 has suggested certain adoption decision, a capital budgeting analysis will
sustainable competitive advantages. Therefore, be required. It involves breaking down the amounts
how does the business case describe in detail the to be paid. For example, with the enterprise in-
innovation, its scope, its goals (the sustainable formation systems, the acquisition costs include
competitive advantages), and its objectives (the initial software, hardware, installations, training,
expected consequences)? configuration, supporting, and consultation costs.
Corporate culture, specifically the power Other cash outflows are used for maintenance,
structure mentioned in the conceptual framework further training requirements, modifications, sup-
of The Role of Business Case Development..., is ports, and upgrades (Cua & Theivananthampillai,
another basic assumption. A business case may 2006; Piedad, 2001). The initial acquisition cost
rely more heavily on the opinions of the execu- represents a capital expenditure. The subsequent
tive sponsor or any other leaders with powerful ongoing costs are operating expenditures. These
opinions. The mention of the “power” person or so-called operating “expenses” accumulate over
people gives credibility (or reputation, the second time, and equate to at least a third of the initial
R) and responsibility (the third R) to the business acquisition cost (Schweitzer, 2003). They must be
case. If the business case is to diffuse successfully, included in the total cost of ownership, otherwise,
the premise of the corporate culture dictates a they will become hidden costs.
disclosure of ownership of the implementation. Combining the net present value, discounted
Furthermore, IT governance essentially involves cash flows, and total cost of ownership, give to
stating who is the executive sponsor who is financial analysis the much needed context as to
responsible to making the IT decision and who valuation, strategy, finance, and corporate gov-
352
ernance (Morin & Jarrell, 2001). There is one the various stakeholders. Some stakeholders are
caveat. TCO is like air. Its absence is fatal, but it the conditions to success. Others affect the creation
must remain discreet and in the background when of value. The purpose of stakeholders analysis is
present. Although determining the lowest TCO is to maximise shareholder value in the longer-term
good, deciding on the option with the lowest TCO (Morin & Jarrell, 2001; Rüegg-Stürm, 2005).
puts emphasis on minimising the cost through
TCO. It is, therefore, not wise. Instead, the ques-
tion to ask is not which option has the lowest structurInG And WrItInG
TCO? Rather, which option enhances the value tHe busIness cAse
to the organisation (Cua & Theivananthampillai,
2006; Morin & Jarrell, 2001)? Using a business report format, the business case
(Table 2) identifies the gaps between the expected
Planning for the Past, future state and the present state (Section 1). The
Present, and Future conclusions on the business case, together with
the recommendations, follow in Section 2. Subse-
In DOI, the innovation-decision process develops quent sections are the methods (Section 3), facts
along a normal course. Imagine the whole project and assumptions (Section 4), financial analysis
as a life cycle consisting of the two major phases (Section 5), nonfinancial analysis (Section 6), and
(the initiation and implementation phases) with the implementation plan (Section 7).
several stages in each phase. The completed As business cases are complex, the executive
business case (the third type of the business case) summary becomes the most important part of the
occurs in the decision stage of the first phase, and business case. Pugh and Bacon (2004) emphasise
effectively links the initiation phase to the imple- the “executive treatment,” and suggest several
mentation phase. Thus, a completed business case approaches of diffusing the innovation via the
must contain the story of the past (what was done executive summary. The upper managers are busy
in the awareness stage and matching stage), the people. They have time to read a summary but
present (the available options and critical factors not an entire proposal. They also have a crucial
to make the decision), and a preview of the project role in the final decision and therefore, they will
management (Nokes, Greenwood, Major, & Good- form their “attitude” towards the business case
man, 2004) with regards to the implementation of in three quick steps: the executive summary
the new enterprise information systems. (first step), the introduction (the second step),
and the conclusions and recommendations (the
nonfinancial Factors third step). By the third step, they usually have
an idea whether or not the executive sponsor has
Other than the sustainable competitive advan- undertaken the necessary rigour in developing the
tages, there are other nonfinancial factors, such business case, and whether or not the business
as the macro environments and the stakeholders. case is worth investing.
These are two of the six categories under the St
Gallen Management Model (Rüegg-Stürm, 2005).
The macro environments are particularly useful conclusIon
in weighing the external circumstances that can
impact on organisation’s growth and sustainability. The key points in the business case are the 3Rs
Organisations often define their success by the (relevance, responsiblity, and reputation) of the
degree to which they are able to meet the needs of strategic value of the new enterprise information
353
systems. What value impacts the organisation the Green, S. (2004). Manager’s guide to the Sar-
most? With the relevance brought about by the banes-Oxley Act: Improving internal controls to
strategic value come the in-depth analysis and the prevent fraud. New York: John Wiley & Sons.
quality (not quantity) of information. The project
Hammer, M. (1996). Beyond reengineering: How
ownership in Section 1 and the stakeholder analy-
the process-centered organization is changing
sis in Table 2 concern responsibility, crucial to
our work and our lives. New York: HarperCollins
Sarbanes-Oxley Act and corporate/IT governance.
Publishers, Inc.
Project ownership sets accountability for the deci-
sion and diffusing the usage of the new enterprise Ministry of Health. (2005). Business case guide-
information systems. Stakeholders’ analysis sets lines for investment in information technology.
the social responsibility. The rigour of the busi- Wellington, New Zealand.
ness case development foster a mindset that the
Monczka, R. M., Carter, P. L., Petersen, K. J., &
organisation is able to fully consider risks, and
McDowell, C. P. (2006). Project 10X: The value
that the organisation is able to mitigate the risk
proposition and strategic impact to sourcing
concerned with the initiation and implementation
and supply effectiveness. In J. H. Cavinato, A.
of the new enterprise information systems. The
E. Flynn, & R. G. Kauffman (Eds.), The supply
reputation of the organisation is at stake in this
management handbook (7th ed.) (pp. 209-232).
instance.
New York: McGraw-Hill.
Lastly, the proposed structure in Table 2 is a
blueprint, but should not be used as a boilerplate. Moore, G. A. (1991). Crossing the chasm: Market-
ing and selling disruptive products to mainstream
customers. New York: HarperCollins Publishers.
reFerences
Morin, R. A., & Jarrell, S. L. (2001). Driving
Borge, D. (2001). The book of risk. New York: shareholder value. New York: McGraw-Hill.
John Wiley & Sons, Inc. Nokes, S., Greenwood, A., Major, I., & Goodman,
Cannon, J. A. (2006). Making the business case: M. (2004). The definitive guide to project manage-
How to create, write, and implement a success- ment: Every executive’s fast-track to delivering on
ful business plan. London: Chartered Institute of time and on budget. New York: Financial Times
Personnel and Development. Prentice Hall.
Carruth, B. (2001). Develop a successful busi- Piedad, F. (2001). Total cost of ownership: Prin-
ness case for board approval. Wellington, New ciples and practical applications. Retrieved 17
Zealand: Institute of Chartered Accountants of April 2006, from http://www.phptr.com/articles/
New Zealand. printerfriendly.asp?p=24404
Cua, F. C., & Theivananthampillai, P. (2006). Value Pugh, D. G., & Bacon, T. R. (2004). Powerful
management of sourcing decisions: The cost of proposals: How to give your business the win-
ownership in performance management systems. ning edge. New York: American Management
Paper presented at the Pacific Asian Consortium Association.
for International Business Education & Research Rogers, E. M. (1962). Diffusion of innovations.
(PACIBER) 2006, Cebu, Philippines. New York: The Free Press of Glencoe.
354
Rogers, E. M. (2003). Diffusion of innovations of enterprise information systems consists of

(5th ed.). New York: Simon & Schuster, Inc. pre-production, production, and post-production
(also known as upgrade and maintenance). Refer
Rüegg-Stürm, J. (2005). The new St. Gallen man-
to innovation-decision process.
agement model: Basic categories of an integrated
Initiation Phase: Consists of awareness stage,
managment. New York: Palgrave Macmillan.
matching stage, and lastly, the decision stage. It is
Schweitzer, D. (Sept 2003). Track the true TCO: the first phase of the innovation-decision process.
Watch out for hidden costs over the long term. The second phase is the implementation phase.
Processor, 25(39). Refer to innovation-decision process.
Innovation: Represents a product, a service, or
Weill, P., & Ross, J. W. (2004). IT governance:
an idea that is perceived, or should be perceived
How top performers manage IT decision rights
by the audience or the market in which this in-
for superior results. Boston, MA: Harvard Busi-
novation is intended to be new and of value.
ness School Press.
Innovation-Decision Process: Starts with an
initiation phase through which the individuals
or decision-making units move from knowing
key terms And deFInItIons (understanding/identifying) the new idea (the
innovation), to forming of an attitude toward the
Business Case: Completed business case innovation, and subsequently, to deciding whether
document. Business case process. to adopt or reject the implementation and use of
Business Case Development: Walks through the new idea. The awareness stage is the agenda
the initiation phase of the innovation-decision setting stage. The attitude formation stage is the
process and talks about the project plans that matching stage. In addition, the decision stage
concern the implementation phase. to adopt or reject the innovation terminates the
Completed Business Case Document: A initiation phase. An adoption decision continues
formal written document that argues a course of the process toward the implementation phase,
action. It contains a point-by-point analysis to which consists of the pre-production, production,
making a decision for a set of alternative courses post-production, and confirmation stages.
of action to accomplish a specific goal. Risk: Connotes a possible negative impact to
Diffusion: Essentially communicating a new something of value. It symbolises the probability
idea (aka, the innovation) within a social system of a loss.
(such as an organisation) with the intention that Total Cost of Ownership: Also known as
the audience of that communication adopts or use TCO, is a rigorous and holistic methodology. It
the innovation. helps to estimate how much an investment will
Diffusion of Innovations: Theory concerns cost to operate over its lifetime. It takes into ac-
the how, why, and at what rate the new idea (com- count all direct and indirect costs. The indirect
monly referred to as innovation) diffuses. costs are generally insignificant individually.
Implementation Phase: Proceeding after However, they become very substantial when
the initiation phase, the implementation phase accumulated over time.
This work was previously published in Encyclopedia of Information Science and Technology, Second Edition, edited by Mehdi
Khosrow-Pour, pp. 3600-3607, copyright 2009 by Information Science Reference (an imprint of IGI Global).
355
356
Chapter 2.4
Extending Enterprise
Architecture with Mobility
Ming-Chien (Mindy) Wu
University of Western Sydney, Australia
Bhuvan Unhelkar
MethodScience.com & University of Western Sydney, Australia
AbstrAct carried out at international software development

organization. Further validation of the model is
The Enterprise Architecture (EA) brings to- envisaged through action research in multinational
gether various business processes, technologies, organizations.
standards, systems, and IT infrastructure of the
organization. This chapter considers extending the
EA with mobility so that it would facilitate easier IntroductIon
implementation of applications that overcomes the
boundaries of time and location. This extension of A carefully thought out and implemented Enter-
EA with mobility will result in a comprehensive prise Architecture (EA) provides the business with
Mobility Enterprise Architecture (M-EA) that will competitive advantage by opening up opportuni-
provide the business with advantages of real-time ties to streamline processes, reduce costs, increase
business processes, reduced costs, increased cli- customer satisfaction and enable thorough strate-
ent satisfaction, and better control. This chapter gic planning (Lan and Unhelkar, 2005). Businesses
outlines the M-EA framework, which is based can further advance these benefits by extending
on the literature review, initial modeling, and a the EA with strategic incorporation of Mobile
case study carried out by the lead authors. Later, Technologies (MT) - including wireless networks
the framework is validated by another case study and handheld devices – into their business plans.
This is so because of the phenomenal impact of
the “time and location” independence provided
DOI: 10.4018/978-1-60566-156-8.ch046
Extending Enterprise Architecture with Mobility
by mobile Technologies (Unhelkar, 2006; Barnes, EA (Linthicum, 2000). While EA has successfully
2002). Furthermore, mobility extends the ability managed to integrate these various technologies
of the organization to create dynamic intercon- used by the enterprises (such as Internet-based
nections, in real time, between various parts of its application, reusable components, security and
information networks. This ability to dynamically database components), enterprises are now seek-
interconnect various parts of its data and informa- ing to capitalize on the MT. This has resulted in a
tion through mobility results in “correlations” that need to further extend and integrate mobility into
provide new insights to the organization’s deci- the EA. MT is a significant emerging technology
sion makers and enhance their decision making. that has the potential to influence various organi-
Thus, a Mobile Enterprise Architecture (M-EA), zational applications (Unhelkar, 2005), in addi-
through location-independence, provides greater tional, impact EA as they are technologies without
opportunities for business information systems wires with the ability to communicate through a
to create these dynamic correlations, resulting multiplicity of hand-held devices. The advantage
in greater business advantage than information of mobility comes from its ability to overcome
exchanges over land-based Internet and other “time and location” boundaries that would enable
communication mechanisms. Thus, the objective enterprises to operate effectively real-time respond
of this chapter is to outline a comprehensive frame- to the ever-increasing changes on this competi-
work for incorporation of mobile technologies tive marketplace. The need, therefore, to have a
in an organization’s Enterprise Architecture that comprehensive EA that would enable delivery of
would provide it with competitive advantage. This services to the “location independent” market has
framework is based on the literature review and grown. Along with that need is the need to have a
initial modeling carried out by the lead author at the formal process to incorporate and extend the EA
Mobile Internet Research and Applications Group with mobility. This chapter outlines the process
(MIRAG) at the University of Western Sydney. of enabling such extension and incorporation of
Later, the framework is validated by case studies mobility in EA.
and action research in multinational organizations.
Initial findings suggests an “all encompassing”
approach to MEA in business that considers the enterPrIse ArcHItecture
business processes as well as the social aspects overvIeW
of mobile technologies, is likely to ensure greater
success, as against a pure technical approach. enterprise Architecture background
In order to increase liveliness in today’s dynamic

bAckGround to reseArcH business environment, enterprise needs to integrate
their business processes, systems, databases, hu-
Information Technology (IT) growth is substanti- man resources, infrastructures and technologies
ated by the large number of infrastructures and together. This integration of various aspects of an
products have been showing up in the market enterprise results in what is known as Enterprise
with ever-increasing frequency (Ramakrishnan Architecture (EA). Kamogawa and Okada (2004)
et. al., 2006). However, in order to increase the state that EA should integrate these various sys-
ability of the enterprise to serve its customers and tems such as Supply Chain Management (SCM)
deal with its business partners in today’s dynamic system, Customer Relationship Management
business environment, there is a need to integrate (CRM) system, and Enterprise Resource Planning
these products and services through a common (ERP) system. The overall EA comprises software
357
systems that may have been created using differ- prise Solution Architecture (ESA), and Enterprise
ent programming languages, databases, and may Technology Architecture (ETA). Enterprise Busi-
be operating on different technology platforms. ness Architecture (EBA) defines the enterprise
Ross et. al. (2006) and Cook (1996) all state EA business model, process cycles and timing; also
allows integration and coordination across whole shows what functions should be integrated into
enterprise, including internal and external enter- the system. Enterprise Information Architecture
prise. So, in Figure 1 this research extends the (EIA) focuses on the data along with the corre-
original idea to integrate not only IS, but also the sponding data models that should be integrated
people, data, processes, applications, platforms, into the system. Enterprise Solution Architecture
and middleware all should be integrated into EA. (ESA), also referred to as an application selection
This integration makes EA could provide best which is the collection of information systems sup-
solution of internal enterprise. Additionally, ap- porting EBA, which also helps the user to easily
plications of other businesses, which are external understand and use the interface and components.
enterprise users, include the customers, partners, Enterprise Technology Architecture (ETA) is a
suppliers, or all members in a supply chain should consistent set of Information Communication
be enabling to see a unified view of the EA as well. Technology standards and selection of technology
This internal and external enterprise architecture device, which uses technology infrastructures to
integration is the blueprint solution for different support EBA, EIA, and ESA.
generation information systems that already exist EBA focus on the integration of business in-
and / or future applications coordination. formation systems and how the business processes
run. EIS considers on data storage, ESA focus on
enterprise Architecture definition supporting people to easy understand and use,
and ETA collects the technology devices, appli-
Enterprise Architecture represents the enterprise’s cations, middleware, platforms, and standards to
key business system, information/ data, applica- support all the others. Based on it, EA (Figure 2)
tion, technology strategies and their impact on in this research has been defined as a structure of
business processes also the users. META Group integrating business processes with IT supported,
Inc. (2006) and VITA (2006) both demonstrate by providing a visualization enterprise solution
that EA consists of four key components, which of the relationships among the System, Process,
are: Enterprise Business Architecture (EBA), People, and Data in an organization.
Enterprise Information Architecture (EIA), Enter-
Figure 1. EA integrates several generations of EIS (Extended from Kamogawa and Okada, 2004)
358
Figure 2. Enterprise architecture - Core structure

Architecture Framework (MODAF) (UK). These
frameworks have influenced each other. The
TOGAF framework (The Open Group website,
2007) is quite popular amongst these frameworks.
TOGAF considers and provides a development
method on the construction of the EA. It includes
two parts of the architecture framework, TOGAF
Architecture Development Method (ADM) and
TOGAF Enterprise Continuum. The Continuum
describes as a series of “current” and “target”,
another set of terms for these are “as-is”, “to-be”,
and the ADM is a “migration plan” to show the
enterprise Architecture Framework clearly process from current different generations
of EIS to the Enterprise target dream architecture.
The critical success factor of structured and imple-
mented EA is having a best practice approach by
using an architecture framework. The framework mobIle tecHnoloGIes
provides a method of organizing architecture overvIeW
documents, dividing them into manageable parts
and defining cross linkages between them. One mobile technologies background
popular reference architecture on which an EA can
be based is the Zachman framework. Zachman “Mobile Technology” is usually used to describe
(1987) created the Zachman Framework which modern wireless connections such as those in cel-
has been described by many authors as basis lular networks and wireless broadband Internet.
for their work. (see EAF work by Finkelstein Wireless technologies encompass communica-
(2006), for example). This framework provides tion that is achieved without land-based or wired
a formal and highly structured way to define mechanisms. In modern usage, wireless is a
the concepts of an enterprise. The 6 x 6 table of method of communication that uses low-powered
the framework provides a classification model radio waves to transmit data between the mobile
based on what (data), how (process), where, who terminals (Shuaib and Boulmalf, 2006; Hammer
(people), when and why – which then crosses with and Champy, 2001). The terminals, such as mobile
the distinct model of planner, owner, designer phones, I-Pods, Personal Digital Assistants (PDA),
(system), builder (technology), implementer, Global Positioning System (GPS), watches, email-
and worker. The strong point on this framework only devices, handheld computers, and “wearable”
is that it provides everyone in the enterprise and technology, are carried by individuals and are far
enterprise architect group a clear picture of how more “personal” than mere desktop PCs. A wide
the enterprise is operating including its people, range of MT comprise “3G” mobile network,
processes and technologies. Mobile satellite networks, Bluetooth, Wireless
There are also several frameworks from Local Area Network (WLAN), Wi-Max and Radio
government agencies, such as The Open Group Frequency Identification (RFID).
Architecture Framework (TOGAF) (US), De-
partment of Defense Architecture Framework
(DOGAF) (US), Federal Enterprise Architecture
Framework (FEAF) (US), and Ministry of Defense
359
leading edge of mobile • Bluetooth

technologies ◦ Application: Bluetooth device,
Bluetooth earphone
• 3G Mobile Network ◦ Characteristics: Transfer data be-
◦ Application: mobile phone device tween a computer, server and
◦ Characteristics: Higher transmission one or more other mobile device
rate, popular used and high marketing “synchronously”
acceptance
Buttery and Sago (2004) describe the Blue-
The development of the third Generation -related tooth application as being built into more and
technologies has overcome the limitation of the more mobile telephones, allowing some very
previous generation of mobile technologies by interesting Mobility Commerce opportunities
allowing higher transmission rates and more com- to be created. As people currently carry mobile
plex e-commerce interactions (Ramakrishnan et. phones equipped with Bluetooth technology, this
al., 2006). According to “The Nielsen/NetRatings technology can be used for making payments and
Australian Internet and Technology Report 2004- related services through simple downloads on
2005” (Film Victoria Australia, 2006), MMS usage their mobile devices. Retailers might also be able
in Australia has grown by 21% in the past years, to provide samples of products to download via
mobile phones have already been popularly used. a Bluetooth link located close to the actual item,
People utilize mobile phones for communication, potentially resulting in better customer service and
working, banking, and shopping has been affecting an enriched shopping experience. Since Bluetooth
all features of daily life in the real world. technology is a radio transmission, it doesn’t need
line-of-sight with another Bluetooth-enabled
• Mobile Satellite device to communicate (Scheniderman, 2002).
◦ Application: GPS device and Internet Once Bluetooth technology is in place, one can
phone (Voice over IP- VoIP) envisage consumers walking around and giving
◦ Characteristics: Space technology- out messages wirelessly via Bluetooth in order
Direction finding and map reading to buy items from vending machines, or buying
low value tickets, or even making small value
Olla (2005) declared that integrating space ‘cashless’ purchases such as newspapers.
technology into mobile communications offers
two main advantages. The first advantage is • Wi-Max
in providing access to voice and data service ◦ Characteristics: Wireless online in
anywhere in the world – of which the current urban by using mobile devices or
popular application is Internet phone (Voice over computers.
IP - VoIP). The second advantage is exact posi-
tioning of useful location sensitive information WiMax Forum (2006) mentioned that the
used for direction-finding and map-reading based purpose of WiMax is to ensure that broadband
services - the current popular application is GPS. wireless radios manufactured for customer use
These applications are becoming commonplace; interoperate from retailer to retailer. The main
with Fitch (2004) pointing out that the technique advantages of the WiMax standard are to enable
for interfacing satellite links to global networks the implementation of advanced radio features in
is well developed. a standardized approach, and provide people in a
360
city with online access via their mobile devices APPlyInG mobIlIty to
or computers. enterPrIse ArcHItecture
• Wireless Local Area Network (WLAN) Information technology and

◦ Characteristics: Wireless link PC or enterprise Architecture
mobile device network connection in
particular surrounding area EA represents a technology-business philosophy
that provides the basis for cooperation between
Currently, laptop, computers and some PDA various systems of the organization that could
devices can be attached to a Wireless Local Area be inside or outside the organizational boundary.
Network (WLAN) using a Compact Flash (CF) or EA also facilitates ability to share data and infor-
a Personal Computer Memory Card International mation with business partners by enabling their
Association (PCMCIA) card. In future, PDAs and applications to ‘talk’ with each other. Cummins
mobile phones might support multiple network (2002) identifies that general characteristics of
technologies. WLAN is expected to continue the EA integration are: distributed computing,
to be an important form of connection in many component-based applications, event-driven
business areas. The market is expected to grow processes, loose coupling of business functions,
as the benefits of WLAN are recognized (Burness decision support information, workflow man-
et. al., 2004). agement, Internet access, and personalization of
interfaces. The infrastructure spans across various
• Radio frequency identification (RFID) technical among EA architectures, which includes
◦ Application: RFID tag, and reader database, applications, devices, middleware, net-
◦ Characteristics: Product tracking and work, platform, security, enterprise service bus,
controlling by system (automatically hosting, Local Area Network, Internet connection,
update from the RFID tag location operation system, servers, systems management
through RFID reader to the system) and so on (Pulkkinen, 2006). There are numer-
ous technologies that can be used for enterprise
Radio frequency identification (RFID) is an application integration, such as bus/hub, applica-
emerging technology that has been increasingly tion connectivity, data format and transformation,
used in logistics and SCM in recent years. RFID integration modules, support for transactions,
technology can help SCM system to identify, sort, enterprise portal, web service, and also Service-
and control the product and information flow all Oriented Architecture (SOA) (Finkelstein, 2006).
through a supply chain. Today RFID is a standard
technology that uses radio waves to automatically Gap between enterprise Architecture
identify people or objects. There are several meth- and mobility-enterprise Architecture
ods of identification, the most common of which
use RFID tags and readers (Kou et. al., 2006). Umar (2005) states that the Next Generation En-
terprises (NGEs) will rely on automation, mobility,
real-time business activity monitoring, agility,
and self-service over widely distributed opera-
tions to conduct business. Many organizations
would like to build their entire systems by using
the today’s emerging technologies of which MT
is a crucial part. Undoubtedly, MT can integrate
361
in EA to provide the enterprise to have M-service simple figure provides a visualization of the
and M-store. The enterprise can provide 24 hours, M-EA integrating System, Process, People, and
7 days, globalization service and any product Data in an organization in a manner that enables
enquire or technical support to the customers. easy incorporation of Mobile Technologies in its
Inter-organizational integrates all process between business processes. As MT integrates into EA,
extended enterprises, such as supply chain and business processes and IS will have to be reen-
customer relations systems. MT help to upgrade gineered. Following on from the changes to the
the traditional supply chain to Mobility Supply systems and processes, people also need to know
Chain Management (M-SCM), traditional CRM how to use the system to operate the process. As
to Mobility Customer Relationship Management well as the data need to be changed to be adopted
(M-CRM) (Lee, 2006) and trading procurement the new system functions. Therefore, the new M-
to Mobility procurement (M-procurement). This EA model will have to handle these four aspects
research looks at these extensions and integrations of an EA as they have a major impact on the way
in greater brief detail considering capabilities of an organization functions.
MT; moreover, these kinds of extensions and
integrations will also bring EA to Mobility - En- Applying mobile technology
terprise Architecture (M-EA). to enterprise Architecture
The time and location independence of
mobility open up tremendous opportunities for Mobile technology is very user centric and per-
organizations to offer integrated services to their sonalized. For example, the mobile phone is clas-
clients and partners. M-EA brings about not only sified as a necessary gadget for most individuals
internal integration but through extension, also in the 21st century. The uses of mobile devices are
offers much more efficiency to its external sup- very popular specifically with the application of
pliers, customers and other trading partners over 3G Mobile network technology. Mobile network
the Internet. Thus, M-EA will connect existing technology provides better service and real-time
and new systems to enable collaborative opera- response communication between business and
tion within the entire organization in real-time – consumer of EA. Some mobile devices have
providing access among systems on production started to provide with Bluetooth device from last
planning and control, inbound and outbound two years. Bluetooth technology is synchroniza-
logistics, material flows, monitoring functions, tion between a personal computer (PC) server
and performance measurements (Rolstadas and and one or more other mobile terminals (Shuaib
Andersen, 2000). However, EA with mobility and Boulmalf, 2006); Synchronization has been
has the challenges of security, privacy, computing particularly successful in cooperative applica-
power and usability. Our project aims to identify tions, providing access to EA. WiMax can help
these challenges, understand and document them, EA integration much more effectively and update
and work out strategies to handle them, resulting unanimity between all retailers of enterprise. The
in successful EA with mobility. An integrated EA general intra-organizational M-EA application
with mobility will provide immense benefits to integration can use WLAN technology to provide
organizations in extending their business IS to all the employees to access the enterprise system
beyond the organization’s boundaries. anytime, anywhere. VoIP technology helps EA
The literature review conducted as part of to extend the globalization business and also
this study provides a better understanding of the overcome the purchase timing problems though
problem. This problem is the gap between IT different countries. In addition, Hurster et. al.
and IT supported and integrated with MT. This (2006) pointed out that GPS devices and RFID
362
Figure 3. Gap from EA to M-EA
tags and readers have already used on SCM base, The technologies of Web Services build on
systems to improve delivery service and track- top of eXtensible Markup Language (XML), Web
ing production location. RFID Technology helps Services Description Language (WSDL) and
highly location-based tracking, reduces the cost Universal Description, Discovery and Integration
and human-mistake risks, also improves the effec- (UDDI), provide an excellent basis for integrating
tively and efficiency of EA (Schilhavy and Salam, the applications of the enterprise – particularly
2006). Consideration of these six MT is important when they are on separate platforms. There is a
in the exercise to integrate EA – particularly as need, however, to provide wireless glue to EA as
the technologies self as well as their applications outlined here. This would result in strategic in-
are maturing rapidly (Ghanbary, 2006). corporation of apply MT into EA starts on an
Figure 4 shows the integration of mobile enterprise repository which is a comprehensive
technology into enterprise technology architecture system containing all applications and the enter-
by describing it as “Mobile Enterprise Technol- prise model. People can reach the system using
ogy Architecture”. The lower part of the Figure the Internet though the native API/ XML, web
4 demonstrates how the application service and forms, and web service. They can use their mobile,
information/data storage go through the enterprise some with Bluetooth headphone, GPS, PDA,
bus or middleware cooperation. This transfer is portable computer (laptop) though service pro-
a basis to help business processes orchestration vider (WAP), or wireless Hub (HTTP) to connect
reengineering. Moreover, the new processes are with enterprise repositories to access the enterprise
the basis of integration different generation of information systems.
enterprise information systems. The most popular The implementation of M-EA should use
four enterprise information systems are SCM “Current” and “Target”, “As-is” and “To-be”
system, CRM system, ERP system, and financial framework to identify the clearly process migra-
(FI) system cooperation within enterprise archi- tion plan from current different generation EIS to
tecture. RFID technology could specifically sup- the Enterprise target dream architecture - M-EA.
port the SCM system to help tracking the location The manner in which this IT and mobile technol-
of the materials and products. This underside ogy incorporation takes place successfully is
section includes the M-EA business process, studied here through the framework outlined in
system, and data structure. People can hold vari- the migration plan table. Table 1 is a migration
ous mobile devices through mobility server and table plan which lists information system life-
access web service to connect to Internet portal, cycles including description, analysis, design, and
including web and mobile portal to run those implementation plan (Irani et. al., 2003) of current
business process, system and access through data state of enterprise, and desired target state M-EA
363
Figure 4. Mobile enterprise technology architecture
Table 1. M-EA migration plan table
Current state “as-is” Desired Target state “To-be”

Description Key factors Reserve assessment indicator
Analysis Assessment indicator result Target M-EA opportunities
Design Construct current EA Construct Target M-EA
Implementation plan Target M-EA opportunities result Target M-EA implementation plan
of enterprise. Firstly, go through people, process, analysis documents of which IT and MT infra-
system, and data migration after applying mobile structures could be extended with current EA as
technology integration to have four M-EA migra- target M-EA opportunities. After the meeting, the
tion plan tables. Then combine and cooperation research team will construct the current EA dia-
these four tables to have the M-EA migration plan grams, also the target M-EA diagram to show
to look down the blueprint of whole structure. how the MT integration to new M-EA can be
Firstly, the research team looks through the matched the enterprise expected extension. More-
enterprise current enterprise information systems over, these diagrams will have be modified through
to understand the key factors of each aspect, and the enterprise meeting to decide which MT ap-
fill into the description section. Then making the plication opportunities they would like to integrate
platform and interoperability diagrams from exist- into their M-EA and how the implementation
ing enterprise information systems to show the processes will be considered. All of the diagrams
reserve assessment indicator for target state. Fol- and documents will be analysis the critical success
low by, the research team sets up a meeting with point to enter into the migration plan table of each
enterprise team to show diagrams and analysis aspect, and then combine them into the compre-
those assessment indicator results and make the hensive M-EA migration plan table.
364
vAlIdAtIon oF m-eA model (cAse 2. Undertake comprehensive interviews to

study And ActIon reseArcH) further identify and document the current
business process and IS of chosen action
Firstly, this research starts on constructing initial research organization
M-EA structure to investigate and analysis the 3. Undertake detailed analysis of the current
changes at 5 aspects, which are data, process, business process and IS need to be dispa-
people, system, and technology from Case Studies. rate or integrated. (How are the company’s
This initial M-EA model provides the implemen- core business processes related to each
tation details of enterprise need to change their other? What information drives these core
architecture before they use mobile technology processes? How must this information be
application. Enterprise can have a perspective integrated?)
view of what Business Process Reengineering 4. Determine the current EA and whether it
(BPR) and Mobile-transformation preparation enables information flow.
need to be done first before they integrate M-EA. 5. Integrate mobility into the EA of the organi-
Secondly, this initial M-EA model would zation. (What technical capabilities should
be implemented and validated by doing Action be standardized companywide to support IT
Research study in chosen organization. The busi- and MT efficiencies and facilitate process
ness processes reengineering, practical issues, standardization and integration?)
benefits, challenges and limitation will be identi- 6. Consider the information sharing privacy
fied, documented and discussed with MIRAG and and security condition apply.
implementation team during and after implementa- 7. Implementation M-EA model to enterprise.
tion. Follow by modifying the model from those 8. Identify and document the key factors influ-
results to reduce the challenges and keep benefits encing the construction and collaborating
to the comprehensive M-EA model. Building on mobility into the EA of these organizations.
the work reported by Spewak (1992), this M-EA 9. Identify and document the key organizational
implementation planning project will based on the benefits and challenges of implementing
following steps: preliminary enterprise analysis M-EA model to enterprise.
and modeling -> analysis of current systems and 10. Modify the model of M-EA from discuss-
technology architecture -> architecture extension ing, present and obtain feedback from the
plan for data and applications -> extension plan research groups.
for Mobility in the overall EA -> implementation 11. Bring the update version model go back ac-
migration plan -> transition to implementation tion research organization to the first step.
-> testing and validation -> release. The major
steps of Action Research in this project are based However, security has to be given due consid-
on Oosthuizen’s (2000) description of a typical eration when the enterprise considered implement
action research circle: action -> results -> reflec- M-EA. The increased mobile computing power
tion -> plan.) of handheld devices introduces new security
Following are the specific steps envisage in threats. The security threats would be in the form
action research of this M-EA model validation to of loss of confidentiality, loss of integrity and loss
comprehensive model from these points onwards: of availability. Password protection is included
in most handheld devices as a countermeasure
1. From case study to construct a draft M-EA (Guizani & Raju, 2005). Furthermore, network
model and application security measures also need to be
considered when using mobile technologies. Hav-
365
ing risk identification and risk planning measures This research envisages continuing effort in
could help to overcome some risks (Ghosh, 2001). the future to enable the researcher to validate the
A possible solution is to build security measures M-EA model. This research currently undertakes
into mobile platforms and applications themselves. case studies and action research in various orga-
Base on above, security is the important part that nizations in practice that will enable validation of
has not been considered in this research. CLEW the initial M-EA model. The initial action research
(Closed Loop Environment for Wireless) is a new suggests enterprises can benefit with the help of
mobile-based technology designed by a company M-EA through its step-by-step framework for
called Alacrity Technology based in Canberra, extending and integrating the existing EA. Further-
Australia (http://www.alacritytech.com.au/). This more, the key factors influencing the construction
technology improves the security of using mobile and collaboration of mobility into the EA of these
technology, and this company- Alacrity is the first organizations; and also the key organizational
action research company of this research project, benefits and challenges of implementing M-EA
aims to improve M-EA security. CLEW technol- model to enterprise are also been focused on in
ogy is introduced as an Intelligent Mobile Internet nearly future.
Interaction (Tatham, 2006). CLEW is more secure
than SMS message, and also records the replies
from the receiver. Messages sent out by CLEW to reFerences
recipients are authenticated by the recipients by
signing off with a password. Therefore, it is much Barnes, S. J. (2002). The mobile commerce value
more secure than SMS, and also if the recipient chain: analysis and future development. Interna-
does not sign off, the backend systems of CLEW tional Journal of Information Management, 22(2),
can send out messages to other pre-agreed people 91–108. doi:10.1016/S0268-4012(01)00047-0
in a list agreed upon with the organization until Burness, L., Higgins, D., Sago, A., & Thorpe,
it schedules the required number of personnel. P. (2004). Wireless LANs – present and future.
Mobile and wireless communications: Key tech-
nologies and future application. British Telecom-
conclusIon And munications Plc.: The IEE.
Future dIrectIon
Buttery, S., & Sago, A. (2004). Future application
This chapter outlines the importance of M-EA of Bluetooth. Mobile and wireless communica-
model as a means of identifying integration chal- tions: Key technologies and future application.
lenges and providing that integration between British Telecommunications Plc: The IEE.
various applications and technologies within the
Cook, M. A. (1996). Building Enterprise Infor-
enterprise. Moreover, this chapter also provides
mation Architectures- Reengineering Information
an overview of EA, MT and applications of M-
Systems. Prentice-Hall, Inc.
EA. This research argued that MT needs to be
integrated with the overall EA and the business Cummins, F. A. (2002). Enterprise Integration:
processes of the enterprise. Such integrated would An Architecture for Enterprise Application and
result in M-EA, it has been identified that would Systems Integration. Canada: Willey Computing
enable the enterprise to conduct business across Publishing, John Wiley & Sons, Inc.
the location and time boundaries. Thus, an inte-
grated M-EA is a powerful tool to help manage
the enterprise’ operation.
366
Film Victoria Australia. (2006). iSH MEDIA. Irani, Z., Themistocleous, M., & Love, P. E. D.
Digital Sanbox Seminar Film Victoria, June (2003). The impact of enterprise application in-
2006. [online]. Available: http://film.vic.gov.au/ tegration on information system lifecycles. [Else-
resources/documents/S2_speaker_Kylie_Robert- vier Science.]. Information & Management, 41,
son_powerpoint.pdf 177–187. doi:10.1016/S0378-7206(03)00046-6
Finkelstein, C. (2006). Enterprise Architecture Kamogawa, T., & Okada, H. (2004). Issues of
for Integration – Rapid delivery methods and E-Business implementation from Enterprise
technologies. Artech House: Boston Lonton. Architecture viewpoint. Proceedings of the 2004
International Symposium on Applications and the
Finkelstein. C. (2006) Enterprise Architecture for
Internet Workshops (SAINTW’04), 2004 IEEE.
Integration: rapid delivery methods and technolo-
gies. U.S.A.: Artech House. Kou, D., Zhao, K., Tao, Y., & Kou, W. (2006).
RFID Technologies and Applications. In W. Kou
Fitch, M. (2004). The use of satellite for mul-
and Y. Yesha (Eds.), Enabling Technologies for
timedia communications. Mobile and wireless
wireless E-Business. Berlin Heidelberg: Springer-
communications: Key technologies and future
Verlag. (pp 89-108).
application. British Telecommunications Plc:
The IEE. Lan, Y., & Unhelkar, B. (2005). Global Enterprise
Transitions: managing the process. Hershey, PA:
Ghanbary, A. (2006). Evaluation of mobile
IGI Golbal.
technologies in the context of their applications,
limitations and transformation. In Unhelkar B. Lee, C. (2006). Mobile CRM: Reaching, acquiring,
(Ed.), Handbook of Research in Mobile Business: and retaining mobile consumers. In Unhelkar B.
Technical, Methodological and Social Perspec- (Ed.), Handbook of Research in Mobile Business:
tives.Hershey, PA, USA: IGI Global. Technical, Methodological and Social Perspec-
tives. Hershey, PA, USA: IGI Global.
Ghosh, A. K. (2001). Security and Privacy for
E-Business. New York: John Wiley & Sons Linthicum, D. S. (2000). Enterprise application
integration. Addison- Wesley information tech-
Guizani, M., & Raju, A. (2005). Wireless Networks
nology series.
and Communications Security. In Y. Xiao, J. Li
and Y. Pan (Eds.), Security and Routing in Wire- META Group, Inc. (2006) Adaptive Enterprise
less Networks,3, 320. New York: Nova Science Architecture. META Group, Inc., CT-USA,
Publishers, Inc. 2004. – Bittler. R. S. private presentation to UWS
Advanced enterprise Information Management
Hammer, M., & Champy, J. (2001). Reengineer-
Systems (AeIMS) Research Group on 13/11/2006
ing the Corporation: A manifesto for business
revolution. London: Nicholas Brealey. Olla, P. (2005). Incorporating commercial space
technology into mobile services: Developing
Hurster, W., Fuychtuller, H., & Fischer, T. (2006).
innovative business models. Hershey, PA, USA:
Mobile Batch Tracking- a breakthrough in supply
Idea Group Inc.
chain management. In Unhelkar B. (Ed.), Hand-
book of Research in Mobile Business: Technical, Oosthuizen, M. (2000). Action research. In K.
Methodological and Social Perspectives., Her- Williamson (Ed.), 2000: Research methods for
shey, PA, USA: IGI Global. students and professionals: information manage-
ment and systems. Centre for information systems:
Charles Sturt University. (pp. 141-158).
367
Pulkkinen, M. (2006). Systemic Management of Tatham, A. (2006). How CLEW could authenti-
Architectural Decisions in Enterprise Architecture cate credit cards to avoid credit card fraud. In
Planning. Four Dimensions and Three Abstraction U. AEIMS research Group (Ed.) (pp. Technical
Levels. Proceedings of the 39th Hawaii Interna- Presentation). Sydney: Alacrity.
tional Conference on System Sciences– 2006 IEEE.
Umar. (2005). IT Infrastructure to Enable Next
Ramakrisham, K. R., Bhattar, R. K., Dasgupta, K. Generation Enterprises. Information Systems
S., & Palsule, V. S. (2006). Review of Wireless Frontiers, 7(3). ISSN:1387-3326.
Technologies and Generations. In Unhelkar B.
Unhelkar, B. (2005, August). Transitioning to a
(Ed.), Handbook of Research in Mobile Business:
Mobile Enterprise: A Three-Dimensional Frame-
Technical, Methodological and Social Perspec-
work. Cutter IT journal, 18(8). Cutter Information
tives., Hershey, PA, USA: IGI Global.
LLC.
Rolstadas & Andersen. B. (2000) Enterprise
Unhelkar, B. (Ed.). (2006). Handbook of Research
Modeling- Improving global industrial competi-
in Mobile Business: Technical, Methodological
tiveness. Kluwer Academic publishers.
and Social Perspectives. Hershey, PA, USA: IGI
Ross, W. J., Weill, P., & Robertson, D. C. (2006). Global.
Enterprise Architecture as Strategy- creating a
U.S. Department of Defense Technical Archi-
foundation for business execution. Boston, MA:
tecture. TOGAF Architecture. The Open Group
Harvard Business School Press.
[Online]. Available: http://www.theopengroup.org
Scheniderman, R. (2002). The Mobile Technology
Virginia Information Technologies Agency.
Question and Answer Book. Amacom: American
(2006). Enterprise Definitions and Models. Com-
Management Association.
monwealth of Virginia [Online]. Available: http://
Schilhavy, R., & Salam, A. F. (2006). Emerging www.vita.virginia.gov/cots/ea/modelAndArchi-
mobile technology and supply chain integration: tecture/index.cfm
Using RFID to streamline the integrated supply
WiMax Forum. (2006). WiMax Forum - Frequency
chain. In Unhelkar B. (Ed.), Handbook of Research
Ask Question, [online]. Available: http://www.
wimaxforum.org/technology/faq, WiMax Forum,
WiMax Forum
Global.
Zachman, J. A. (1987). Zachman Framework. The
Shuaib, K., & Boulmalf, M. (2006). Co-Existence
Zachman institute for framework advancement
of WLAN and WPAN Communication Systems.
[Online]. Available: http://www.zifa.com/
In Unhelkar B. (Ed.), Handbook of Research
and Social Perspectives, Hershey, PA, USA: IGI
Global. key terms And deFInItIons
Spewak, S. H., & Hill, S. C. (1992). Enterprise Enterprise Architecture (EA): A structure of
Architecture Planning: Developing a Bluepoint for integrating business processes with IT supported,
data, applications, and technology. A Wiley-QED by providing a visualization enterprise solution
publication, John Wiley & Sons, Inc. of the relationships among the System, Process,
People, and Data in an organization.
368
Mobile Enterprise Architecture (M-EA): A lular networks and wireless broadband Internet,
structure of integrating business processes with also encompass communication that is achieved
Information Technology that are supported and without land-based or wired mechanisms
integrated with MT. It integrates business Systems, Service-Oriented Architecture (SOA): Us-
Processes, People, and Data in an organization ing of services to support the business and user
in a manner that enables easy incorporation of requirements, one kind of modern today enterprise
Mobile Technologies in its business processes. architecture.
Mobile Technologies (MT): Used to describe
modern wireless connections such as those in cel-
This work was previously published in Handbook of Research in Mobile Business, Second Edition: Technical, Methodologi-
cal and Social Perspectives, edited by Bhuvan Unhelkar, pp. 487-498, copyright 2009 by Information Science Reference (an
imprint of IGI Global).
369
370
Chapter 2.5
Enterprise Modelling in
Support of Organisation
Design and Change
Joseph Ajaefobi
Loughborough University, UK
Aysin Rahimifard
Richard Weston
AbstrAct ganisation (re)design and change approaches have

been proposed and tried with varying degrees of
Enterprises (business organisations) are increas- successes. This chapter contribution discusses
ingly operating under uncertain conditions arising the role of enterprise and simulation modelling
from: governments that introduce new regula- in support of organisation (re)design and change.
tions; a market place which is shaped by ongoing The capabilities and constraints of some widely
change in customer requirements; change in capital acknowledged public domain enterprise modelling
markets that orient overall market directions; an frameworks and methods are reviewed. A model-
advancing base of technology; and increasing ling approach which integrates the use of enterprise
competition which can arise from a growing num- modelling (EM), causal loop modelling (CLM),
ber of sources (Monfared, 2000). Consequently, and simulation modelling (SM) is described. The
organisations are expected to change rapidly in approach enables the generation of coherent and
response to emerging requirements. Classical semantically rich models of organisations. The in-
theories and more recently ‘method-based’ or- tegrated modelling approach has been applied and
tested in a number of manufacturing enterprises
(MEs) and one case study application is described.
DOI: 10.4018/978-1-60566-856-7.ch018
Enterprise Modelling in Support of Organisation Design and Change
comPlexIty oF orGAnIsAtIons With growing uncertainty in the world most

And tHe need For modellInG organizations need to operate and compete in a
volatile environment (Warnecke 1993, Verna-
From a systems engineering perspective an ‘or- dat 1996, Weston 1998 and Mills et al 2003).
ganisation’ is an entity which consists of func- Successful organisations can take many forms
tional parts or members that contribute to the (such as be large or small, centralized or dis-
achievement of that purpose (Blethyn & Parker, tributed, manual or automated and transactional
1990). Farnham & Horton (Mullins, 2005) state or transformationally led). Consequently the
that organisations are social constructs created by process of designing and changing organisations
people to achieve specific purposes by means of is complex and through their lifetime, various
planned and co-ordinated activities. Organisations ‘organisation design and change’ (OD&C)
deploy people that work in association with other projects are needed to maintain alignment
resource systems to realise well ordered sets of between the composition of the organisation
activities that lead to achievement of specified and emerging requirements. This paper is
objectives. Martin (2005) identified four common concerned with providing improved means of
aspects of any organisation: engineering manufacturing organisations, or
so called Manufacturing Enterprises (MEs),
• a system of coordinated activities that typically realise multiple product types
• a group of people & other resources that in uncertain quantities for various customers.
realise those activities Those MEs can be considered to be complex
• defined goal(s) for the following reasons:
• leadership
1. they may deploy large numbers and varieties
Mills et al (2003) describe an organisation as of system components; including people,
a system of resources that collaboratively execute electromechanical machines and IT systems
coordinated routines so as to realise product & that execute their many processes, possibly
services. Siemienuch et al, (1998) suggest that concurrently, to generate values for their
an organisation is configuration of knowledge, customers;
embodied in people and machines, which uti- 2. causal and temporal dependencies exist be-
lises data to create information (e.g. product tween system components such that changes
data models) and their physical manifestations in one component can impact significantly
(products for sale). Conventional means of devel- on the entire ME;
oping an organisation centres on bringing people 3. complexity arises because of need to deploy
together and providing them with a structure various philosophies, methods, frameworks
(action plans) and technology for doing work and technologies to integrate the operation
(Davis, 1982). Here a common requirement is of ME components;
to realised organised association between people 4. significant uncertainty arises from the en-
(competences) and jobs (related set of activities) vironment in which MEs operate, due to
(Vernadat, 1996); thereby developing needed increased global competition, rapid techno-
behaviour and constraining unwanted behaviour logical changes and product customization
(Weston 1998) . This leads to a unit of society or (that often necessitate changes in product
‘organisations’ that function to realise products mix, product properties & production
and services (Drucker, 1990, Warnecke, 1993, volumes).
Handy, 1993).
371
Despite their inherent complexity, typically ness and regularity exist in nature; thus there is
MEs are expected to generate unique behavioural consistency in terms of causes and effects (Alison
responses such that (a) in the short term, profitable et al 2001). Classic organisation theorists: Taylor,
responses are made to changing market require- Fayol, Urwickand Gulick, Mooney and Reilly
ments and (b) in the long term, the enterprise and Weber generally focused on the question of
recomposes and renews itself so that it continues organisational structure and their prescriptions
to thrive. Consequently managers and engineers tended to reflect mechanistic structural prescrip-
responsible for any given ME should deploy tions for centralisation and specialisation of
suitable means of defining ME goals and needed work (Scott & Mitchell, 1976). With increased
leadership and integration mechanisms that will: uncertainty worldwide this kind of approach, and
improve the choice and operation of processes its assumptions about roles in organisations, has
and resource systems (human & technical) lead- increasingly becoming inadequate. However, the
ing to quality, timely and cost effective delivery merits of the mechanistic approach to organisa-
of products and services in response to emerging tion design and change lies in is predictability.
customer requirements, within constraints of By assuming that aspects of activities, jobs, roles
stakeholder desires. It follows that implementing & tasks are predictable then their outcomes and
change in any ME requires understandings about performances can in part be predicted and planned.
and manipulation of complex causal and temporal There may for example be specialised systems of
dependencies between ‘strategic intentions’; ‘pro- roles that are relatively stable over long period
cess designs’; ‘product (and service) portfolios’; of time that have clear reporting relationships.
‘resource configurations’; ‘departmental boundar- Nowadays, however, increasing rates of change
ies and people reporting structures’; ‘manufactur- in organisations is affecting the stability of role
ing policies’; ‘work-organising methods’; and requirements and thereby the suitability of role
‘human and technical resource systems’ (Zhen incumbents. Consequent upon increasing rates
& Weston 2006). of change, traditional assumptions about stable
OD&C (organisation design and change) jobs and thereby roles that are punctuated with
necessitates coherent thinking and action-taking periodic changes may become obsolete in some
amongst groups of people with differing agendas, types of ME. Roles in organisations may endure
desires, concerns, roles, affiliations, responsibili- for relatively long periods but role requirements
ties, competencies and experiences related to the may not (Ashforth, 2001). Though the classic ap-
organisation and its environment (Ajaefobi et al proach (to organisation design and engineering)
2008). Generally OD&C implementation will satisfied the requirements of a steady state business
significantly impact on all ME personnel, such environment (Weston et al, 2001), current busi-
as by changing their status, roles, role relation- ness environments may not be deterministic but
ships, work patterns, use of (personal and group) rather stochastic (Ramenyi, 1998). Furthermore,
productivity tools, fitness to fulfil their roles it has also been observed that classic approaches
and designated workloads, working conditions, to organisational design have often led to rigid
opportunities for personal advancement and organisation structures and systems that can not be
remuneration and so forth. Classic approaches changed with sufficient regularity. The same struc-
to OD&C have previously been developed and tures deliberately imposed to organise targeted
applied by industry and academia (Mintzberg, behaviours and constrain unwanted behaviours
1989). However, most of those classic approaches can placed severe restrictions on responses to
are rooted in developing mechanistic views of emerging requirements. Consequently the abil-
reality with underlying assumptions that: orderli- ity to respond to any new set of requirements or
372
to assimilate the use of new technologies within enterPrIse modellInG:

products and products realising processes may meAnInG And scoPe
be compromised (Goldman et al, 1995, Barber &
Weston 1998). It follows that classic approaches To effectively define complex organisations that
and their solution provision are satisfying a dimin- give rise to stochastic behaviours, it is essential
ishing fraction of present organisational change to think in terms of models. Enterprise Modelling
project requirements. Alternatively stated, MEs (EM) uses an appropriate language or modelling
adopting improved OD&C practice can have constructs to describe a complex object such as
significantly improved competitive advantage an enterprise, in a way that the model developed
over MEs who retain conventional practices. In can be read and understood by anyone who knows
recent decades, ‘method – based’ approaches to the modelling language. Thus EM is a means of
OD&C have been suggested and tried in indus- externalising and formalising knowledge about
try with varying degrees of successes. Popular an enterprise in an unambiguous form; making
‘methods-based’ approaches to OD&C include: it possible to be exploited by other enterprise
Business Process Re-Engineering, Continuous agents and those outside the enterprise boundary.
Process Improvement, Kaisen, Total Quality Most enterprise knowledge resides in the minds
Management, Just-In-Time Manufacture, Push- of people who comprise the enterprise. However,
Pull Manufacturing, and Lean and Agile Manu- by applying suitable decomposition principles for-
facturing (Hammer & Champy, 2001, Davenport, mally by using well defined modelling constructs,
1993, Evans & Lindsay, 1993, Womack, et al, holistic models of a complete enterprise can be
1990). Their underlying structures and concepts created to facilitate shared understandings about
‘systemise’ and ‘focus’ OD&C thinking and ac- different aspects of an organisation thereby facili-
tion taking and are reported to yield beneficial tating organisation design and change. Any model
outcomes relative to preceding classic approaches. can be viewed as being a useful representation of
Generally though, the industrial application of an object. To improve its utility when modelling
method-based OD&C leads to long lead-times, a complex object the model will necessarily be
high costs, poorly justified engineering projects an abstraction of reality (universe of discourse)
that do not prepare the organisation for future expressed in terms of some formalism or modelling
change. These outcomes are to be expected be- constructs for the purpose of the user (Vernadat,
cause (1) invariably manufacturing organisations 1996). State-of-the-art enterprise models (EMs)
constitute very complex and dynamic systems are a graphical & possibly computational repre-
that naturally require complex design and change sentation of the structure, processes, information,
processes and (2) current method-based OD&C resources, people, goals that comprise a subject
are not analytically well founded. The present enterprise, possibly with reference to business,
authors argue that unless suitable decomposition government, social, technical and environmental
and quantitative and qualitative modelling prin- constraints. EMs can be both descriptive and defi-
ciples are used to underpin method-based OD&C nitional - spanning what the enterprise is at present
they will remain deficient. Enterprise modelling (so called ‘as-is’ models) to what the enterprise
(EM) and enterprise engineering (EE) can now could be at some point in time into the future.
provide a form for model driven OD&C and many The role of most EMs is to achieve model-driven
virtual engineering (VE) tools and methods are enterprise design, analysis, operation and change
emerging that will build upon such a foundation (Fox, & Gruninger, 1998).
to revolutionise the way in which organisations Generally, enterprises have structures, func-
are conceived and engineered. tions and behaviours that can be depicted in differ-
373
ent models (Christensen et al, 1995). Modelling an d. resource view

enterprise from a structural perspective involves: e. business rule view (economic)
(a) developing models of mission, vision, values, 3. three modelling levels
culture and (b) modelling the enterprise resources a. requirement definition – what the
and their relationships. Modelling enterprise customer wants to model to achieve
functions encompasses developing models of b. desig n specification – to design for-
(a) what the enterprise does to create values for mally one or more solutions of satisfy-
its customers (an ‘as is’ EM) and (b) what the ing set of requirements
enterprise intends doing in future (possible ‘to c. implementation description – detailed
be’ models). implementation
Berio & Vernadat (1999) suggest that Enterprise
models should describe among other things: Therefore, any given enterprise model is not a
monolithic model but is an assemblage of models
1. three fundamental types of flows within or that represent different aspects of the enterprise.
across the enterprise, namely: Figure 1 (Verndat, 2004) is an illustrative descrip-
a. material flows (physical objects such tion of different enterprise aspects that can be
as products, tools & raw materials) captured by EMs.
b. information flows (documents, data, The development of useful models of MEs
computer files), and requires a modelling architecture (or a framework
c. decision/control flows (sequence of within which modelised elements are created
operations) and integrated using well defined modelling
2. five modelling views: constructs) and a modelling methodology (i.e.
a. function view (what the enterprise a well defined set of modelling stages or steps
does) which the modelling takes to create, integrate
b. organisation view and deploy the use of modelled elements). En-
c. information view terprise modelling architectures are designed to
Figure 1. Main views represented by enterprise model
374
treat enterprises as complex systems that can be process instances will often not be readily
designed or improved in a structured and orderly predicted.
manner. Thereby better results can be achieved 2. organisations deploy multiple and often
relative to the use of an organisation (re)design interoperating resource systems (human &
approach (Bernus, 2003). As previously discussed, technical) that need to be modelled in terms
real world MEs are complex systems that oper- of needed resource system attributes, such
ate within complex environments. Consequently as competences, capabilities & capacities
EM models will invariable be simplified and they can bring to assigned roles, including
abstracted representations of complex realities. In potential and actual abilities to reach states
theory EM allows the representation of multiple and change states so as to quantify and qualify
aspects of the enterprise in so called model views. relative performances of alternative choices
Those views are aspects of the enterprise relevant of resource system.
to multiple model users who will have differing 3. organisational processes can be modelled at
responsibilities for the ME, thereby allowing the different layers of granularity to represent
user to deploy and possibly develop models of value streams, roles, activities and operations
interest without being disturbed by the overall that can be explicitly be described in terms of
enterprise complexity. required competences to enable marshalling
Public domain approaches to EM almost and matching of available competences that
exclusively adopt a process thinking approach can be brought to bear on multiple process
to representing, defining, analysing and com- instants by people and their supporting tech-
municating organisation entities and systems. nical systems in timely and effective ways
Such a process oriented approach can facilitate 4. a suitable integrating infrastructure is needed
requirements modelling and provides a unifying to link aspects of processes and systems
glue to bring together different aspects (namely: across organisational boundaries to ensure
function, information, resources and organisa- interoperability between processes and re-
tion aspects) of the various ME views required. source systems.
To deploy EM, so as to facilitate organisation
design and change the present authors made the Also assumed is that the realisation of those
following assumptions: modelling requirements necessitates the use of
proven modelling architectures and modelling
1. organisations are set of multiple, dependent, methods with capabilities to abstract and repre-
concurrent operating business processes: sent any organisation aspects of different views
where processes can be conceptualise at that suit modeller, user and other stakeholders’
various levels of abstraction (suited to requirements. Public domain modelling frame-
model users) in terms of ordered set of works were observed to exist that in part provide
value adding activities, inputs, outputs, a viable modelling framework such as: CIMOSA,
events (and their associated materials, IDEF0, PERA, ARIS and GRAI/GIM. Most of
information and control flows). Related to these frameworks are supported by modelling
this assumption is the notion that typically methods that are designed to systemise enterprise
many different instances of these processes model capture and reuse. They usefully provide
need to be organised and resourced so as means of handling organisational complexity, by
to realise the products and services created offering modelling constructs to decompose (gen-
by the enterprise in its life time; and that eral and specific) process networks into process
the qualities, mixes and timings of these segments. Also they were observed to provide
375
means of documenting and visualising associated IdeF suit of methods

flows of activities, materials, information, controls
and so forth. Subsequent sections of this Chapter IDEF modelling suite was developed in the early
will briefly review the capabilities of some of 1980’s as part of the US Air Force ICAM pro-
these widely known modelling frameworks and gramme. It is a derivative and further extension
methods and suggest reasons why MSI mainly of SADT. The IDEF modelling suite consist of
deploy CIMOSA in its modelling ventures. (a) IDEF0 for function modelling, (b) IDEF1X for
information modelling and (c) IDEF2 for dynamic
modelling. The IDEF suite was designed to model
modellInG metHodoloGIes decisions, actions, and activities of a system in a
revIeWed structured graphical manner (Bravoco & Yadav,
1985, Jang, 2003). Figure 2 illustrates the IDEF0
structured analysis and modelling approach. Like its SADT predeces-
design technique (sAdt) sor, IDEF0 adopts a hierarchical decomposition
approach to complexity management; thereby
SADT was originally designed as a ‘blue print’ decomposing complex problems to lower levels
for software engineering and not as complex of granularity. It uses the ICOM constructs of
systems design methodology (Ross & Schoman SADT to describe the system being modelled.
1977). However, the scope of SADT has been Despite its capabilities, IDEF0 has some widely
widened into a full scale methodology for require- known limitations especially in terms of its limited
ments definition, problem analysis and function time-based capabilities for example, its technique
specification and has since proven useful in many generates ambiguous activity specifications that
application domains. The strength of SADT lies are essentially static in nature. (Wu, 1992, Jang,
in its system decomposition principles that uses 2003). IDEF0 models are static in the sense that
four modelling constructs, namely: input, control, they are paper based and are not directly computer
output & mechanism (ICOM) to model a system. executable. Furthermore, IDEF2 which models
In this way, it provides stakeholders (using the system behaviours is ill defined requiring the
model) with a clear knowledge of the workings support of other tools such as Petri nets to analyse
of the system being modelled including who and system behaviours (Vernadat, 1996).
what performs what and necessary operating con-
straints. But a key limitation of SADT (Vernadat, the Purdue enterprise reference
1996) is that it cannot sufficiently describe well Architecture (PerA)
discrete event dynamic systems (like manufactur-
ing systems) because: PERA was designed to support industry in the
development and implementation of integrated
• SADT does not handle flows but focuses manufacturing system (Williams, 1994, Verna-
on dependences dat, 1996). Developed at Purdue University in
• SADT provides no dynamic behaviour de- the USA in collaboration with a consortium of
scription e.g. time dependent control flows industry collaborators, PERA methodology starts
• SADT is not precise in its semantics. with requirements definition, which is structured
in two respective branches, namely: requirement
definitions for information and manufacturing
systems. Next to this are three compartments
376
Figure 2. IDEF0 decomposition principles illustrated
that respectively encode: manufacturing system odology. The conceptual model presumes that as
architecture and the information system architec- complex system, an organisation is comprised of
ture with human and organisational architecture three parts namely;
sandwiched between the two. The human and
organisational section provides an architecture • physical system that comprises technical
for describing organisational structures, human and human systems, structured work cen-
positions, roles and training requirements dur- tres and materials that flow through those
ing the life cycle of MEs. PERA prides itself in centres
comprehensiveness, claiming to cover enterprise • operating system that is dedicated to real
life cycle from inception and mission specification time control of physical system
to operational and then plant disposal. • decision system – locus of decisions for the
whole organisation via a hierarchical struc-
Group de recherché en ture organised into decision levels made of
Automatisation Integrere (GrAI) decision centres
• the information system that links the
GRAI methodology (Domeingts, et al, (1995), other systems with one another and the
Wainwright & Ridgeway, (1994), Vernadat, environment
(1996)) was developed at the University of Bor-
deaux in early 1980’s. GRAI is predominantly a GRAI uses two graphical tools: (a) GRAI grid
graphical technique using decomposition principle used to analyse focused parts of the enterprise and
comprising a combination of ‘top-down’ and (b) GRAI nets used to analyse decision centres
‘bottom-up’ approaches to define decision cen- in terms of their activities, resources and mecha-
tres and information required to make a decision nisms. GRAI is supported by GIM methodology
within those centres. GRAI approach is based on which extends the capability of GRAI modelling
three elements: (a) conceptual model, (b) tools framework to cover the entire CIM systems.
& representation rules and (c) application meth-
377
cIm open system gies has impacted upon industry with respect to
Architecture (cImosA) specifying requirements (process models) and
structuring and supporting the design of resource
CIMOSA was developed by the ESPRIT consor- systems (human & technical); significant applica-
tium AMICE and validated with ESPRIT projects tion constraints still remain. The present authors
CIMPRES and CODE in multiple case studies observe that enterprise models created naturally
and pilot implementations (Kosanke, 1995). CI- provide semantically rich descriptions of ‘rela-
MOSA reference architecture supports modelling tively enduring’ or static aspects of enterprises
of complete enterprise life cycle (requirement but those models do not readily transform into
definitions, design specifications & implemen- equivalent simulation models needed to analyse
tation descriptions). The goal of CIMOSA is to and thereby to predict short and medium term
provide the industry community with widely enterprise behaviours. In reality, however, most if
usable and accepted CIM concepts to structure not all enterprise systems are subject to significant
CIM systems; namely: (a) enterprise modelling and often on going changes during their life cycle
framework that can accurately represent business and thus necessitate simulation to support ‘what-
operations, support their analysis and design, if’ evaluation of potential process and system
and lead to executable enterprise models, (b) alternatives needed to implement organisational
integrating infrastructure to support application design and change. Standards related to enterprise
and business integration and (d) a methodology engineering and integration such as CEN ENV
based on CIM system life cycle. CIMOSA views 13550 (model execution services) provide a basis
an enterprise as a collection of functional entities for computer based development and execution but
processing enterprise objects to realise products as yet no standard model transformation methods
and services. CIMOSA modelling methodology are reported in the public domain literature that
adopts systematic and hierarchical approach to allow selected segments of enterprise models to
complex systems decomposition. At highest level, be executed within virtual environments (Zelm,
CIMOSA views an enterprise as a collection of 2003). Table 1 highlights the modelling capabili-
domains, each domain consisting of domain pro- ties and constraints associated with the current
cesses (DPs). DPs are core-stand alone processes enterprise modelling methods (Rahimifard &
that interact with one another through exchange Weston, 2007).
of events. DPs can further be decomposed into Researchers at the MSI Research Institute at
business processes (BPs), enterprise activities Loughborough University, UK, have deployed
(EAs) and functional operations (FOs). CIMOSA various public domain modelling architectures
details are summarised in the so called CIMOSA and concepts to developing coherent models of
cube and the reader is referred to the references industrial processes and systems. Those models
mentioned (AMICE, 1993, Zelm, et al 1995 & have supported both large and small scale organi-
Vernadat, 1996). sation design and engineering exercises at sites of
ME collaborators. In so doing MSI has produced
various integrated modelling methodologies in-
enHAncInG modellInG cluding those reported by Monfared et al (2002),
cAPAbIlItIes oF current Chatha et al (2007); Rahimifard & Weston (2007)
em solutIons and Agyapong-Kodua et al (2008). For most of
the integrated modelling methods developed
Although the emergence of public domain en- CIMOSA has provided a backbone modelling
terprise modelling architectures and methodolo- architecture and representational formalisms for
378
Table 1. Current EM capabilities & constrains
Capabilities of Current EM Solutions Primary Modelling Constraints

Can formally and explicitly handle enterprise complexity, decom- Resultant models are focused on relatively enduring aspects of
pose complex processes into process segments and their resource MEs and do not encode time dependencies related to process in-
systems and related structural dependences. stants and related product flows, control flows, exception instants,
and other time dependent causal effects.
Enables attribution of multi-perspective model views to models of Resultant models cannot replicate real ME behaviours, such as in
process segments and their related resource systems. support of model validation.
Can represent and help communicate and visualize unified views Resultant models do not provide a basis for prediction of future
about multiple stakeholders. ME behaviours.
Generates explicit representation of different aspects of ME Models created cannot be used to visualise possible future
processes that can inform and support various forms of decision scenarios such as changes in resource systems properties and
making. characteristics.
Has abstraction and generalisation concepts that facilitate the Models are essentially static and therefore cannot fully capture
integration and reuse of multi-perspective and decomposed models enterprise stochastic behaviours.
processes and resource systems.
Has life cycle related concepts which coupled with formal decom- Limited integration and interoperability.
position concepts facilitates system integration and interoperation.
graphical capture of relatively enduring aspects fines of general enterprise modelling principles.
of case study enterprises. The choice of CIMOSA For example, Monfared et al (2002), described
as a foundation backbone for MSI’s modelling how four types of CIMOSA diagramming template
methodologies was made essentially for reasons, (namely: ‘context diagram’, ‘interaction diagram’,
which include: ‘structure diagram’ and ‘activity diagram’) can be
used as a source of enterprise knowledge which can
• CIMOSA enterprise modelling architec- be reloaded using simulation modelling technolo-
ture and method was through experimental gies. Those templates support enterprise engineer-
application by MSI researchers considered ing projects though stages of model development
to provide the most comprehensive public so that modellers and process engineers can move
domain architecture available; iteratively as they structure and organise:
• CIMOSA provides a public domain open
system architectures that support modelling 1. the ongoing capture of coherent and seman-
of different life phases of complex systems tically rich picture of dependent processes,
from requirements definitions through in such a way that key dependencies can be
implementation descriptions (Kosanke, explicitly represented;
1995, Gransier & Werner, 1995, Vernadat, 2. the reuse of multiple coherent views of
1996); multi- process models in support of process
• CIMOSA provides an extensive set of and system simulation.
modelling constructs and representational
formalisms (particularly process-oriented The use of these templates was reported to be
constructs) to enable users create particular effective in structuring the capture and coherent
models of any aspects of enterprise. representation of multiple static views of enterprise
process attributes at needed levels of abstraction.
The eclectic nature of CIMOSA allows expan- Context diagrams are used to organise process
sion of its modelling capabilities within the con- models into interrelated CIMOSA conformant and
379
Figure 3. Approach used to create CIMOSA enterprise models
non CIMOSA conformant domains (DMs), the aspects of specific process instants and lack the
former being the concerned domains. CIMOSA capability to be used in determining and possibly
DMs in the context diagrams are further decom- predicting individual and or collective behaviours
posed into more detailed domain processes (DPs). of system elements. But such predictions are core
Interactions among the DPs triggered by events requirements for organisation design and change
typically result in exchange of (informational, projects in MEs. This is the case because static
material, human, technical and financial) resources EMs are focused on an abstract big picture and its
amongst DPs and these interactions can be rep- decomposition, i.e. on relatively enduring aspects
resented using interaction diagrams. Structure of MEs and do not encode specific time depen-
diagrams are used to formerly attribute relative dencies related to process instants that result as a
enduring organisational relationships that couple consequence of product flows, control flows, ex-
business processes (BPs) and enterprise activities ception instants and time dependent causal effects
(EAs) of a DP while activity diagram templates are that determine factors such as resource utilizations
deployed to explicitly encode relatively short-lived and consumption and associated wastes. On the
descriptions of precedence links used to organise contrary discrete event simulation modelling
the interoperation of BPs and EAs. however can capture and quantitatively describe
Use of CIMOSA ‘context’, ‘interaction’, aspects of dynamic behaviours including queues,
‘structure’ and ‘activity’ diagramming templates stochastic events, product flows, process routes,
enables explicit definition of key structural re- resource utilization, efficiencies, break downs, etc,
lationships between DMs, DPs, BPs and EAs. while dynamic systems modelling instrumented
Typically, those structures will indicate process by qualitative and quantitative causal loop and
segment ownerships, hierarchical ordering, se- continuous simulation models (that can be nu-
quential ordering and precedence relationships, merically integrated) can computer execute and
and necessary interchanges of information, mate- predict dynamic behaviours arising from causal
rial, human resource and finances. dependencies between system variables (Ajaefobi,
Though the diagramming templates enhanced et al 2008). It follows that to support organisation
model creation and descriptions, the models still design and change static models created during
remained static in the sense that they do not encode enterprise engineering projects will in general need
380
to be re-encoded into dynamic views using suitable • bottlenecks generating long queues of
simulation techniques with sufficient constructs WIP
to represent enterprise systems behaviours and • long lead times and unbalanced workflows
properties. In simulation views, change parameters • idle resources (within human or technical
that can trigger system redesign and change can systems)
clearly be observed and different possible solu- • poor quality jobs or high error rates
tion options tried in a virtual environment. In • etc.
general the authors’ experiences of ME modelling
have confirmed the commonly held view that In theory enterprise modelling has significant
continuous simulation is best suited to modelling potential to support organisation design and
dynamic behaviours of either large or small scale change projects because EMs:
systems but with essentially a simplified view of
the stochastic nature of products. While discrete a. present interrelated enterprise sub systems
event simulation (DES) is best suited to modelling as models from different view points
small scale systems in which stochastic behaviours b. models created in respect of (a) ‘paint the
need to be clearly understood. Also observed is big picture’ of an enterprise by explicitly
that the use of DES can be key when a common depicting what the enterprise does and clearly
resource is used to realise multiple products with separating this from who/what does what and
a significant product dynamic but that in general at what time. This can enhance stakeholder
the use of DES should be preceded by creating an understandings of the enterprise.
EM (and possibly causal loop modelling) when c. In dynamic views (where EMs are trans-
supporting enterprise engineering projects. formed into models of ME segments using
suitable simulation technology), enterprise
engineers can observe:
common orGAnIsAtIon desIGn ◦ throughputs from which assessment
& cHAnGe (od&c) trIGGers can be made as to whether the organi-
And tHe role oF em sation’s performance is acceptable
given the available resources and time
OD&C projects can be strategically planned or ◦ make informed decision about wheth-
it can come as part of a response to emerging er current best performance can still
requirements. Generally though, OD&C proj- be improved upon, such as by raising
ects in a given enterprise can be triggered by the work input rates and reducing cycle
need to respond to: perceived opportunities; new times by deploying resources with in-
requirements and threats; or in anticipation to creased efficiency
certain events such as new government legisla- ◦ take decisions about what can be
tion, new product launch by a competitor, new changed to improve the quality,
customer requirements and technological change quantity and timelines of the work
(Mullins, 2005). Planned OD&C projects may throughput
also constitute a program within an organisation ◦ observe production system perfor-
to improve its operational efficiency and current mance under changing scenarios in-
best performance. Practical examples of common cluding the impacts if performance
triggers that would naturally lead to organisation metric of the technical systems and
(re)design and change in a typical production the proficiency with which people use
enterprise include: of those technologies are changed
381
◦ assess impacts and observe impacts and applications. Manufacturing variables change
of changing competences, commit- at discrete points in time. For example, the time
ment and motivation of the people to complete a manufacturing operation may vary
◦ assess impacts on performances each time the operation is undertaken, which can
of alternative resource systems make discrete event simulators more suitable in
configurations comparison to continuous simulations. In many
◦ can determine policies for improving production situations orders from customers arrive
interoperability and communications at unpredictable times and commonly involve un-
and feedback between people on the certain quantities and mixes of product realisation
shop floor with their managers. at any point in time. This makes it very difficult
to develop equations that adequately characterise
different production scenarios which can be solved
sImulAtIon modellInG (sm) via numerical integration methods. Whereas DES
In suPPort oF od&c models can support enterprise systems perfor-
mance observation and analysis under different
Simulation modelling is an established means of possible scenarios where alternative resource
understanding and predicting systems behaviours configurations are loaded by stochastic order,
which has received great attention in literature. information and material workflows. Knowledge
Simulation technologies have been widely ap- obtained from such performance analysis could
plied in manufacturing and service organisations inform and support management decisions about
(Barber, et al, 2003). Generally, simulated systems OD&C parameters such as:
imitate real systems. Harrison & Petty, (2002)
described why simulation is becoming important 1. what needs to be changed, improved or
tool in science and engineering: supported amongst the processes & systems
deployed by the enterprise to realising spe-
• safety – simulation allows a system to be cific objectives
tested under extreme conditions that could 2. what control measures are needed to improve
not be justified in a real world scenario systems performances and the appropriate
• practicality – with large systems, simula- times to put such measures in place
tion may be the only practical way of ap- 3. how informed decisions can be made about
proaching a problem new competences, capabilities and capacities
• simulation allows what if analysis – this that may be needed by resource systems to
allows the physical system to be built and improve their throughput and the over all
tested under different solution options performance of systems
• simulation is repeatable 4. what training system operators require and
• simulation promotes understanding of sys- the contents of such training based on cur-
tems, particularly if allied with graphics. rent performance and observed gaps
• simulation can be used to verify analytical 5. needed technology to enhance system per-
solutions to problems formance (software and hardware)
6. how to enhance cost engineering by using value
Though continuous simulation has been use- stream maps and activity based costing
fully applied in manufacturing cases, it is discrete 7. what wastes, errors and reworks, etc can be
event simulation (DES) technology that has been reducing thus setting the ground for Lean
adopted in the majority of manufacturing studies manufacturing implementation
382
8. ways of reconfiguring enterprise processes relationship modelling notations, provided by

& systems to realised enhanced performance causal loop modelling (CLM) technologies, (c)
in terms of quality, quantity, timeliness and discrete event simulation models (SM) of selected
overall efficiency configurations of workloaded process segments
9. ways of improving organisation structure and their underpinning resource systems and
and reporting lines can also be observed (d) mixed reality modelling, based on the use
of workflow modelling (WFM) techniques that
To summarise DES models have potential to enable interaction and information interchange
(a) detect and eliminate problems that otherwise between simulation models and real resource
would require cost and time consuming correc- systems. Figure 4 is an illustrative description of
tive measures during production ramp up, (b) the modelling approach adopted, which integrates
determine optimum times such as processing EM, CLM, SM & WFM to create comprehen-
times, failure times, recovery times, etc, and sive, coherent and semantically rich process and
throughput of the plant, (c) determine size of system models in support of organisation design
buffers and numbers of machines with intended and change projects.
throughputs and (d) investigate how failures af- This modelling approach is usually realised
fect the throughput and the utilisation of resource by taking the following steps:
elements and systems.
However DES models of production situations 1. elicit and capture actual plant data from a tar-
typically soon become very complex if the real- get organisation, about its process network,
ity is closely modelled. Hence the scope of DES product and work flows and its resource
models need to be restricted. As a consequence the systems
modeller needs to guard against optimising only 2. create ‘as is’ EM using suitable a enterprise
a small part of a given ME rather than consider modelling method
optimising overall ME performance. It follows 3. validate/modify the ‘as is’ EM
that EM and SM techniques are complementary 4. encode focused aspects of the ‘as is’ EM into
with the former designed to provide an explicit simulation views using a suitable simulation
structural decomposition of the whole ME and tool
the latter can enable detailed analysis of possible 5. validate/modify the ‘as is’ simulation models
uncertain behaviours in specific segments of the (SMs)
EM where that analysis can be made with the 6. develop candidate ‘to be’ SMs
context of the big picture captured by the EM. 7. validate/modify the behaviours of the ‘to
For more than two decades, researchers in be’ SMs
MSI Research Institute, Loughborough University 8. integrate the operations of ‘to be’ and ‘as is’
have collaborated with various industrial partners SMs to predict an organisation’s performance
while developing and deploying coherent and in selected aspects when subjected to alterna-
semantically rich ME models drawn from real tive strategies/policies and requirements
industrial data. Essentially, MSI’s modelling 9. validate and modify as required the integrated
approach unifies the use of: (a) manufacturing SMs
enterprise decomposition principles and multi- 10. recommend and justify improve organisation
perspective graphical modelling templates as designs
defined by public domain enterprise modelling
(EM) methodologies, (b) causal and temporal
383
Figure 4. Illustrative description of the modelling approach
IllustrAtIve cAse the current trend (market share increase) continue.

study APPlIcAtIon To extend its current market share naturally re-
quire continuous performance improvement of
The modelling approach and steps described in its processes and resource systems. Funded by
previous sections have been deployed in a wide EPSRC in the UK, the authors are collaborating
range of case studies carried out within automo- with Woodlands managers in seeking to deploy
tive, machine building, composite and furniture best industry practices to support the company in
industries. An instance of such applications in a reorganising its resource systems to ensure profit-
SME furniture making organisation is described ability and competitiveness. At the commence-
in outline in this section to illustrate the complement of the project, the following observations
mentary natures of EM and SM techniques when were made:
used in organisation design and change situations.
• Woodland’s processes were not formerly
case study background & documented; rather understandings about
modelling requirements them existed in the ‘minds’ of the manag-
ers and operators. A Woodland’s organisa-
Woodland is a furniture making SME located in tional chart depicted people and their roles,
East midlands in England. It is a make to order but documented information about people
furniture business with over 60 regular workers. responsibilities was very limited and
Recent increase in market share resulted in man- could not show dynamics associated with
agement concerns about the company’s capacity roles including ‘requirements to perform
to cope with increasing customer demands, should role’, ‘inputs to roles’, ‘role performance
384
outputs’, ‘resource utilisation’ and ‘im- thereafter to suggest improved ways of

pacts of requirement changes on roles and deploying a finite resource
role incumbents’. 4. Observe and monitor production system
• Woodland’s production systems operate reaction when processing ‘cognate’ and
in a ‘push’ production fashion with inputs ‘distinct’ work item types; so as to analyze
and outputs of both up & down streams and the impact of variations in required
not properly regulated with respect to a quantities of such work items on the system
give takt time. Consequently during nor- capability and capacity
mal operations there is significant WIP and 5. Suggest improved production planning and
long queues which also results in an untidy control methods
working environment and raises questions 6. Investigate the implications of operating a
about conformance to health and safety full automated system in comparison with
regulations. the current semi-automated status quo
• Woodland’s production systems are semi- 7. Suggest alternative ways of running the
automated, deploying competent operators production system, thereby making it more
to dynamically resource their wood cut- flexible, lean and agile
ting, furniture making machines and work 8. Suggest possible changes in organisational
benches as significant fluctuation in cus- structure, company policies and management
tomer order is the norm philosophy
• Though Woodland currently meets its
lead time targets, the management re- case study modelling
quires further lead time reduction to re- Approach & methodology
main competitive with alternative overseas
manufacturers The modelling approach described was deployed
• Woodland management desires improved to elicit data and formally decompose and explic-
resource utilisation and/or better perfor- itly document three main processes identified in
mance without compromising the well be- Woodland, namely:
ing of its resource systems.
(a) ‘production processes’ that on a day to day
Against the backdrop of the observations made, basis make and deliver furniture to aggre-
project objectives were agreed as follows: gated order
(b) ‘new product introduction processes’ that
1. Using suitable EM constructs and represen- operate annually to develop and introduce
tational formalisms to abstract, represent and new furniture products, replacing around
document Woodland processes and systems, ten percent of products
so as to identify who does what and with (c) ‘strategy making processes’ that operate as
what competences required to conceive and realise improved
2. Encode aspects of Woodland processes competitiveness.
and systems in a dynamic (virtual reality)
view, thereby better understanding current The processes identified and modelled corre-
throughput, performances and resource spond to three generic processes commonly found
utilisations in MEs namely strategic, tactical and operational
3. Observe and monitor system performance process types (Pandya et al, 1997). Details of these
using different resource configurations and key processes were modelled using CIMOSA ISO
385
standard modelling techniques. Figure 5 shows In such systems, work item flows will typically
a high level context diagram created to model involve flows of raw material, (sub) components,
interacting domains identified when Woodland sub assemblies and products. Additional work
and its supply chain partners realise a ‘make and items flows may take the form of different in-
deliver furniture to order’ process; while figure formation types (such as customer order types,
6 is one of the structure diagrams generated to works order types, product specification types,
explicitly model elements of business processes etc) which need processing into alternative forms
and enterprise activities. The domains shown in that have increased value in the business context
figure 5 interact with one another at different under study. The term ‘distinctive work item types’
levels via exchange of information and resources implies that ‘significant differences’ in processing
in other to successfully realise ‘make & deliver requirements will be necessary to achieve value
furniture to aggregated order’ process addition as different work item types pass through
the system. Here it is assumed that:
Woodland Production system
(A) ‘cognate work item types’ (such as sub-
Woodlands production system was observed to be product and product item types within the
a multiple value stream production system. Here, same product family) can share a similar
the present authors consider a multiple value value adding process. Therefore also as-
stream production system as being: “any system sumed is that for cognate work item types,
which adds value to two or more ‘distinctive work relatively minor differences between item
item flows’; by realising necessary value add- types (such as dimensional or operation time
ing processing operations through an organised differences) will need to be accommodated
sharing of the (human and technical) resources by the assigned production system during
available to that system”. process set up and process execution. In
Figure 5. Top level context diagram showing Woodland domains
386
such cases, variation within a product family generate distinctive process dynamics (such
might be accommodated via use of program- as process bottlenecks, inventory level pat-
mable machines and product type sensing terns and so forth). Hence production systems
mechanisms, or by deploying people with used to realise distinctive work item types
the requisite competencies needed to set will likely require correspondingly distinc-
up and process all input item types within tive work centre processing capabilities and
a family. competencies, and distinctive workflow
rules and controls. This requirement is also
(B) ‘distinctive work item types’ will belong likely to be additional to the need to cope
to different product families, in the sense with in-family product variation via the use
that the sub-product and product item types of flexible resource system elements.
belonging to different families will each
require a distinctive value adding process. The Woodland project is still on going but the
Typically this will require alternative pro- progress reported considers the processing of cog-
cessing routes through work centres of nate product types in the assembly shop and how
production systems, and the various work this shop operations could be changed to improve
item flows within each family will likely performance. Following the modelling method
Figure 6. Woodland Structure diagram showing some business processes

Context Diagram Site Map
BP71
BP72 BP73
Make Furniture
to Order Spray & Finish Package & Deliver
BP71-1 Furniture Production Furniture
Machine Furniture BP72-1 BP73-1
Components Spray & Rag Carcasses Inspect & Package
and Components Furniture
Store/Transport Components For Assembly (EA711)
BP73-2
Prepare Components for Assembly (EA712) Dry Furniture Carcasses and components (EA721)
BP72-2 Position Furniture in Queue
BP71-2
De-nibble & Reassemble
Assemble Carcasses
Furniture
& Fit Components Load Furniture in Lorries (EA731)
BP72-2
BP71-3 Fix Fittings Notify OP about Dispatch (EA732)
Disassemble Carcasses
& Components for Assembly
Fine Finish Furniture (EA722) Transport furniture to Stockists (EA733)

Verify Satisfactory Delivery (EA734)
DP7
BP74 BP75
Support & Implement Produce & Deliver Manage & Maintain Production
Product Introduction Furniture & Transport Capabilities
BP74-1 BP75-1
Program CNC M/Cs & Create Manage Day to Day Utilisation
Jigs for New Furniture of Production Resources
BP74-2 BP75-2
Prototype Machine New Comp Maintain Production Machines
& Facilities
Identify Needed Component Machining & Storage Procedures (EA741) BP75-3
BP74-3 Manage Day to Day Utilisation
Program CNC M/Cs & Create of Transport Resources
Jigs for New Furniture Assembly BP75-4
BP74-4 Service and Maintain Lorries
Prototype Assemble New Comp
BP75-5
Identify Needed Component Assembly & Storage Procedures (EA742) Monitor Use of Domain
BP74-5 Resources & Recommend Change
Prototype Spray & Finish New Comp BP75-6
Monitor Operations of BPs and
Identify Needed Component Spraying Procedures (EA743)
Recommend Change
BP74-6
Prototype Package New Comp
Identify Needed Component Packaging Procedures (EA744)
CIMOSA Domain Activity Event Information Human Resource Physical Resource Finance
External Links RHW
Non CIMOSA Domain Flow of Res./Mat. Flow of Process Alternative Flow 1
- Structure Diagram Design by: J.O.A

Woodland Furniture- Process Modelling
Checked by: RHW Woodland
Title:
Make To Order Furniture Domain Number:
CTX- Furniture Prd/Assy
Last update Furniture
387
described in section 6, an assembly process model grams. Figure 7 is an illustrative description of

captured using EM technology was re-encoded decomposition of a CIMOSA domain into lower
in a dynamic view using Siml8, a discrete event level BPs, EAs and FOs.
simulator. The initial objective was to replicate the Here, the rationale was to develop dynamic
assembly shop behaviours over time, including how (simulation) models of different process segments
it responds to changing work item types. Simul8 and subsequently to link the modular models
has four basic building blocks namely: into a coherent system model that captures more
holistically entire system behaviours. Following
1. Work Entry Points through which work items this approach, the operations of the door assem-
to be processed enter into the system. The bly section of the assembly shop were encoded
arrival pattern is controlled either to follow a in a simulation view. The ‘as is’ weekly average
scheduled pattern (deterministic behaviour) throughput of 800 – 900 completed product units
or a stochastic (probabilistic) behaviour was assumed as a benchmark. Resource systems
2. Queues – hold work items that are waiting utilisations and queues associated with this
to be processed throughput were observed based on Woodland’s
3. Work centres that perform actual value single shift work pattern of 8 hours (excluding
adding operations under varying prob- break and lunch times). To cope with anticipated
ability distribution of processing times and customer demands, Woodland management de-
efficiencies sired that the throughput of this section should be
4. Work Exist Points where completed work raised to: 1200 units per week in the short term
are stored. (2-3 months) using the present available resources
and resource configuration; and to double that in
Two other important Simul8 modelling con- the long term (12 months time) when Woodland’s
structs that were found useful in dynamic model management expected to increase its investment
development are work item types (objects to be in resources, including increasing the number of
processed) and the resources that are required to machine tools and possibly its human resource as-
realise operations. These main building blocks signments used to complete work done in the door
were used to encode static models previously assembly section. Presently, the technical resource
developed. Subsequent sections of this paper systems in this section of Woodlands production
describe how simulation models of Woodland’s system include; a cutting machine, three parallel
assembly system were created. milling machines, a sanding machine; assembly
benches (where operators use different types of
Woodland simulation modelling (sm) electrically powered hand tools) and final assem-
bly benches (where operators used different hand
A modular approach was adopted while develop- tools to complete product builds). In the ‘as is’
ing Woodland dynamic models using simulation simulation model, batches of 10 and 25 of the two
technology. Modularisation was naturally sup- components (‘panels’ & ‘frames’) respectively are
ported by the decomposition mechanisms used routed to the cutting machine at hourly intervals.
previously to capture CIMOSA conformant ‘static’ The operator begins with any batch and alternates
models that had also previously been validated by between batch types while taking an average of
the company management. CIMOSA modelling 20 to 30 minutes to switch over between jobs.
constructs were used to further define top level This order and change over times (which in effect
processes into lower levels of granularity (BPs, are down times) are necessary in the real system
EAs & FOs) using interaction and structure dia- because the two components share the same ma-
388
Figure 7. CIMOSA hierarchical decomposition modelling approach
DM2
Raw Material
Supplier
DM1 DM3
Stockists Sub-Product (e.g.
Chair) Supplier
Make & Deliver

DM7 Furniture to DM4
Woodlands Produce (aggregated) Order Miscellaneous
& Deliver Furniture Fixture Supplier
BP71
BP72 BP73
Make Furniture
to Order Spray & Finish Package & Deliver
Furniture Production Furniture
BP71-1 DM6 DM5
Machine Furniture
Components
BP72-1
Spray & Rag Carcasses
BP73-1
Inspect & Package
Business Technology Vendors
and Components Furniture Management
BP73-2
Prepare Components for Assembly (EA712) Dry Furniture Carcasses and components (EA721)
BP72-2 Position Furniture in Queue
BP71-2
De-nibble & Reassemble
Assemble Carcasses
Furniture
& Fit Components Load Furniture in Lorries (EA731)
BP72-2
BP71-3 Fix Fittings Notify OP about Dispatch (EA732)
Disassemble Carcasses
& Components for Assembly
Fine Finish Furniture (EA722) Transport furniture to Stockists (EA733)

Verify Satisfactory Delivery (EA734)
BP74
DP7
Support & Implement
Product Introduction Produce & Deliver BP75
Furniture Manage & Maintain Production

BP74-1
Program CNC M/Cs & Create & Transport Capabilities
Jigs for New Furniture
BP75-1
BP74-2 Manage Day to Day Utilisation
Prototype Machine New Comp of Production Resources
BP75-2
Identify Needed Component Machining & Storage Procedures (EA741)
Maintain Production Machines
BP74-3 & Facilities
Program CNC M/Cs & Create
BP75-3
Jigs for New Furniture Assembly
Manage Day to Day Utilisation
BP74-4 of Transport Resources
Prototype Assemble New Comp BP75-4
Identify Needed Component Assembly & Storage Procedures (EA742) Service and Maintain Lorries
BP74-5
BP75-5
Prototype Spray & Finish New Comp
Monitor Use of Domain
Identify Needed Component Spraying Procedures (EA743)
Resources & Recommend Change
BP74-6 BP75-6
Prototype Package New Comp Monitor Operations of BPs and
Recommend Change
Identify Needed Component Packaging Procedures (EA744)
Produce Customer
BP72 BP73
BP71
Orders DP7
Spray and Finish Package & Deliver

Make Furniture to Order Production Furniture
EA711
Set –up NC Machine
Package & Deliver Furniture BP73

EA712
Assemble Cut frames Assembly BPs
NC Machine, Jigs &
Fixtures,Raw Materials
Machine Furniture Doors
Components Cut panel
Picking List EA713
Machinists Store/Transport
Components for Assembly Collect components
EA714 Assemble
Prepare Components for Assembly
Tables
Mill components
BP71 -2 Collect
Assembly
Tools Assemble Carcasses & components
Components into Furniture EA716 Assemble Sand components
Picking List Assemble
Make Furniture Disassemble Furniture D’Units under frame
to Order BP71 Assemblers for Spraying Join components
Fit frame
To top
Install hinges
Fit Legs
Assemble
Install handles
Others Install
features
chine for cutting operations and secondly, both related resources were also observed. Similarly it
components are needed further down stream for was observed that the human operators working
subsequent assembling. in this section, though apparently were seen to be
From the SM results however, it was observed working all day, their utilisation in terms of time
that the down time due to switching between the spent on value adding operations was also low.
two jobs was rather high. The implication here was It follows that much time is spent on non value
that the machine spends one third of its production adding operations, which is a trigger for change.
time per day on ‘changing over’, thereby contrib- Table 2 shows the ‘as is’ simulation results and
uting significantly to under-utilisation of the said observed utilisation of some selected resources.
machine and its human operator. Furthermore, In summary, 927 product units were assembled
since the said machine feeds other machines further and the minimum time spent by a unit in the
down stream, delays and under-utilisations of other system was about 60 minutes, even though the
389
Table 2. Observed ‘As - Is’ simulation results
As – Is Scenario
Resource Work Machine Change waiting Queue Operator Total
Types done Utilisation % Over % Utilisation output
%
Cutting machine 1145 61 33 6 114 100 Total = 927
units made
Milling 1 1091 69 _ 31 2 69
Assemble bench 1 633 66 _ 34 2 43
Panel Finishing 1169 97 _ 3 110 98
Finishing 69 _
minimum ‘make span’ of a single unit was as low Consequently, work entry rate and batch sizes
as 12 minutes. This implies that there are delays were increased and safety buffers introduced so
in the system. Figure 8 is the screen print out of that down stream resources are not kept waiting
the ‘as-is’ door assembly simulation. while the cutting machine is processing the first
Following the observations made, a possible batch of the day. Furthermore, it was observed
to be scenario was simulated for this section. that switching between jobs in the cutting ma-
Here, the target was to realise the management’s chine could be done in an average of 10 minutes
goal of 1200 units per week from this section. and that the operators had previously been over
Figure 8. Screen print out of ‘as-is’ door assembly operations
390
indulging themselves, hence the new change over in the preceding paragraph, the throughput and the
time for cutting machine was set between 10 -15 resource system utilisations were observed. Fol-
minutes depending on the operator’s competences lowing the observations, suggestions were made
and so called tacit knowledge. The simulation on how the section throughput and over all per-
results showed significant increase in throughput formances could be improved by implementation
from 927 to 1251, representing a 35% increase. certain OD&C principles observed . Suggestions
Furthermore, queues were significantly reduced made were based on consideration of different
especially the queue to the cutting machine. A resource units as components that need for them
closer investigation of the behaviour of the queue to interwork to improve the over all performance
to the cutting machine showed a rising and falling of the assembly shop. Subsequent section briefly
pattern, with a peak value occurring at mid day. The describes the background for component based
same pattern is repeated during weekly simulation approach to organisation design and the role of
run with peak at mid week. This possibly captures enterprise modelling and engineering.
the mood of the human operator who slows down
his pace as the days goes on and then increases
after lunch break. Table 3 shows the performance comPonent-bAsed
results of some selected resource systems. orGAnIsAtIon desIGn And
Adjacent to the door assembly unit is the table cHAnGe APProAcH
assembly section. This section was also encoded
in a simulation (virtual reality) view. Over 3 dozen BPR (Hammer & Champy, 2001) advocates
table types are currently being assembled by a radical approach to organisation design and
Woodland in this unit. However, the authors con- change, involving fundamental rethinking and
solidated them into three broad groups based on the radical redesign of business process to bring about
commonality table features; including complexity, dramatic improvement all critical performance
routing and processing times. In general, products measures. Yogesh (1998) however observed that
belonging to the Woodlands table family are as- 70% of BPR projects reviewed did not yield the
sembled from three main components namely; envisaged dramatic results. Essentially, the reason
table tops (TT), under frames (UDF) and legs given for failed BPR projects was that BPR focus
(TLG). The table assemble operations involved mainly on organisation’s processes leaving other
produced the tables to a design specification us- component parts of the organisation unchanged.
ing primary components that are delivered from OD & C project should take a multi perspective
Woodland machine shop. Like the case described view of the organisation. Similarly, OD&C can-
Table 3. ‘To - Be’ scenario simulation results
To – Be Scenario
Resource Types Work Machine Change Waiting Queue Operator Total
done Utilisation % Over % Time % Utilisation % output
Saw machine 1240 66 13 15 0 81 Total =1251
units made
Milling 1 1235 78 _ 22 0 78
% increase
Assemble bench 1 745 78 _ 22 0 80 = 35
Panel finishing 1178 98 _ 2 71 98
Finishing 1267 79 _ 21 268/0 79
391
not be realised by merely reshuffling of workers ments. One way of viewing resource systems is
followed by changes in organisational structure that they are building blocks or components with
without changing the process design, procedures embedded range of attributes and characteristics
and work item flows. It follows that OD&C is such as competences/or capabilities that enables
not only a radical process reengineering which them to act individually or collectively to realise
seeks to change process designs and sequences specified business requirements. It follows that
of operation without change to resource systems for organisation design and change to be effec-
compositions, structures and configurations. tive, resource components should be agile and
OD&C projects should take a holistic view of the responsive with inherent ability to be reformed
entire organisation bearing in mind the causal and or reconfigured so that individual and collective
temporal relationships between all organisation behaviours of organisation system components
system components. This can be instrumented could rapidly be aligned to meet specific and
by EM & engineering concepts, which deploys changing business process needs. This implies
multi perspective views of organisation (structure, that organisations should have:
processes, resources & information) and how these
views dynamically interact and impact on one 1. suitable components or building blocks from
another; including the impacts of such interactions which high performance systems are to be
on the over all organisation’s goals. configured
In general, a manufacturing organisation is an 2. organisational structures and associated
open system with inputs from the: mechanisms capable of establishing flexible
linkages between different components, so
• environment (which provides constraints, that systems can be configured rapidly and
demands and opportunities) yet have inherent ability to be reconfigured
• resource systems that realise value adding and reengineered in the event of requirement
operations to deliver products and services changes
• mission and strategies that link resource 3. means of supporting rapid and holistic defi-
systems to processes, opportunities, con- nition and redefinition of system behaviours
straints and demands so that unwanted be- in the form that can be used to help specify,
haviours are constrained implement, control and change individual and
collective behaviours of system components
Responding to requirement changes which to facilitate organisation design and change.
generally will necessitate OD&C is complex and
difficult because it implies changes in most if not Generally though, high performing dynamic
all the organisation inputs including: strategies, systems should be built from components by
policies, processes, procedures, work organisation (a) selecting and programming components
structures and resource systems configurations behaviours and (b) defining and programming
(Nadler, 1993). Weston (1998) suggests that ef- system architectures. This will result in composed
fective and timely responses to change require- resource systems that can be defined in terms
ments will necessitate change capable organisation of their reachable states. To respond to require-
systems and structures. Resource systems are ment changes; (a) component behaviours are
integral parts of organisation that are configured re-selected and re-programmed and (b) system
in a variety of ways to generate different behav- architectures are re-defined and re-programmed,
ioural responses needed by the organisation to thereby resulting in a re-composed or changed
realise specific and emerging business require- system. It follows that system components of
392
organisations should possess sufficient flexibility 1. incorporating sufficient redundant capabili-

characteristics especially ‘change capability’ and ties into initial system composition so that
‘change capability rate’. Harrison et al, (2001) possible alternative behaviours could be
present change capability classes and related readily programmed to enable the system
them to change scenarios that can beneficially reach predictable change of states, possibly
be applied as follows: on frequent basis
2. structuring the system and its integrating
1. ‘programmability’: ability to programme mechanisms in such a manner that its com-
system behaviours and or composition so position could be modified rapidly and ef-
that the system can reach a range of well fectively in response to change requirements
known states, thereby providing means of 3. designing and building the organisation in
addressing changes of predictable nature such a way that it can easily and rapidly be
2. ‘reactivity’: ability to react to changes of decommissioned and or replaced(if need
unpredictable nature by modifying system be) in response to change requirements and
behaviours or compositions environmental conditions beyond the scope
3. ‘proactivity’: ability to prepared for of the constituting system capabilities
modification of system behaviours and
compositions
conclusIon
Similarly, the ‘change capability rate’ includes
notions of engineering ‘costs and time’ when Models provide means of abstracting and manag-
a composed system is moving between states. ing different components of complex organisa-
Change capability rate therefore depends on: tions. An organisation such as a manufacturing
enterprise (ME) has many interacting component
• the properties of the components (modular parts that can be represented at different levels
building blocks) from which systems are of granularity. Current business requirements
composed, which impact on the ease and demand that most if not all organisation parts and
extent to which system programming, sys- systems should be responsive and change capable.
tem reaction and system pro-action could Models of concerned parts of any given ME make
be specified and realised it easier to visualise and analyse characteristic
• the structure and mechanisms used to inte- behaviours of alternative configurations of enter-
grate various parts of a system as this can prise components thereby supporting production
enable or constrain subsequent changes to planning and performance monitoring. By so do-
system composition and behaviours ing, organisation design and change projects are
• the type of change processes and support- supported. EM and its supporting architectures
ing technology that are available to modify and methodologies provide solutions to complex
system composition or behaviours systems modelling, thereby facilitating organisa-
tion design and change projects. The case study
Therefore, to design and build change capable reported described how EM & SM can be used to
organisation systems from system components, inform OD&C project via top-down and bottom-up
the following measures are needed to support modelling approaches. In the former, Woodland
improved organisation flexibility: was modelled in context with all the domains that
interact to realise ‘make furniture to aggregated
order’ while the later approach described how
393
elemental low level enterprise activities and opera- Bernus, P. (2003). Enterprise models for enter-
tions and resource units assigned to execute them prise architecture and ISO9000:2000. Annual
can be reorganised to realise improved system Reviews in Control, 27, 211–220. doi:10.1016/j.
performance. arcontrol.2003.09.004
Blethyn, S. G., & Parker, C. S. (1990). Designing
information systems. Butterworth-Heinemann.
reFerences
Bravocco, R. R., & Yadav, S. B. (1985). Require-
Agyapong-Kodua, K., Ajaefobi, J.O., & Weston, ments definitions architecture – an overview. Com-
R.H. (2009). Modelling dynamic value streams puters in Industry, 6, 237–251. doi:10.1016/0166-
to support process design and evaluation. Int. J. 3615(85)90046-6
of Computer Integrated Manufacturing.
Chatha, K. A., Ajaefobi, J. O., & Weston, R. H.
Ajaefobi, J. O., Wahid, B., & Weston, R. H. (2007). Enriched multi-process modelling in
(2008). Proc. of the ICAM, Western Michigan support of the life cycle engineering of business
University, USA. processes. International Journal of Production
Alison, B., O’Sullivia, O.A., Rice, J., Rothwell, A., Research, 45(1), 103–141.
& Sanders, C. (2001). Research skills for students. Christensen, L. C., Johansen, B. W., Midjo, N.,
AMICE (Ed.). (1993). Open system architecture Ornarheim, J., Syvertsen, T. G., & Totland, T.
for CIM, ESPRIT Consortium (2nd ed.). New York: (1995). Enterprise modelling-practices and per-
Springer-Verlag. spectives. In Proc. of the ASME 9th Engineering
Database Symposium, Boston, US.
Ashfort, B. E. (2001). Role transition in organi-
sational life: An identity-based perspective. Mah- Davenport, T. H. (1993). Process innovation: Re-
wah, NJ: Lawrence Erlbaum Associates. engineering work through information technology.
Boston, MA: Harvard Business Press.
Barber, K. D., Dewhurst, F. W., Burns, R. L. D.
H., & Rogers, J. B. B. (2003). Business–pro- Davis, L. E. (1982). Organisational design. In G.
cess modelling and simulation for manufactur- Salvendy (Ed.), Handbook of industrial engineer-
ing management: A practical way forward . ing (pp. 2.1.1-2.1.29). New York: Wiley & Sons.
Business Process Management Journal, 9(4). Doumeingts, G., Vallespir, B., & Marcotte, F.
doi:10.1108/14637150310484544 (1995). A proposal for integrated modelling of
Barber, M. I., & Weston, R. H. (1998). Scoping manufacturing system: Application to the re-
study on business process reengineering: Towards engineering of an assembly shop. Control Engi-
successful IT application . International Journal neering Practice, 3(1), 59–67. doi:10.1016/0967-
of Production Research, 36, 574–601. 0661(94)00065-O
Berio, G., & Vernadat, F. B. (1999). New de- Drucker, P.F. (1990). The emerging theory of
velopments in enterprise modelling using CI- manufacturing. Harvard Business Review, May-
MOSA. Computers in Industry, 40(2), 99–114. June(3), 94-104.
doi:10.1016/S0166-3615(99)00017-2 Evans, J. R., & Lindsay, W. M. (1993). The
management and control of quality (2nd ed.). Min-
neapolis, MN: West Publishing Coy.
394
Fox, M. S., & Gruninger, M. (1998). Enterprise Mintzberg, H. (1989). Mintzberg on manage-
modelling. AI Magazine, 109–112. ment: Inside our strange world of organisations.
London: MacMillan.
Goldman, S. L., Nagel, R. N., & Preiss, K. (1995).
Agile competitors and virtual organisation. New Monfared, R. P. (2000). A component–based
York: Van Nostrand Reinhold. approach to design and construction of change
capable manufacturing cell control systems. Un-
Gransier, T., & Werner, S. (1995). Validation of
published doctoral dissertation, Loughborough
CIMOSA. Computers in Industry, 27, 95–100.
University.
doi:10.1016/0166-3615(95)00048-0
Monfared, R. P., West, A. A., Harrison, R., &
Hammer, M., & Champy, J. (2001). Reengineering
Weston, R. H. (2002). An implementation of
the corporation, a manifesto for business revolu-
business process modelling in automotive in-
tion. New York: Harper Business.
dustry. In Proc. of the Instn. of Mech. Engineers,
Handy, C. (1993). Understanding organisations Vol. 216 Part B: Engineering Manufacture (pp.
(4th ed.). New York: Harmondsworth, Penguin. 1413-1427).
Harrison, D. K., & Petty, D. J. (2002). Systems Mullins, L. J. (2005). Management & organisa-
for planning and control in manufacturing. New- tional behaviours (7th ed.). Upper Saddle River,
ness, Oxford. NJ: Prentice Hall.
Harrison, R., West, A. A., Weston, R. H., & Nadler, D. A. (1993). Concepts of management
Monfared, R. P. (2001). Distributed engineering of organisational change. In Mabey, et al (Eds.),
of manufacturing machines. In Proc. of the Instn. Managing Change (2nd ed.). PCB Ltd.
of Mech. Engrs., Vol. 215, Part B Journal of En-
Pandya, K. V., Karlsson, A., Sega, S., & Carrie,
gineering Manufacture (pp. 217-231).
A. (1997). Towards the manufacturing enterprise
Jang, K.-J. (2003). A model decomposition ap- of the future. Int. Journal of Production. Manage-
proach for a manufacturing enterprise in business ment, 17, 502–521.
process reengineering. International Journal
Rahmifard, A., & Weston, R. H. (2007). The
of Computer Integrated Manufacturing, 16(3),
enhanced use of enterprise and simulation model-
210–218. doi:10.1080/0951192021000039594
ling techniques to support factory changeability.
Kosanke, K. (1995). CIMOSA- overview and International Journal of Computer Integrated
status. Computers in Industry, 27, 101–109. Manufacturing, 20(4), 303–328.
doi:10.1016/0166-3615(95)00016-9
Ramenyi, D. (1998). Doing research in business
Martin, J. (2005). Organisation behaviour and and management: An introduction to process &
management (3rd ed.). Florence, KY: Thomson method. London: SAGE Publications.
Learning.
Ross, D. T., & Schoman, K. E. (1977). Structured
Mills, J., Platts, K., & Bourne, M. (2003). Com- analysis requirements definitions. IEEE Trans-
petence and resource architectures. International actions on Software Engineering, SE-3, 6–15.
Journal of Operations & Production Management, doi:10.1109/TSE.1977.229899
3(9), 977–994. doi:10.1108/01443570310491738
395
Siemieniuch, C.E., Sinclair, M.A., & Vaughan, Williams, T. J. (1994). The Purdue enterprise refer-
G.M.C. (1998). A method for decision support for ence architecture. Computers in Industry, 24(2-3),
allocation of functions and design of jobs in manu- 141–158. doi:10.1016/0166-3615(94)90017-5
facturing based on knowledge requirements. Int.
Womack, J. P., Jones, D. T., & Roos, D. (1990).
Journal of Computer Integrated Manufacturing.
The machine that changed the world. New York:
Vernadat, F. B. (1996). Enterprise modelling & Oxford Associates.
integrations: Principles and applications. Lon-
Wu, I. L. (1992). Manufacturing systems design
don: Chapman & Hall.
and analysis. London: Chapman & Hall.
Vernadat, F. B. (2004). Enterprise modelling &
Yogesh, M. (1998). Business process redesign:
integration: Myth or reality? France: MACSI/
Business change of mythic proportions? MIS
INRIA & LGIPM, University of Metz.
Quarterly, 121–127.
Wainright, C. E. R., & Ridgeway, K. (1994). The
Zelm, M. (2003). Towards user oriented enterprise
application of GRAI as a framework for manufac-
modelling- comparison of modelling language
turing strategy process, Factory 2000-Advanced
constructs. In R. Jardim-Goncalves, et al (Eds.),
Factory Automation, Conference Publication, No.
Proc. of the 10th ISPE Int. Conf. on CE: Research
398. IEEE Press.
and applications. Madeira Portugal: Swets &
Warnecke, H. J. (1993). The fractal company Zeitilinger.
- a revolution in corporate culture. New York:
Zelm, M., Vernadat, F. B., & Kosanke, K. (1995).
Springer-Verlag.
The CIMOSA business modelling process. Com-
Weston, R. H. (1998). Integration infrastructure puters in Industry, 27, 123–142. doi:10.1016/0166-
requirements for agile manufacturing systems. 3615(95)00018-2
Proc. of Instn of Mech. Engineers., 212(Part B),
Zhen, M., & Weston, R. H. (2006). Simulating
423-437.
dynamic behaviours in complex organisations:
Weston, R. H., Clements, P. E., Shorter, D. N., Case study application of a well structure model-
Hodgson, A. J., & West, A. A. (2001). On the ex- ling approach. In A. Nketsa, et al (Eds.), Proc. of
plicit modelling of systems of human resources. In- the European Simulation and Modelling Confer-
ternational Journal of Production Research, 39(2), ence, Modelling and Simulation (ESM’2006),
185–204. doi:10.1080/00207540010001857 EUROSIS-ETI, Toulouse, France (pp. 390-395).
IGI Global).
396
397
Chapter 2.6
The Enterprise Systems
Approach
Andrew Targowski
Western Michigan University, USA
AbstrAct sustainable enterprise development should lead to

new development of enterprise systems.
The enterprise system approach is defined by its
evolution and major milestones of architectural
planning. The ES architectures are multi-faceted IntroductIon
solutions, hence it is defined in the scope of the en-
terprise organization architecture (EOA), enterprise The purpose of this study is to define the Enter-
functional architecture (EFA), enterprise processive prise Systems approach, its evolution, and major
architecture (EPA), enterprise information architec- milestones of its architectural planning. The former
ture (EIA), enterprise software architecture (ESA), is done mostly in a graphic manner and based on
enterprise network architecture (ENA), enterprise graphic models, which should be self-explanatory.
service architecture (ESA), business component The ES architectures are multi-faceted solutions,
architecture (BCA), enterprise information in- hence they will defined in the scope of the Enter-
frastructure (EII), and enterprise configurations. prise Organization Architecture (EOA), Enterprise
A composite ES architecture is presented as a Functional Architecture (EFA), Enterprise Proces-
transitional architecture, which is currently prac- sive Architecture (EPA), Enterprise Information Ar-
ticed by most enterprises. The near future of the chitecture (EIA), Enterprise Software Architecture
ES approach will be rather limited to the ways of (ESA), Enterprise Network Architecture (ENA),
delivering ES’ applications within a framework of Enterprise Service Architecture (ESA), Business
service-oriented architecture (SOA) and the cloud Component Architecture (BCA), Enterprise Infor-
computing, which satisfies effective large-scale mation Infrastructure (EII), and Enterprise Configu-
operations. The progressive process of organization/ rations. Such enterprise architectures concerning
business virtualization and the urgent need for more hardware and data have been left undefined due to
the limits of this chapter.
DOI: 10.4018/978-1-60566-856-7.ch001
The Enterprise Systems Approach
The ES approach became necessary in the “systems administrators” who are the craftsmen
1990s when the complexity of enterprise systems of the Information Age. There are an estimated
became the major issue in systems development 7,000 such data centers in America alone. It is not
and was integrated into thousands of IT solutions. It surprising that that they are inefficient. On average,
was necessary to provide a general, well-modeled only 6% of server capacity is used and nearly the
map of IT systems and services thatcould help in 30% that are no longer in use at all, no one has
understanding the rising enterprise complexity, bothered to remove. Many data centers will be
which had to be contained and explored for the consolidated and given a big reengineering. For
sake of enterprise operations. example, Hewlett-Packard (HP) used to have 85
data centers with 19,000 IT workers worldwide.
trends of enterprise One can expect that it will be cut down to six
systems development facilities in America with just 8,000 employees
by the end of 2008 and the budget cut from 4%
The process of IT-driven enterprise formation will to 2% of revenue (The Economist, October 28,
take sharp turns in the 21st century while more 2008, p. 6).
technologies and standards will be developed
and further challenge business and IT execu- • As a result of such operations integrations,
tives. The single major question for application data centers are becoming factories for
acquisition in the past was “make or buy?” It computing services on an industrial scale.
was assumed that whether an application was Software is increasingly being delivered
“made” or “bought,” it would almost inevitably as an online service and wireless networks
be run inside the enterprise firewall. But in the are connecting to more and more devices..
21st century the IT industry is entering a period All these allow computing to be disag-
of massive innovation and growth in alternative gregated into components or “services,”
delivery models, ranging from in IT terminology. This trend leads to the
development of cloud computing, in which
• EAI through to full-blown BPI, SaaS, and information is permanently stored in serv-
SOA. So now, the costs, risks, benefits, ers on the Internet and cached temporarily
and sustainability associated with each one on clients that include desktops, entertain-
must be carefully evaluated. A 21st-centu- ment centers, table computers, notebooks,
ry enterprise systems portfolio will contain wall computers, handhelds, sensors, moni-
an eclectic mix of delivery models as the tors, etc. Cloud computing is a general
economics of delivery change and technol- concept that incorporates software as a
ogy progresses. It is extremely important service (SaaS), Web 2.0, and other technol-
that any enterprise application strategy is ogy concepts in which the common theme
fully informed about current capabilities of is reliance on the Internet for satisfying the
different delivery models and is aware of computing needs of the users. For exam-
how these may develop over time. ple, Google Apps provides common busi-
ness applications online that are accessed
Before Ford revolutionized carmaking, au- from a user’s web browser, while the soft-
tomobiles were assembled by teams of skilled ware and data are stored on the Google’s
craftsmen in custom-built workshops. Similarly, servers farms. Since cloud computing ap-
most corporate data centers today house armies of plies the Internet for data transportation, it
398
will be a questionable path for distributing operating worldwide and in which students
corporate data, which, up until now, were in many locations come together in a “vir-
transmitted through private networks. tual classroom” to form a virtual business.
They do not conduct real transactions, nor
The enormous applications of the Internet and do they sell real merchandise. The purpose
its cost and time cutting practices in delivering of the class is simply to learn how to start
services lead to the development of virtual enter- and run a business. The business is usu-
prises, which are organizations without walls but ally replaced at the end of the year for the
with a cyberspace. The virtualization of business next group of students. The students do
and the society takes very convincing forms in business with many companies, and even
the beginning of the 21st century. Hence, one can hold trade fairs and grand openings. One
predict that the following trend in the development of examples of new emerging ES are: e-
of ES will take place: commerce, collaborative communication
(MS SharePoint), and so forth. A more new
• To support the development and operations ES’ development and implementations one
of virtual enterprise (VE), which is a tem- can expect in the near future (Targowski
porary alliance of enterprises that come to- 2009a).
gether to share skills or core competencies
and resources in order to better respond to The financial crisis which touched the world
business opportunities, and whose cooper- in October 2008 should change the development
ation is supported by computer networks. of ES. Since the failure of banks and financial
It is a manifestation of Collaborative institutions has something to do with the enormous
Networks and a particular case of Virtual complexity of created enterprises, which get out
Organization (VO). VO defies the conven- of control? For example Lehman Brothers and
tional rule for operating an organization. It Wachovia Bank were not economically vital and
does so by accomplishing tasks traditional- sustainable firms in 2008 because their business
ly meant for an organization much bigger, practices did not adequately consider the long-
better resourced, and financially stable. A term consequences of their actions. The collapse
company having the technical capability of Enron in the 2000s showed the importance of
with the right human skill set and another internal policies such as maintaining adequate
with the solution may come together to controls and good corporate governance practices
create a VO. The VE (a kind of VO) does on a firm’s sustainability. Environmental steward-
not exist in the physical sense but on an ship requires consideration of how to use natural
electronic network representing a partner- resources in a productive, efficient, and profitable
ship of businesses existing as a nebulous manner. Concern for how goods and services
form of business organization that only ex- are produced, packaged, transported, used, and
ists to meet a market opportunity. Today, recycled calls for examining supply chain poli-
there are numerous virtual enterprises on cies as well as energy and environmental impacts.
the Internet. Virtual Music Enterprises is Social responsibility starts, but does not end, with
one example of a virtual enterprise Virtual the employee and a firm’s hiring, opportunity and
Enterprise California lists several other training practices. The needs of other stakeholders
examples. Another example of a Virtual (suppliers, customers, investors) and their cultural
Enterprise would be an educational class concerns, as well as the community at large and
399
even the future generations are increasingly impor- tested market space, ripping for growth. Such
tant. Hence, the expected trend in the development strategic moves, termed “value innovation,” create
of ES should lead towards the development of: powerful leaps in value for both the firm and its
buyers, rendering rivals obsolete and unleashing
• Sustainable enterprise and supporting an new demand. It leads to:
appropriate information infrastructure.
Sustainability has been defined as “meet- • Development of value-creation enterprise.
ing the needs of the present without com- Examples of such enterprises include:
promising the ability of future generations Barnes & Noble Bookstores, Cirque du
to meet their needs.” This definition calls Soleil, and Dell Computers which created
for business organizations to operate in a new market space (with a new value for
manner that values the well-being of stake- customers) that outperformed other books
holders (customers, suppliers, employ- stores, cirques, and PC makers. This new
ees and investors) in sustaining economic emerging kind of enterprise will need a
prosperity while also taking care of their new set of key enterprise systems.
employees and the environment. The core
concept of sustainability seeks to encour- Table 1 illustrates a typology of emerging new
age the adoption and implementation of kinds of enterprises, which determine appropriate
practices that aim at the “triple bottom support of enterprise systems.
line:” The ES are not only being implemented in
i. Economic vitality, industrial practice. They also are taught at the uni-
ii. Environmental stewardship, and versity level where courses on enterprise projects
iii. Social responsibility. are not only undertaken, but also are supported by
computer laboratories that introduce students to
While initially, sustainability was viewed as hands-on use of enterprise software such as SAP
primarily focusing on preserving the earth’s re- and MS Dynamics. One of the leaders in this kind
sources, its broader meaning today encompasses of education is the Western Michigan University
the achievement of economic prosperity while and its Business Information Systems Department
caring not only for the planet’s resources, but (Targowski & Tarn 2006).
also for the well-being of employees, society, and The presented solutions define the current state
culture and future generations. Each of the three of the art and provide generic rather than specific
segments of the triple-bottom line will require solutions. Perhaps such an approach will make
businesses to examine and refocus their current the presented solutions more permanent. The
policies and practices, as well as supporting the presented future trends of enterprise systems lead
information infrastructure. to more complex implementations of ES opera-
Companies have long engaged in head-to-head tions. However, more new enterprise systems one
competition in search of sustained, profitable must expect, particularly such ones, which will
growth. They have fought for competitive advan- be supporting virtual enterprise/organization,
tage, battled over market shares, and struggled sustainable enterprise, and value-creation enter-
for differentiation. Yet in today’s overcrowded prise. Solutions as SOA and cloud computing at
industries, competing head-on results in nothing first glance seem to be more effective. However,
but a bloody “Red Ocean” of rivals fighting over a whether will they be more reliable it is a big ques-
shrinking profit pool. Kim and Mauborgne (2005) tion which will be answered in coming practice
developed a strategy of Blue Ocean for uncon- of successful and failing users.
400
Table 1. A typology of emerging new kinds of enterprises
CUSTOMER KINDS OF ENTERPRISES

VISION
Survival Sustainable
(wisdom) Enterprise
New Value Value-
(satisfaction) Creation
Enterprise
Fast and Effective Virtual
Transaction Enterprise
(usefulness and con-
venience)
Low Cost Electronic
(usefulness) Enterprise
Activities Service
(convenience) Enterprise
Product Industrial
(productivity) Enterprise
CORE Production Relations Networking Innovations Sustainability
COMPETENCE
STRATEGY Red Ocean Strategy Blue Ocean Strategy Green Ocean
Strategy
This chapter first defines a concept of the en- as system engineering (Schlager 1956, Goode
terprise systems approach, its evolution, and then & Robert Machol 1957, Chestnut 1967). The
reviews major theoretical and industrial solutions system analysis applications in civil applications
being practiced by this approach. It concludes with led to the development of the system approach for
the assumption that this approach’s success will computer applications design (McDonough 1963,
depend on more defined and applied standards Kanter 1966) and system-oriented approach to the
which should smooth the flaws of information, conceptualization of social dynamics in sociology
communications, and other processes within an (Buckley 1967). Eventually, the system engineer-
enterprise and among enterprises. ing approach was applied as a method of designing
“hard” complex products. Its generalization led
to the conceptualization of the system approach
tHe emerGence oF tHe for “soft” products (Van Gigch (1974).
enterPrIse systems APProAcH In the 1970s and 1980s the supply of com-
prehensive families of mainframe computers
The enterprise systems approach has roots in the (ex. IBM 360 and 370) facilitated the strong
General System Theory, defined by Ludwig von development of computer information systems in
Bertalanffy in 1951 and in Cybernetics defined business and civic/governmental organizations. It
by Norbert Wiener in 1948. Based upon these two was the beginning stage of this kind of applica-
theories, the system analysis has been emerging tions, which eventually led to the development
for military large-scale systems, practiced at the of the “information archipelago” (McFarlan,
RAND Corporation (Hitch 1955). Its applica- McKenny 1983) of disintegrated applications,
tions in military systems were intensified during currently called “legacy systems.” In response to
the first stage of the Cold War and early defined this growing crisis, McFarlan offered a concept
401
of the “applications portfolio” (1981), a set of a Enterprise Systems expand ERP toward SCM
few key applications which should be controlled (Supply Chain Management) and CRM (Customer
carefully by IT managers, currently called Chief Relationships Management). Today, due to off-
Information Officers (CIO). This strategy was shore outsourcing of manufacturing, the impetus
widely accepted by computer executive since they towards the further development of CAD/CAM
had no challenge of more advanced applications. within ES is less energetic than used to be in the
Eventually, the Application Portfolio led past in the U.S and in Western Europe.
to the long stagnation of the system approach At the beginning 21st century, the enterprise sys-
and its vigorous application in the computer tems approach is popular and has many scholars,
information systems development in practice. practitioners, and writers occupied which improv-
Targowski contrasted this approach with his ideas ing this approach, mostly in the scope of enterprise
of the enterprise systems approach in his book, architecture (Rechting & Maier 1997, Carbone
Architecture and Planning of Enterprise-wide 2004). IBM promotes the so -called the Zachman
Information Management Systems (1990). His Framework as a classification structure often used
hypothesis was that if the enterprise systems are in Information Technology departments by the
properly planned and designed, they will limit the teams responsible for developing and document-
“information archipelago” and provide the tools ing an Enterprise Architecture. The Framework is
for the logical integration of systems within the used for organizing architectural “artifacts” in a
enterprise. A few years later, as a result of Ameri- way that takes into account both who the artifact
can enterprise dynamics stagnation, Hammer and targets (for example, business owner, and builder)
Champy called for Reengineering the Corporation and what particular issue is being addressed (for
(1993), which must lead to radical redesign of a example, data and functionality). These artifacts
company’s processes, organization, and culture may include design documents, specifications,
to achieve a quantum leap in performance. Ever and models. The Zachman Framework is one of
since, the system approach was aimed toward the the earliest Enterprise Architecture frameworks
whole enterprise. (Zachman 1987).
Targowski’s scholar message (1990) was Scott Bernard defines enterprise architecture
weak in the industrial circles, but Hammer and as:
Champy’s bestseller showed the right direction
on how to invigorate ES, at least for IT profes- The analysis and documentation of an enterprise
sionals. Its first step led towards the development in its current and future states from an integrated
of Enterprise Resources Planning (ERP), which strategy, business, and technology perspective
integrated MRP I (Material Requirements Plan- (2005:31).
ning) with MRP II (Manufacturing Resources
Planning). Hence, such software systems, such Its components are limited to: goals and initia-
as SAP, gained world-wide popularity. However, tives; products and services; data and information;
only IBM meant to apply the enterprise systems systems and applications; and networks and in-
approach holistically when it offered Computer frastructure as an Enterprise Architecture Cube.
Integrated Manufacturing (CIM), which contained Enterprise architecture as a business strategy
CAD, CAM, MIS and Office Automation Systems. was recognized by Ross, Weill, and Robertson
Unfortunately, IBM quit this kind of systems (2006) who researched more than 200 compa-
delivery and switched to the promotion of SAP, nies. They found that the successful companies
which is based on early IBM PICS and COPICS were those which operated through the enterprise
systems (without CAD/CAM). Nowadays, the architecture concept. The authors noticed that,
402
A high-level enterprise architecture creates shared The Schekkerman Model of Enterprise Ar-
understanding of how a company will operate, chitecture is shown in Figure 1. It is a model of
but the convergence of people, process, and tech- extended enterprise, which goes far beyond the
nology necessary to implement that architecture information enterprise architecture.
demands shared understanding of process and For example, a very strong user of enterprise
data at a more detailed level (2005:49). architecture concept is the U.S. Department of
Defense (DOD), which applies a concept of En-
Schekkerman (1995) is one of leading spe- terprise Architecture Framework (DODAF) in
cialists who not only recognized the importance order to (Mosto 2004):
of enterprise architecture but in order to pursue
its farther right development, founded the in- • Define a common approach for describing,
dependent Institute for Enterprise Architecture presenting, and comparing DoD enterprise
Developments (IFEAD) in the Netherlands. architectures
IFEAD publishes professional recommendations • Facilitate the use of common principles,
and provides solutions based on its definition – assumptions and terminology
‘Enterprise Architecture is about understanding The principal objective of DODAF is:

all of the different elements that go to make up the
enterprise and how those elements interrelate.” To ensure that architecture descriptions can be
In that sense, examples of elements are: strate- compared and related across organizational
gies, business drivers, principles, stakeholders, boundaries, including Joint and multi-national
units, locations, budgets, domains, functions, boundaries
activities, processes, services, products, infor-
mation, communications, applications, systems, The DoD recognizes a difference between an
infrastructure, etc. architecture and systems in the following manner:
Figure 1. The J. Schekkerman model of enterprise architecture
403
• System Architecture is like blueprints for • Users’ groups and their system needs
a building • Operational Environment (OE)
• Enterprise Architecture is like urban • Solution Logic (SL)
planning • Solution Assessment (SA)
• State-of-the-art of profession and technol-
However, as far as information enterprise ogy (SA)
architecture is concerned, the main attention in • Peaceful cooperation with the nature (N)
business practice at the beginning of the 21st cen-
tury is turning towards a system of servicing of The enterprise systems’ planning/modeling is
application software and information processing done in such a way that the resulting solutions are
on a large scale within the concept of SOA and expressive and correct. The architectural dimen-
cloud computing. A concept of extended enterprise sions of ES are shown in Figure 2. As George Klir
architecture is still in status nascendi but offers (1985, p. 27) indicates, the architectural system
very promising solutions for more advanced and planning should be described from a proper per-
ambitious firms. spective so as to recognize its overall structure
without being distracted by details.
The ES approach is defined as the art and
tHe enterPrIse systems science of solutions planning applied to fulfill
APProAcH deFIned the practical and the synergetic requirements of
information technology (IT) advanced enterprise
The enterprise systems (ES) approach has roots environments, including users and customers. The
in the system approach, which was initiated at the characteristics that distinguish the work of an sys-
RAND Think Tank of the Air Force in the 1950, tem architect from other man-made structures are:
by such pioneers as Charles Hitch (1955) and
others, who applied this approach in developing • The suitability of the solutions for use by
complex military systems. In the 1990s, when all stakeholders
information systems strategic planning became • The stability and relative permanence of
popular in business enterprises it was evident that solution construction for 15 to 25 years
the enterprise systems’ big-picture can be best • The communication of solutions’ ideas
defined under the form of the enterprise systems and assessment through the open-ended
architecture. configuration.
The enterprise systems approach is based
on the philosophy of the system approach (Klir The modeling of the ES architecture begins,
1985) and management cybernetics (Beer 1981), when the business and IT strategies are defined
graphic-architectural modeling, and was first and is based on the following principles:
defined by Targowski (1990) as a comprehensive
and cohesive solution to the problem of system 1. Cybernetization. A good ES architecture is
development, thus eliminating the fuzziness of viable and capable of growing, and adapting
the “Applications Portfolio” and the “Information to complex, dynamic enterprise operations.
Archipelago” (McFarlan 1981). This architecture should be self-organizing,
The enterprise systems architectural approach with a feedback, and requisite variety.
involves thought and graphic visualization in the 2. Systematization. A good ES architecture is
following dimensions: modeled as a set of components and inter-
relationships in order to accomplish a given
404
Figure 2. The dimensions of the enterprise systems approach
goal and produce results outside of ES, 7. Value Engineering. A good ES architecture
despite of the external obstacles. does not contain unnecessary components
3. Cohesiveness. A good ES architecture makes but only these ones that are proper for a
all components of ES sticking together and given solution.
it supports the enterprise functions and 8. Sustainability. A good ES architecture
processes in a harmonious manner. supports the reliable development and
4. Categorization. A good ES architecture operations of an enterprise in a long-term
requires that ES components are self- perspective, taking into account; economic
contained, independent of each other, and vitality, environmental stewardship, and
planned without repetition in different forms. social responsibility.
5. Primitiveness. A good ES architecture is 9. Open-ended structures. A good ES archi-
based on generic elements and relationships. tecture is open for future improvements
Further extensions of the ES’s functionality and for the insertion of components that are
should have roots in the primitive, generic perceivable though not yet available.
model of a given element.
6. Completeness. A good ES architecture is The application of these principles in the
perceives all major and possible components modeling of ES architecture determines the reli-
of solution logic (SL) and technical orienta- ability, quality, and efficiency of a solution for the
tion (TO). whole enterprise. To the extent that these general
405
principles are known, they provide limits on what social change in the growing com-
is possible and guidance on what is efficient plexity of social issues,
(Clemson 1984). ◦ Pluralism – representing interest
Functions of the ES architect. The functions groups affected by computer use with
of the ES architect are numerous and include the “fair information practices” and with
technical as well as the sociological aspects of a combination of legal, regulatory,
information management. The ES architect must and security measures to protect the
be an expert in application systems and their hard- public and users,
ware/network infrastructures. He/she must know ◦ Radical criticism – protesting the
how to arrange the many components of these ap- philosophy that computers should be
plications/infrastructures in the most economical allowed to have their own logic of in-
and convenient manner by using graphic modeling dependence, and that mega computer
tools. Such applied sciences are used as system systems should be developed to op-
science, decision science, management, econom- erate automatically without human
ics, information engineering, etc.The ES architect control,
must know how to draft, render and finish in a ◦ De-evolutionism – gaining power of
very good visual manner the expected solution. design and loosing control over use,
The ES architect as a social engineer has a ◦ Computer surveillance – producing
choice of using: organizational benefits, since em-
ployees “better” work.
• Tools of information technology, such as
system analysis and design, programming Because these choices of ES factors, the archi-
high level languages, databases, knowl- tect should have of social awareness about what is
edge bases, utility software for rapid pro- he/she is going to develop. Social awareness can
totyping solutions, be composed of social biases, beliefs, and expecta-
• Stimulations defined in information poli- tions. Social biases are influenced by the choice of
cies on security, privacy, trade, computer IT tools. For example, a technical bias advocates
crime, intellectual property, ethics, and so that technology can solve a social problem, while
forth, a non-technical bias supports the view that such a
• Positions that will influence ES modeling problem can be solved through managerial action
(Mowshowitz 1980): such as leadership or an improved market strategy.
◦ Technicism – using IT as an instru- The choice of stimulations can influence social
ment of progress, where success or beliefs. For example, computers can be seen as
failure depends on the system design strength of the economy or as a threat to an indi-
and implementation; social and po- vidual privacy and autonomy. Social expectations
litical consequences are ignore, are the result of a position taken by the ES architect
◦ Progressive individualism – human- and are determined by the information culture in
izing the system with computers as the scope of values (human and civil rights ver-
helpers to achieve desirable change, sus totalitarian slavery, creativity, and electronic
◦ Elitism – informing and rescuing so- friendship versus alienation), symbols (credits
ciety as the mission of the computer cards equate to a cashless society, computer screens
specialist becomes a mission of a equate to a paperless society), competence stan-
social engineer, which is the most dards (a lack of computer competence equates to
sophisticated method of steering the illiteracy), knowledge centers (data, knowledge,
406
Figure 3. The ES architect as a social engineer

tHe evolutIon oF tHe
enterPrIse systems APProAcH
In the early period (the early 1960s) of computer

information systems (CIS) development, the para-
digm of design was application programming. As
shown in Figure 4, the IS concept came as a result
of preparing an operational environment for the
application of the IBM 1400 and IBM 360 tape/
disk computers. The operational environment was
organized as a bureaucracy specialized in routines;
consequently, this was the logic orientation (SL)
of design. The routines were programmed in
COBOL, the state of the art (SR) language and
processed in batches as a technical orientation
(TO). The system assessment (SA) was based on
the cost-benefits analysis, which was aiming at
the replacement of office clerks by computerized
routines. The leading specialist in this time was a
programmer, a master handler of computers, who
wisdom bases), know-how (individual computer applied the MIND III (practical) (Skolimowski
skills, social skills that control the information 1984).
transformations), and futurology (“Star Wars”). In the 1970s with the development of database
The Figure 3 illustrates the ES architect’s social technology, the system design switched from
awareness about modeled solutions. sequential processing on magnetic tape storage
to on-line processing transactions on magnetic
disks storage. The operational environment (OE)
was slowly transforming into a technocracy which
was emphasizing project management.
Figure 4. Pre-architecture stage of is development in the 1960s
407
Solution logic (SL) was oriented towards In the 1990s the first ES architectures are ap-
subsystems using databases; among the most pearing due to the integration of MRP I and MRP
popular were such ones as customer orders, master II into the ERP system complex on the business
schedules, inventory control, and bill of material side of an enterprise. On the technical side of the
processing (BOMP), cost control, and other. The enterprise, CAD and CAM evolved into a CIM
dominant professional was a system designer. The solution (SL). The operational environment (OE)
technical orientation (TO) was on-line process- embraced the whole enterprise and strategic role
ing via remote terminals accessing databases in of IT become a very important factor in a search
real time. The state of the art (SR) was database for business competitive advantage (SA) by Chief
technology and systems assessment was still a Information Officers (CIO). The wide proliferation
cost-benefit analysis. Figure 5 shows a set of of enterprise systems required good GUIs to make
key factors defining systems development in the systems easier to operate by the end-users (SR).
1970s. A system designer applied the MIND III Furthermore, stand-alone PCs were organized into
(practical). LANs, and building networks were connected
In the 1980s the creeping revolution of into MANs and eventually in WANs and GANs,
personal computers (TO) involved in the Com- creating dynamic private computer network grids
puter Age all workers of an enterprise, particu- (TO). Every 18 months personal computers have
larly business professionals, who soon learned been doubled their computing speed, so millions
spreadsheets and word editors (OE). Solution of them had to be recycled as they were replaced
logic (SL) put emphasis on sure applications, by faster units (N). Figure 7 reflects the situation
which were organized in a application portfo- in ES in the 1990s. The MIND I (theoretical) and
lio. The complexity of applications grown and MIND II (creative) as well as MIND III (practi-
required that more attention should be paid on cal) are engaged in the process of developing the
the system life cycle development and end-user ES architectures.
involvement in applications implementation In the 2000s the trend in developing ES takes
(“Joint Application Development”). Hence, a a sharp turn. The public Internet’s web technology
system developer became a key specialist in passes to enterprise networks, creating Intranets
this process (Figure 6). and Extranets (TO).
Figure 5. Pre-architecture stage of IS development in the 1970s
408
Figure 6. Pre-architecture stage of IS development in the 1980s
Figure 7. The architecture of enterprise systems

enterprise due to such system as SCM and CIM,
in the 1990s
supported by the Extranet within an emerging
Enterprise Information Infrastructure (EII) (SL).
The same enterprise is even more expanding,
since the Internet allows for instant communi-
cation and manufacturing and IT projects are
outsourced to offshore subcontractors, mostly in
Asia (OP). Since computer programming is done
in India, the local IT professionals transform into
business process management (BPM) specialists,
who conceptualize how to speed up the flow of
info-materials and money through the dispersed,
electronic enterprise (SR). The offshore outsourc-
ing trend is led by business executives who look
for cost minimization and fat bonuses given to
them by stockholders, taking full advantage of
electronized globalization. Figure 8 explains the
challenges for IT architects in the 2000s.
The further evolution of the ES architecture
probably will put emphasis on how to operate a
very complex Enterprise Information Infrastruc-
ture in the global environment in a reliable and
secure manner.
This leads to the fast emergence of efficient

Global Economy, where “the distance is dead.”
A closed enterprise transforms into an extended
409
Figure 8. The architecture of enterprise systems

• World Dimension with 4 views (provided
in the 2000s
in Figure 9)
• Business Dimension with 13 views
• Behavior Dimension with 6 views
To identify the enterprise organization com-

plexity let’s assume that the number of elements
(e), relationships (r), and behavior states (s)
measures system complexity. The number of
relationships is computed based on the formula:
r = (e –1)e: 2
The minimal number of behavior states (On
and OFF) can be computed by the formula:
s = 2e
The number of elements, relationships and
states characterizes systems shown in Figure 10.
Let’s apply measures of system complexity to
a model of an enterprise system shown in Figure
8. There are two possible approaches to this issue.
In the Three-Tier approach, called “complete,”
system complexity is provided in Table 2. In the
Two-Tier approach, called “holistic,” system
complexity is provided in Table 3. It was assumed
that a view is composed of e=7, r=21, s=128 for
all 23 views.
tHe enterPrIse orGAnIZAtIon
In “complete” management of a business
ArcHItecture
enterprise, executives, responsible for the so
called “big-picture” decisions, must take into
Organizers, managers, and workers of a business
account: 27 entities (23 views, 3 dimensions, 1
organization system deal with enormous system
inter-dimension), 187 elements, 585 relationships,
complexity. Why? Because, a business organi-
and 11,224 system states in a three-tier hierarchy.
zation system is a social system, which in the
This is a huge number of “things” to care for.
system hierarchy is located on the highest level
Therefore, “big-picture” is divided into a series
available for mankind. A business organization
of “small pictures” that are associated with senior,
system complexity is illustrated in Figure 9. Any
middle, and lower levels of management. In addi-
business organization system in the contemporary
tion, automation of information systems takes care
economy of developed nations, such as an enter-
of the majority of well-structured relationships
prise is multi-dimensional and multi-viewing. To
and elements’ states. In such a manner system
be successful in the global economy a firm must
complexity is decomposed and “processable” by
apply the holistic rule for business development
individual managers.
and operations. One can recognize the following
In “holistic” management of a business enter-
dimensions and views of an enterprise:
prise system, which has emerged as a new mode
of management in the 1990s, relationships are
410
Figure 9. A model of enterprise dimensions (D) and views (V)
Figure 10.
411
Table 2. Complexity of a business enterprise system – “complete” (three-tier)
System Level Elements Relationships States Entities Elements Relationships States

Number
Intra-view 7 21 128 23 161 483 2944
Business 13 78 8192 1 13 78 8192
Dimension
Behavior 6 15 64 1 6 15 64
Dimension
World 4 6 16 1 4 6 16
Dimension
Inter- 3 3 8 1 3 3 8
Dimensional
Total 27 187 585 11224
Table 3. Complexity of a business enterprise system – “holistic” (two-tier)
System Elements Relationships States Entities Elements Relationships States

Level Number
Intra-view 7 21 128 23 161 483 2944
Inter-view 23 2553 8388608 1 23 2553 8388608
Total 24 184 3036 8391552
taking place in a two-tier hierarchy. At the first tHe enterPrIse FunctIonAl

level, relationships are established for each view. ArcHItecture
At the second level, relationships are established
among all views of the three system dimensions. The first task of the ES architect is to define the
“Big-picture” executives must be involved in Enterprise Functional Architecture (EFA), which
184 system elements, 3036 relationships, a nd identifies major enterprise business, technical,
8,391,552 system elements’ states. and operational functions, according the manage-
The complexity of “holistic” management is ment chart as it is shown in Figure 11. Similar
much bigger than the complexity of “complete” approach is the Value Chain (Figure 12), which
management. The former is a case of the global identifies primary and secondary activities (de
economy challenge that requires executives to facto functions) of the enterprise. Both architec-
include more system elements, relationships, and tures provide a clue of what kind of information
states. The need for automation of global informa- systems are required to support operations of a
tion systems is acute. To reduce the complexity of given enterprise.
the global economy new institutions are emerg-
ing, for example the World Trade Organization
and regional treaties such as NAFTA and the tHe enterPrIse ProcessIve
European Union. ArcHItecture
The ES architect must minimize the enterprise
complexity in order to manage such a multifaceted In their book Reengineering the Corporation
enterprise. (1993), Michael Hammer and co-author James
Champy define a reengineering goal as “achieving
412
Figure 11. The enterprise functional architecture
dramatic improvements in critical contemporary 2. Analyze existing processes – reconfigurate

measures of performance such as cost, quality, your work for smoother workflow;
service and speed.” The basic steps of reengi- 3. Invent new ways to work – reconstruct your
neering are: jobs to match reality;
4. Implement new processes – for ongoing
1. Define business objectives – reassess your competitiveness.
business purpose and reposition for greater
market penetration;
Figure 12. The Value Chain – The Enterprise Activities Architecture
413
Hammer and Champy say that ability to use every thing is right, the computer sends a
insight, imagination, and a willingness to challenge payment to a supplier, even without its in-
all assumptions are key to Business Process Re- voice. About 500 workers were moved to
engineering (BPR). In practice, BPR lets people other jobs, saving on cost.
change not only procedures but their rules. In • IBM Credit, the organization that finances
effect, they may change not only rules of a given computers, software, and service sold by
process but of the entire business. IBM, learned that the actual work required
Reengineering in the 1990s inspired executives to process a new customer could be com-
and managers in thousands of companies to start pleted in only 90 minutes, instead of the
rethinking and redesigning such basic business six days to two weeks it used to take.
practices as customer service, order fulfillment,
product development, etc. For example: The main premise of BPR is based on the
reorientation of business procedures from
• Ford Motor Company found that it em- function-driven to process-driven. Whereas the
ployed 100 times more people in its ac- former has roots in the 19th century bureaucracy,
counts payable department than smaller when each department had its own internal flow
Mazda. After reengineering the process, of information, the latter is based on the cross-
the company cut through the territory functional flow of information, which supports a
of accountants, purchasing department given process, for instance order fulfillment. The
staff, warehouse receiving clerks, and re- BPR Model in Figure 13 illustrates this change
ordered them. Now, a receiving clerk at in business practice.
Ford checks the database for the order and Reengineering approaches defined by Currid
delivery compatibility and quality and if (1995) are:
Figure 13. Functions versus Processes
414
• Streamlining business processes; compatibility, while BPI applies business rules

• Integrating business processes; to operate integrated applications in order to run
• Transforming business processes. a business more efficiently. The BPI by Oracle
Enterprise Software is shown in Figure 14.
The seven reengineering business principles
defined by Currid (1995) are:
tHe enterPrIse InFormAtIon
1. Organize work around results, not tasks; ArcHItecture
2. Capture data only one time-when it is first
created; Once the Enterprise Functional Architecture is
3. Allow decision points where work is modeled, the next step is the modeling of the
performed; Enterprise Information Architecture (EIA) shown
4. Incorporate control into information in Figure 15. It should contain the main, already
processing; developed, information systems and their relation-
5. Make people who use a process do the work; ships, and also those, which should be develop
6. Work in parallel instead of sequentially; in the future.
7. Treat geographically dispersed resources as Among the Business Information Systems
one. (BIS) one can recognize: systems for marketing
& sales (CRM), finance, accounting, legal, man-
The BPR projects lead to Business Process agement and other business functions. Product
Integration (BPI) via Enterprise Application Inte- information Systems (PIS) contain such ones as
gration (EAI). However, EAI typically involves the CADD (Computer-aided Drafting and Design),
exchange of information between two applications
without regard for business process. BPI, on the Figure 15. The enterprise information architec-
other hand, takes into account the overall work ture
flow and the multiple applications required to
complete a business process. EIA ensures software
Figure 14. The Oracle Enterprise Suite Addressing

the Task of BPI by Extracting Business Processes
from Applications
415
CAPP (Computer-aide Process Planning), and has 500,000 lines of code and 80,000 database
other ones supporting engineering. Among Op- tables, for 22+ application modules and 800
eration Information Systems one can include; business processes, becoming one of the most
CAM (Computer-aided Manufacturing), CAP complex software in the market, more complex
(Computer-aided Publishing), and other. The than MS Windows.
Inter-organization Systems may be identified by Figure 18 describes the Oracle Enterprise Suite
such ones as; SCM (Supply Chain Management) Software Architecture, which is less complex than
and e-Commerce (B2B, B2C). SAP and more intuitive, design according to the
Based on the EIA the next task will be to se- business cycle.
lect the right enterprise software and develop the Figure 19 depicts the MS Dynamics Enterprise
Enterprise Software Architecture. Software, which is designed mostly for small and
medium size companies in a very “compact” manner.
Microsoft Dynamics enterprise software sup-
tHe enterPrIse soFtWAre ports three most critical functional areas: Financial
ArcHItecture Management; Customer Relationship Manage-
ment and Supply Chain Management, indicating
The Enterprise Software Architecture (ESA) is the trend of integrating business functions/pro-
well described in Figure 16, where the SAP soft- cesses into self-contained system federations. The
ware collaborates with no-SAP software. generic and vendor-independent ESA is defined
The more detailed SAP Software Architecture in Figure 20, where shadow areas identify web-
is shown in Figure 17, where 22 modules/systems technology driven most popular applications in
are presented. Needless to say that this software 2006 (Targowski 2003).
Figure 16. The general enterprise SAP R/3 software architecture (the Targowski model)
416
Figure 17. The enterprise SAP R/3 software modules architecture
Figure 18. The main applications of the Oracle Enterprise suite
tHe enterPrIse netWork • Private computer grid of networks (Figure

ArcHItecture 21)
• Public computer set of networks based on
The Enterprise Network Architecture (ENA) web technology, where Intranets are pri-
can be composed of two kinds of interconnected vate networks (Figure 22)
networks:
417
Figure 19. The enterprise ms dynamics software architecture
Figure 20. The enterprise popular software architecture - 2006
418
Figure 21. The enterprise grip of private computer networks
Figure 22. The enterprise public computer networks
tHe enterPrIse servIce- tried to address these concerns. The introduction

orIented ArcHItecture (soA) of the Service-oriented Architecture (SOA) in
Figure 23, along with associated technologies
Integration has always been high on the list such as the emergence of the Enterprise Service
of concerns for IT professionals, and over the Bus (ESB) and Business Process Management
years there have been a number of models and (BPM) solutions has helped to create more open
technologies (e.g., DCOM, ORB, COBRA) that information infrastructures, which can be pen-
419
Figure 23. The enterprise services-oriented architecture (SOA)
etrated by different service-oriented software are designed as a single whole (e.g., SCM, ERP,
modules, particularly web-driven ones. In the CRM), BCA consists of a coalition of Business
past, the integration efforts represented data and Components that communicate either via events
information, now they add processes. (EDA) or via request/reply calls (SOA). The BCA
In order to apply the SOA, applications need offers the following benefits:
to be “deconstructed” into their component ser-
vices – each of which performs a specific business 1) BC are easier to develop since are smaller
function, such as address lookup, customer credit and less complex than whole applications,
check, and so on (Figure 23). Within the SOA 2) BC are reusable,
such services can be assembled and delivered 3) BC are easily deployable and modifiable,
by specialized services throughout the Internet, 4). BC contributes to easier software con-
according to required standards. figuration management.
The Business Component Architecture (BCA) The way of using the web-driven services is
refers to the development of enterprise systems’ described in Figure 25. In this mode, it is necessary
service as a collection of reusable business com- to apply the Web Service Description Language
ponents/services that interact via request/reply (WSDL). The following steps are involved in
(SOA) only. The BCA in some projects includes providing and consuming a service:
the Event-driven Architecture (EDA), which
continue to exchange between different service- 1. A service provider describes its service using
oriented software components (Figure 24). In WSDL, which is inserted into the Service
contrast to traditional, monolithic applications that Directory, which is accessible through the
420
Figure 24. The business component architecture (BCA)
UDDI (Universal Description, Discovery, guages. Since 1998, the Java Message Service
and Integration) registry for the retrieval of (JMS) has emerged as the dominant industry
services provided by various vendors. standard for enterprise communication, imple-
2. A service consumer queries the Service mented by thousands of companies. Many other
Directory to locate a service provider and find communication standards have been developed
out how to communicate with that service. and the Enterprise Service Bus (ESB) emerged
3. Part of the WSDL provided by the service as a channel for communicating all standardized
provider is passed to the service consumer, messages (Figure 26).
informing about a system of request and The ESB supports development in multiple
responses applied by the vendor. programming languages, becoming a multi-plat-
4. The service consumer uses the WSDL to form enterprise backbone, perhaps the Enterprise
send a request to the service provider. Nervous System (ENS) facilitating; communica-
5. The service provider provides the expected tions, connectivity, transformation, portability
response to service consumer. and security.
Several companies developed Enterprise Ap-

plication Integration (EAI) solutions during the
mid 1990s, but these solutions called middleware
were proprietary. Most of these solutions were
developed in C/C++ and other proprietary lan-
421
Figure 25. The system of using Web-available services (SOAP – SOA Protocol)
Figure 26. The Enterprise Service Bus (ESB) Architecture
422
tHe enterPrIse InFormAtIon • Integrated enterprise, which applies a

InFrAstructure common,w enterprise database for the ma-
jority of applications,
The information management discipline is very • Agile enterprise,
young; it is at most 50 years old. However, its • Informated enterprise, which applies
professional development has been very strong knowledge management systems in
especially in the last two decades of the 20th decision-making,
century. Since the beginning, IS have been de- • Communicated enterprise,
veloped as islands of automation that have been a • Mobile enterprise,
consequence of an autonomous and eclectic (non- • Electronic entreprise,
architectural) approach. In the 1980s personal • Virtual enterprise, which applies commu-
computers triggered a creeping, quiet revolution, nication technology in connecting different
known as the Computer Age. In the 1990s, the worker locations, when the workers very
Internet launched the Communication Age and the often operate from their own home, a hotel,
phenomena of telecommunication networking and a car, or a customer’s location1.
video services. We no longer deal with IS only; we
now have to include into an information architec-
ture other forms of information-communication Figure 27. The Enterprise Information Infrastruc-
systems and services. ture Architecture
These new systems and services require a
new approach towards IT applications in the
enterprise. A set of these systems and services
is what can be called the Enterprise Information
Infrastructure (EII) as it is depicted in Figure 27
(Targowski 2004).
The EII can be considered as an info-com-
munication construction of an enterprise, whose
bricks, floors, and walls are made of electronic
elements. For many, these elements are invisible
even wireless, but their cost is in millions of dollars.
tHe enterPrIse
conFIGurAtIons
According to the IT criterion, one can distinguish

the following configurations of enterprise (Tar-
gowski 2003):
• Off-line enterprise, which data processing

operates in a batch mode, not on-line and
not in real-time,
• On-line enterprise, which processes infor-
mation on-line through computer networks,
423
• Sustainable enterprise, which is economi- with CIOs and triggered the outsourcing the IT
cally vital, environmentally friendly and projects, first inshore and later offshore (looking
socially responsible. for cost advantage). Eventually, due to the latter,
the quality of IT solutions could be in question
The above sequence of the enterprise evolu- and the reliability of solution may be lower.
tion is a simplified model, as is each model. In The nowadays IT-driven enterprise can look
practice, some components of these enterprises as it is shown in Figure 28.
can appear in just one enterprise and very often in
a sequence different than the evolutionary model.
However, very often such enterprises, which have conclusIon
only a piece of each configuration are considered
unfinished solutions with questionable benefits The future of the enterprise systems approach will
of information technology. The purpose of the be more complex as IT is still considered as the main
enterprise evolution is to know its requirements factor in achieving competitive advantage in the
for each configuration’s architecture, budget, Global Economy (Targowski 2009b). Only in the
developmental skills, and the timeline. U.S. are expenditures for information management
So far the development of IT-driven enterprise 15% of GNP, in the range of $ 2 billion (2006).
has been incremental, leading to a mixture of all This enormous amount of financial means will
the presented enterprise configurations with over support the development of IT-driven enterprises.
the budget projects, overwhelming IT workers, Their success will depend on more defined and ap-
who could not cope with such increasing com- plied standards, which should smooth the flows of
plexities. As a result of it, CEOs disconnected information, communications, and other processes.
Figure 28. The multi-dimensional, composite enterprise architecture in the 2000s
424
reFerences Hammer, M., & Champy, J. (1993). Reengineer-

ing the corporation. New York: HarperBusiness.
Ackoff, R. L. (1999). Re-creating the corporation.
New York: Oxford University Press. Hitch, Ch. (1955). An application of system
analysis. In St. L. Opter (Ed.), System analysis.
Barbour, I. (1993). Ethics in age of technology. Middlesex, England.
San Francisco: Harper.
Kanter, J. (1966). Integrated management and
Beer, S. (1981). The brain of firm. Chichester, control systems. In The computer and executive
UK: John Wiley & Sons. (pp. 45-56). Englewood Cliffs, NJ: Prentice Hall.
Bernard, S. A. (2005). An introduction to enterprise Kim, W. Ch., & Mauborgne, R. (2005). Blue ocean
architecture. Bloomington, IN: AuthorHouse. strategy. Boston: Harvard Business School Press.
Bernus, P. (2003). Handbook on enterprise archi- Klir, G. J. (1985). Architect of problem solving.
tecture. New York: Springer. New York: Plenum Press.
Bertalanfyy, L. v. (1951). General system theory - a Krafzig, D., et al. (2004). Enterprise SOA. Upper
new approach to unity of science (Symposium). Saddle River, NJ: Prentice Hall.
Human Biology, 23, 303–361.
Laszlo, E. (1972). Introduction to systems phi-
Buckley, W. (1967). Sociology and modern sys- losophy. New York: Harper Torchbooks.
tems theory. Englewood Cliffs, NJ: Prentice Hall.
Marin, J. (1997). Systems engineering guidebook.
Carbone, J. A. (2004). IT architecture toolkit. Boca Raton, FL: CRC Press LLC.
Upper Saddle River: Prentice Hall PTR.
McDonough, A. M. (1963). Information manage-
Chestnut, H. (1967). Systems engineering meth- ment. In Information economics and management
ods. New York: Wiley. systems. New York: McGraw-Hill.
Clemson, B. (1984). Cybernetics: A new manage- McFarlan, F. W. (1981). Portfolio approach to
ment tool. Tumbridge Wells, UK: Abacus Howe. information systems. Harvard Business Review,
Currid, Ch. (1995). Strategies for reengineer- 59, 142–150.
ing your organization. Roseville, CA: Prima McFarlan, F. W., & McKenney, J. L. (1983). The
Lifestyles. information archipelago governing the new world.
Erl, Th. (2004). Service-oriented architecture. Harvard Business Review, 61, 145–156.
Upper Saddle River, NJ: Pearson Professional McGovern, J. (2003). The practical guide to enter-
Edition. prise architecture planning. Hoboken, NJ: Wiley.
Fowler, M. (2002). Patterns of enterprise ap- Microsoft. (2002). Application architecture for
plication architecture. Boston: Addison-Wesley .NET.
Professional.
Mosto, A. (2004). DoD architecture framework
Goode, H. H., & Machol, R. E. (1957). System overview. Retrieved November 2, 2008, from
engineering: An introduction to the design of http://www.enterprise-architecture.info/Images/
large-scale systems. New York: McGraw-Hill. Defence%20C4ISR/DODAF.ppt
425
Mowshowitz, A. (1980). On an approach Targowski, A. (1990). The architecture and plan-

to the study of social issues in computing. ning of enterprise-wide information management
Communications of the ACM, 24(3), 146. systems. Hershey, PA: Idea Group Publishing.
doi:10.1145/358568.358592
Targowski, A. (1996). Global information infra-
Mowshowitz, A. (1980). Human choice and com- structure. Hershey PA: Idea Group Publishing.
puters, 2. Amsterdam: North-Holland.
Targowski, A. (2003). Electronic enterprise: Strat-
Nadler, D. A., et al. (1992). Organizational archi- egy and architecture. Hershey, PA: IRM Press.
tecture. San Francisco: Jossey-Bass.
Targowski, A. (2009a). The evolution from data
Perks, C., & Beveridge, T. (2002). Guide to IT to wisdom in decision-making at the level of real
architecture. New York: Springer. and virtual networks. In C. Camison et al. (Eds.),
Connectivity and knowledge management in
Porter, M. (1990). The competitive advantage of
virtual organizations. Hershey, PA: Information
actions. New York: Free Press.
Science Reference.
Rechtin, E., & Maier, M. (1997). The art of system
Targowski, A. (2009b). Information technology
architecting. Boca Raton, FL: CRC Press LLC.
and societal development. Hershey, PA: Informa-
Ross, J. W., Weill, P., & Robertson, D. (2006). tion Science Reference.
Enterprise architecture as strategy: Creating a
Targowski, A., & Rienzo, T. (2004). Enterprise
foundation for business execution. Boston: Har-
information infrastructure. Kalamazoo, MI:
vard Business Press.
Paradox Associates.
Schekkerman, J. (2008). Enterprise architecture
Targowski, A., & Tarn, M. (2006). Enterprise
good practices guide. Victoria, Canada: Trafford
systems education in the 21st century. Hershey,
Publishing.
PA: IGI Global.
Schlager, J. (1956). Systems engineering: Key to
Van Gigch, J. P. (1974). Applied general systems
modern development. IRE Transactions, EM-3,
theory. New York: Harper & Row.
64–66. doi:10.1109/IRET-EM.1956.5007383
Wiener, N. (1948). Cybernetics: Or the control and
Shnaidt, P. (1992). Enterprise-wide networking.
communication in the animal and the machine.
Carmel, IN: SAMS.
Cambridge, MA: MIT Press.
Skolimowski, H. (1981). Eco-philosophy. design-
Zachman, J. A. (1987). A framework for informa-
ing new tactics for living. Salem, NH: Marion
tion systems architecture. IBM Systems Journal,
Boyers.
26(3).
Skolimowski, H. (1984). The theater of mind.
Wheaton, IL: The Theosophical Publishing House.
Spewak, St. H. (1993). Enterprise architecture endnote
planning. Hoboken, NJ: Wiley.
1
A virtual enterprise emerged historically
before an electronic enterprise, however,
the former performs better if is electronic.
tems, edited by Maria Manuela Cruz-Cunha, pp. 1-30, copyright 2010 by Information Science Reference (an imprint of IGI
Global).
426
427
Chapter 2.7
Designing Data Marts from XML
and Relational Data Sources
Yasser Hachaichi
Mir@cl Laboratory, Faculté des Sciences Economiques et de Gestion, Tunisia
Jamel Feki
Hanene Ben-Abdallah
AbstrAct structures, the originality of our method lies in its

being decision maker centered, its automatic extrac-
Due to the international economic competition, tion of loadable data mart schemas and its genericity.
enterprises are ever looking for efficient methods
to build data marts/warehouses to analyze the large
data volume in their decision making process. On IntroductIon
the other hand, even though the relational data
model is the most commonly used model, any data Faced with the ever increasing economic competi-
mart/warehouse construction method must now tion, today’s enterprises are hard-pressed to rely on
deal with other data types and in particular XML decision support systems (DSS) to assist them in the
documents which represent the dominant type of analysis of very large data volumes. As a response
data exchanged between partners and retrieved from to this constraint, data warehousing technologies
the Web. This chapter presents a data mart design have been proposed as a means to extract pertinent
method that starts from both a relational database data from information systems and present it as his-
source and XML documents compliant to a given torical snapshots used for ad hoc analytical queries
DTD. Besides considering these two types of data and scheduled reporting. Indeed, a data warehouse
(DW) is organized in such a way that relevant data
is clustered together for an easy access. In addition,
DOI: 10.4018/978-1-60566-756-0.ch004
Designing Data Marts from XML and Relational Data Sources
a DW can be used as a source for building data precisely, it presents a DM design method that
marts (DM) that are oriented to specific subjects starts from both a relational database source and
of analyses. XML documents compliant to a given DTD.
Traditionally, the data loaded into a DW/ Besides considering these two types of data struc-
DM is mainly issued from the enterprise’s own tures, our method has three additional advantages.
operational information system. Thus, most cur- First, it provides for a DSS development centered
rently proposed DW/DM construction approaches on decision makers: it assists them in defining
suppose a single, often relational data source; cf., their analytical needs by proposing all analytical
(List, Bruckner, Machacze, & Schiefer, 2002), subjects that could be automatically extracted
(Golfarelli, Maio, & Rizz, 1998), (Cabibbo, L., from their data sources; the automatic extraction
& Torlone, R. 1998), (Moody , & Kortnik, 2000), of DM schemas distinguishes our method from
(Prat, Akoka , & Comyn-Wattiau, 2006), (Zribi, currently proposed ones. Secondly, it guarantees
& feki, 2007), (Golfarelli, Rizzi, & Vrdoljak, that the extracted subjects are loadable fromthe
2001), (Vrdoljak, Banek, & Rizzi, 2003), (Jensen, enterprise information system and/or the external
Møller, & Pedersen, 2001). However, due to the data sources. The third advantage of our design
international competition, enterprises are increas- method is its genericity: It is domain independent
ingly forced to enrich their own data repository since it relies on the structural properties of the
with data coming from external sources. Besides data sources independently of their semantics.
data received from partners, the web constitutes It automatically applies a set of rules to extract,
the main external data source for all enterprises. from the relational database and XML docu-
For instance, an enterprise may need to retrieve ments, all possible facts with their dimensions
from the web data about the exchange rates in and hierarchies.
order to analyze the variation of the quantities To achieve these advantages, our method oper-
of its sold products with respect to the exchange ates in four steps. First, it structurally homogenizes
rates during a period of time. the two types of data sources by converting the
To deal with such an open data source, a DW/ DTD into a relational model. Secondly, it classifies
DM construction approach must, hence, overcome the set of relations issued from both the converted
the main difficulty behind the use of multiple DTD and the repository of the source relational
data sources: the structural and semantic hetero- DBMS. This classification is then used to identify
geneities of the sources. In fact, even though the automatically the facts, measures, dimensions
relational data model is the most commonly used and their attributes organized into hierarchies;
model (Wikipedia encyclopedia, 2008), a DW these identified multidimensional elements are
construction approach must now deal with other modeled as star DM schemas. Finally, the result-
data types and in particular XML documents which ing DM schemas can be manually adapted by
represent the dominant data type on the web. On the decision makers/designers to specify their
the other hand, the semantic data heterogeneity particular analytical needs. The automatic steps
comes into play when the internal and external of our design method allowed us to incorporate
data sources are complementary, e.g., the case of it into a CASE toolset that interactively provides
transactional data between partners. This type of for DM schema adaptation.
heterogeneity remains a challenging problem that The remainder of this chapter is organized as
can be treated either at the data source level or follows. First we overview current DW design
the DW/DM level (Boufares, & Hamdoun, 2005). approaches for relational and XML data sources.
This chapter deals with the structural data Then, we illustrate our four-step DM design
heterogeneity when designing a data mart. More method through a relational data source and a
428
set of XML documents extracted from the web. e/r diagram-based

Finally, we summarize the presented work and design Approaches
outline our ongoing research efforts.
(Golfarelli, Maio, & Rizz, 1998) propose a semi-
automated method to carry out a DM conceptual
current dAtA WAreHouse modeling from an E/R diagram. Their method
desIGn APProAcHes produces a fact schema by: 1) defining facts from
relationships or from frequently updated entities;
Data-driven (or bottom-up) development ap- 2) building an attribute tree for each fact, useful to
proaches rely on the analysis of the corporate data construct a fact schema; 3) pruning and grafting
model and relevant transactions (List, Bruckner, the attribute tree in order to eliminate unnecessary
Machacze, & Schiefer, 2002), cf., (Golfarelli, levels of detail; 4) defining dimensions; 5) defin-
Maio, & Rizz, 1998), (Cabibbo, L., & Torlone, R. ing measures; and 6) defining hierarchies. The
1998), (Moody , & Kortnik, 2000), (Prat, Akoka , third, fourth and fifth steps are supported by the
& Comyn-Wattiau, 2006), (Zribi, & feki, 2007), preliminary workload declared by decisional users.
(Golfarelli, Rizzi, & Vrdoljak, 2001), (Vrdoljak, Also starting from a given E/R diagram,
Banek, & Rizzi, 2003), (Jensen, Møller, & Ped- (Cabibbo, & Torlone, 1998) propose a method to
ersen, 2001). Data-driven approaches were justi- build a multidimensional database in four steps:
fied by Bill Inmon (Inmon, 1996) by the fact that 1) identifying facts, dimensions and measures
unlike transactional systems whose development through a thorough, manual analysis of the given
lifecycle is requirements-driven, decision support E/R diagram; 2) restructuring the E/R diagram to
systems (DSS) have data-driven development describe facts and dimensions in a more explicit
lifecycle. In addition, Inmon argues that require- way. The produced version of the E/R diagram can
ments are the last thing to be considered in a DSS be directly mapped onto the source data model;
development; they are understood after the data 3) deriving a dimensional graph that succinctly
warehouse is populated with data and query results represents facts and dimensions from the restruc-
are analyzed by the decision makers. Hence, data- tured E/R diagram; and finally, 4) translating the
driven approaches enjoy a double advantage: they dimensional graph into the multidimensional
reduce the task of decision makers by proposing model. In this method, the first step which is the
potential analytical needs and they guarantee that most crucial is manually done. In this method, the
the enterprise’s information system can feed the designer must have domain expertise to correctly
selected needs with pertinent data. identify all potentially needed multidimensional
Considering the advantages of data-driven elements.
approaches, we elected to propose a data mart (Moody, & Kortnik, 2000) propose a three-
design method within this category. Hence, we step method to design a DM/DW also from an
next limit ourselves to over viewing works perti- E/R data model: 1) entity classification which
nent to data-driven approaches where the majority classifies entities into three categories: transaction
starts from conceptual schemas modelled through entity (describes an event that happens at a point
E/R, XML models, or UML cf. (Prat, Akoka, & in time), component entity (directly related to a
Comyn-Wattiau, 2006) and (Zribi, & feki, 2007). transaction entity via a one-to-many relationship)
We focus on those works that start from E/R and and classification entity (related to component
XML models since they are more pertinent to entities by a chain of one-to-many relationships)
our method. 2) hierarchy identification using to-many rela-
tionship and 3) data mart schema development
429
for each transaction entity. Each schema is a star Taniar, 2005) also propose a generic method for
schema represented through a fact table and a building an XML DW for XML documents. Their
number of dimension and sub-dimension tables. method first applies a set of cleaning and integra-
Separates star schemas can be combined to form tion operations in order to minimize the number
constellations or galaxies. of occurrences of dirty data, XML structural er-
rors, duplications or inconsistencies. Secondly, it
xml-based design Approaches summarizes data in XML documents to extract
only useful and valuable information in order to
(Golfarelli, Rizzi, & Vrdoljak, 2001) propose a create other XML document(s) used for the con-
method for the design of an XML DW from XML struction of the dimensions. Thirdly, the method
sources. Their method relies on two assumptions: creates intermediate XML documents from the
the existence of a DTD for the XML documents, initial documents; this step focuses on determining
and the conformity of these documents to their the main activity data (data involved in queries,
corresponding DTD. This method designs a DW calculations etc.). Thus, each intermediate docu-
in three steps: 1) DTD simplification mainly to ment linked to other documents represents a fact
flatten nested elements; 2) DTD graph creation document. Finally, the method updates/links all
in order to represent graphically the source struc- intermediate XML documents (fact and dimen-
ture and simplify the manual fact selection; and sion documents), in such a way that relationships
3) construction of an attribute tree for each fact between keys are established, and an XML DW
in the graph; within the attribute tree dimensions is created. In this method, several sub-steps have
and measures are found among the nodes imme- to be accomplished manually by an expert in the
diately linked to the chosen fact. In this method, XML document domain.
the selection of facts and measures is manual Furthermore, (Jensen, Møller, & Pedersen,
and requires the intervention of an expert in the 2001) studied how an OLAP cube can be obtained
domain of the XML documents that will load the from XML data. To build a cube, the DTD of the
future data warehouse. XML documents is transformed into a UML class
In an attempt to improve their method, (Vr- diagram using a set of transformation rules. Once
doljak, Banek, & Rizzi, 2003) developed a semi- the class diagram is obtained, the designer uses
automated process to design XML data warehouses it to specify an OLAP DB model (named a UML
from XML schemas. Once again, in this process, snowflake diagram) through a graphical user in-
facts and measures are chosen manually. For terface. Finally, the UML snowflake diagram is
each selected fact, they 1) build the dependency transformed into relational structures to prepare
graph from the XML schema graph; 2) rearrange the implementation of the OLAP cube. This ap-
the dependency graph to define dimensions and proach is also used by (Ouaret, Z., Bellatreche,
measures; and then 3) create a logical schema. One L., and Boussaid, O., 2007) who starts from XML
main drawback of this method is that it requires schemas instead of the DTD.
an intensive intervention of the designer. In addi- In summary, we notice that:
tion to manually identifying the fact, dimensions
and measures, the designer must also identify the 1. although data-driven approaches for multi-
many-to-many relationships among elements; dimensional design proceed automatically
these relationships are needed to construct the from either E/R or UML, they are based on
dependency graph. conceptual models that companies either do
On the other hand, the authors of (Rusu, not always have, or detain obsolete versions;
Rahayu, & Taniar, 2004) and (Rusu, Rahayu, &
430
2. the few proposed approaches for the design or structured as XML documents compliant to a
of XML-based DM/DW suppose that the given DTD. This method enjoys four main advan-
designer is able to manually identify the tages: 1) It overcomes the problem of absence/
interesting facts to be analyzed. However, obsoleteness of conventional documentation
this identification requires a high expertise (i.e., E/A diagram, UML class diagram). In fact,
both in OLAP domain and, the XML docu- it exploits the recent version of the data source
ment domain; extracted from the DBMS repository or described
3. all methods consider that the candidate DM by XML documents in use; 2) it automates the
built are of the same pertinence whereas main design steps; 3) it assists the designer in
some of these DM may not be useful for the choice of relevant multidimensional concepts
the decisional process; and among those extracted by assigning to each one a
4. the proposed approaches try to represent the relevance level reflecting its analytical potential
main DM properties at a conceptual level by for the decision-making; and 4) it keeps track of
abstracting away details of an envisaged DW the origin of each component in the generated DM
implementation platform. “Unfortunately, schema. This traceability is fundamental both to
none of these approaches defines a set of automatically derive logical representations and
formal transformations in order to: (i) univo- to define ETL processes.
cally and automatically derive every possible
logical representation from the conceptual
model; or (ii) help the designer to obtain dAtA mArt desIGn For
the most suitable logical representation of relAtIonAl And xml sources
the developed conceptual model” (Mazón,
& Trujillo, 2008). Our design method is a bottom-up method for
DM that starts from a relational database source
To overcome these problems, we propose a and XML documents compliant to a given DTD.
bi-source method that builds DM schemas from Unlike existing approaches, ours is composed of
data sources either modeled as relational databases four steps (Figure 1) among which only the last
Figure 1. Designing data mart schemas from XML and relational data sources
DATA SOURCE PRETREATMENT RELATIONS CLASSIFICATION DATA MART CONSTRUCTION DATA MART ADAPTATION
1
DTD
DTD
simplification DTDS
AN_UNIV
DTD DCS AN_UNIV
simplification GRAD_ENS COD_ENS

ENSEIGNEMENT_ASSURE COD_MAT
COD_AUD COD_SEC
MATIERE
ENSEIGNANT NBR_GRP_ENS INT_SEC
NOM_ENS
PRE_ENS NBRE_GRP INT_MAT INT_AUD
Customer VOL_HOR_ENS_MAT
2
TYP_ENS NUM_CYC
Entities Relationships
ID NBRE_ETUD
I
Transitiontree
tree
E_MAIL
VOL_HOR_MAT NAT_ENSMT
Transition
NUM_SEM DCS NAT_ENSMT
NUM_TEL_F (ENSEIGNANT)
DCS NUM_SEM
NUM_TEL_M (ENSEIGNANT)
CustomerID Name Address
construction
AN_UNIV
construction 7 DCS AN_UNIV
R1 (a,b,…,c) R4 (a#,g#,…,v) Multidimensionalconcept

concept
Multidimensional
COD_AUD COD_SEC
Locality Zip
Rm (a#,e#,…,z)
GRAD_ENS COD_ENS
R2 (e,f,…,h) ENSEIGNANT MATIERE
identification NOM_ENS NBR_GRP_ENS INT_SEC
identification
XML
Customer
R3 (g,k,…,l) Rn (e#,g#,…,y) TYP_ENS
VOL_HOR_ENS_MAT
NBRE_ETUD
NUM_CYC
3 ID E_MAIL
VOL_HOR_MAT
I NAT_ENSMT
Transitiontree
tree
XML
DCS NUM_SEM
Transition
CustomerID Name Address
XML enrichment
enrichment
number string
Locality Zip
string string
Adapted Data marts
AN_UNIV
DCS AN_UNIV
6
COD_AUD COD_SEC
GRAD_ENS COD_ENS
MATIERE
Conceptualclass
class
ENSEIGNANT
DB 1’ 4
NOM_ENS NBR_GRP_ENS INT_SEC
schema
DB schema Relationalschema
schema Conceptual AN_UNIV
PRE_ENS NBRE_GRP
VOL_HOR_ENS_MAT
INT_MAT INT_AUD
Relational Relational
TYP_ENS NUM_CYC
identification
DCS AN_UNIV NBRE_ETUD
extraction
E_MAIL
identification Data8mart
VOL_HOR_MAT
generation
NAT_ENSMT
extraction mart
NUM_SEM NUM_TEL_F (ENSEIGNANT) DCS NAT_ENSMT
generation
COD_AUD COD_SEC DCS NUM_SEM
Data base Data

GRAD_ENS COD_ENS NUM_TEL_M (ENSEIGNANT)
MATIERE
NOM_ENS
adaptation
adaptation
TYP_ENS
VOL_HOR_ENS_MAT NUM_CYC AN_UNIV
NBRE_ETUD DCS AN_UNIV
E_MAIL
DCS NUM_SEM ENSEIGNEMENT_ASSURE COD_MAT
R2 (e,f,…,h) R1 (a,b,…,c)
COD_AUD COD_SEC
NUM_TEL_M (ENSEIGNANT) GRAD_ENS COD_ENS
R1 (a,b,…,c)
MATIERE
NOM_ENS
… … PRE_ENS NBRE_GRP INT_MAT INT_AUD
R2 (e,f,…,h) TYP_ENS
VOL_HOR_ENS_MAT NUM_CYC
NBRE_ETUD
Rm(e#,g#,…,y) Rn (a#,e#,…,z) E_MAIL

…. Data marts
DCS NUM_SEM NUM_TEL_F (ENSEIGNANT)
Rm (a#,e#,…,z)
Rn (e#,g#,…,y)
5
Relational
Relational schema
schema
integration
integration
431
is manually conducted by the decision makers; in The third step of our design method (data mart
this step, they adapt the automatically constructed schema construction) extracts the multidimen-
DM schemas for their particular needs. sional concepts (facts and their measures, dimen-
As illustrated in Figure 1, our design method sions and attributes organized into hierarchies)
starts with a data source pretreatment step to re- from the classified relations and produces star
solve the structural heterogeneity of the two types models. To automate this step, we define for each
of data sources. For the relational database, this multidimensional concept a set of extraction rules.
step is trivial since it only extracts the database Our rules are independent of the semantics of the
schema (table names, column names and types, data sources and their domain. They rather rely on
and key constraints) from the repository of the the structural semantics of the relations, which is
relational DBMS. On the other hand, for the XML mainly disseminated through the key constraints
documents, it transforms their DTD into a rela- (primary and foreign keys). In addition, our rules
tional schema (i.e., a set of relational tables). To have the merit to keep track of the origin of each
do so, the DTD is first simplified to eliminate any multidimensional concept in the generated data
redundancies. Then, the simplified DTD (noted mart schemas. This traceability is fundamental
DTDs) is reorganized into a set of linked trees during the definition of ETL processes.
that we call transition trees. In addition, since a Finally, the decision makers/designers are
DTD is poor in typing information required for presented with a set of potential data mart sche-
the identification of multidimensional concepts mas that they can adjust to meet their particular
(measures, dimensional attributes …), this step analytical requirements. In this final step (data
scans sample XML documents in order to extract mart adaptation), the decision makers/designers
richer typing information; the extracted types are can add derived data, remove, and/or rename
assigned to the attributes and packed data elements DM schema elements. The application of these
in the transition trees. Based on the existing links adaptation operations is constrained to ensure that
among the typed transition trees, these latter are the resulting schemas are well-formed (Schneider,
transformed into a relational schema. This step 2003) (Salem, Ghozzi, & Ben-Abdallah, 2008),
is concluded by a schema integration phase that e.g., a fact must have at least two dimensions.
merges the two source schemas to produce one Before explaining in detail the above four steps,
semantically coherent schema; it applies exist- in the remainder of this section, we overview the
ing propositions for relational database schema concepts of XML structures and the relational
integration, cf., (Bright, Hurson, & Pakzad, 1994), model.
(Sheth, & Larson, 1990), (Ceri, Widom, 1993),
(Hull, 1997) and (Zhang, & Yang, 2008). basic xml structural concepts
Once the data source pretreatment produces the
integrated relational schema, the design continues An XML document has two types of informa-
with the relation classification step. This latter tion: the document structure and data content;
performs a reverse engineering task by examining XML provides a means for separating one from
the structure of the relations in the source sche- the other in the electronic document. The docu-
mas obtained from the first step. It automatically ment structure is given by opening and closing,
determines the conceptual class of each relation: matching tag pairs (each called an element) and
A relation conceptually either models a relation- the data content is given by the information be-
ship or an entity. This classification optimizes the tween matching tags. In addition, an element can
automatic fact and dimension identification and have attributes whose values are assigned in the
improves its results. opening tag of the element.
432
To define the structure of a set of XML docu- attributes (A1, A2,…, An) each of which is associ-
ments, a DTD document can be used. A DTD is ated with one domain representing the set of its
a context free grammar specifying all allowable possible values. Each attribute Ai is the name of
elements, their attributes, and the element nest- a role played by its domain D in the relation R.
ing structure. Given one DTD, it can be verified Thus, a relation R represents a set of tuples t1, t2,
whether an XML document is conforming to/ … tm. If for any two distinct tuples ti and tj in R
respects the DTD, and if so, the XML document there exists an attribute K such that ti(K) ≠ tj(K)_,
is said to be valid. then such an attribute (set of attributes) is called
Note that there exist other formalisms for a candidate key. It is common to choose one of
describing XML document structures, e.g., XML the candidate keys as the primary key used to
Schema and DCD (Document Content Descrip- uniquely identify tuples in the relation.
tion). However, the DTD is a formalism recom- Furthermore, a set of attributes FK in a relation
mended by the World Wide Web Consortium R1 is a foreign key if the two following rules are
(W3C) (W3C, 2008). For this, we assume in our satisfied: (1) the attributes in FK have the same
method that the structure of an XML document domain as the primary key PK of another relation
is described by a DTD and that the XML source R2, and (2) every value of FK in any tuple t1 in
documents are valid. R1 either occurs as a value of PK for some tuple
A DTD is composed of element types, sub- t2 in R2 or it is null.
element types, attributes, and terminal strings Figure 3 shows an example of a relational
such as ENTITY, PCDATA and CDATA. The database modeling a hotel room booking system.
DTD types are however very limited since all of This example is adapted from the one presented
the types are considered as strings. In addition, a in (Databasedev.co.uk, 2008). In our schema, the
DTD can constrain the occurrences of an element primary keys are underlined and the foreign keys
and a sub-element type through the symbols: “*” are followed by the sharp sign (#) and the name
(a set with zero or more elements), “+” (a set with of the referenced relation.
one or more elements), “?” (an optional element),
and “|” (alternative elements). For more details
about DTD, the reader is referred to (Sahuguet, dAtA source PretreAtment
2000) and (W3C, 2008).
Figure 2 depicts an example of a DTD describ- This first step of our design method aims at
ing e-Ticket documents. An e-Ticket document structurally homogenizing the data sources by
describes the booking in a hotel that a consumer transforming the source DTD into a set of relations
can do and/or the list of concerts that a consumer (i.e., tables). It is conducted through an automatic
can attend. Such documents can be used by an process composed of four stages: DTD simplifica-
online broker that deals with a particular hotel tion, transition tree construction, transition tree
and offers entertainment services (in this case, enrichment and relational schema generation.
concert ticket purchase).
dtd simplification
basic relational model concepts
The simplification of a DTD removes empty
In the relational model (Codd, 1970), a database elements, substitutes and transforms other ele-
is modeled by a set of relations (also called tables) ments. The empty element removal is applied
that forms a relational schema. A relation, denoted to every element that is tagged EMPTY and that
R ( (A1, A2,…, An _), has a name (R) and a list of does not declare an ATTLIST. Such an element
433
Figure 2. An example of DTD for e-Ticket documents
has no content in valid XML documents. Thus, Recall that an ENTITY is a variable used to define
it is not useful for the decision process. On the shortcuts for either commonly used text, or text
other hand, the element substitution first replaces that is difficult to type.
each reference to an ENTITY type with the text Finally, the third simplification step applies
corresponding to that entity, and secondly it re- a set of transformations to the DTD to reduce
moves the corresponding ENTITY declaration. the components of its ELEMENT declarations.
434
Figure 3. Relational database schema for hotel room booking system
Our transformations slightly differ from those 1. Flattening which converts a nested definition
presented in (Shanmugasundarma, Tufte, Zhang, into a flat representation where the binary
DeWitt, & Naughton, 1999) and (Yan, & ADA, operators “,” and “|” do not appear inside
2001) and resemble the transformations in (Jensen, any parentheses;
Møller, & Pedersen, 2001). They are of the three 2. Reduction which reduces several consecutive
types depicted in Figure 4: unary operators to a single one; and
Figure 4. Transformations to simplify a DTD
435
3. Grouping which replaces multiple occur- hand, a leaf denotes either a PCDATA element,
rences of one sub-element with a single an attribute, or an element identified as the root
occurrence having the most generic operator. node of a transition tree (the same tree if the DTD
contains a recursion). In addition, the arcs of the
All of these transformations are iteratively transition tree can be labeled with the attribute
applied to the DTD until no more transforma- type (ID, IDREF).
tions could be applied. Note that when applying
the third transformation, we may lose the initial Root Determination
order of elements; however, this will not affect the
final result (i.e., the candidate multidimensional We determine the root node of a transition tree
schemas). by one of the following four rules that we have
To illustrate these transformations, let us con- developed in (Hachaichi, Feki, & Ben-Abdallah,
sider the two following XML element declarations 2008-a):
extracted from the example of Figure 2:<!ELE-
MENT Buy (BuyDate, Customer, (ConcertDate, R1. Each DTD element that does not appear in
Concert+)*,TotalPayementB)><!ELEMENT the declaration of any other element is a
Bookings (Customer, DateBookingMade, Time- root of a transition tree.
bookingMade, (BookedStartDate, BookedEnd-
Date, Room )+, TotalPayementDueAmount)> In general, each XML document can be seen as
These elements will be simplified one root element that contains all the elements in
into:<!ELEMENT Buy (BuyDate, Customer, the document. Thus, rule R1 will extract this top-
ConcertDate*, Concert*,TotalPayementB) most element as a root of one transition tree. The
><!ELEMENT Bookings (Customer, Date- application of R1 must however exclude recursive
BookingMade, TimebookingMade, Booked- DTD where all elements are sub-elements of other
StartDate+, BookedEndDate+, Room+, elements (Yan, & ADA, 2001). We treat the case
TotalPayementDueAmount)>where the inner of a recursive DTD in the rule R4.
parentheses are dropped and each element has at
most one operator. R2. Each element that contains at least one non
PCDATA element is a root for a transition
transition tree construction tree.
As shown in Figure 1, the DTD simplification Rule R2 excludes transition trees composed of
step is followed by the construction of transition the root connected to only leaf nodes (PCDATA).
trees for the simplified DTD. In this step, a simpli- Such a tree, in the XML document, can be con-
fied DTD is split into substructures reorganized sidered as a basic data entity, in the sense that it
as trees that we call transitiontrees. These latter cannot represent a relationship. In addition, by
facilitate the relational schema generation from imposing that a transition tree contains at least one
the initial DTD. complex element, R2 ensures that the transition
Each transition tree has a root node, interme- tree represents a relationship.
diate nodes, terminal nodes (leaves) all of which
are connected by directed arcs. The root node R3. Each element contained in the declaration of
and intermediate nodes refer to one element in n elements (n ≥ 2) is the root of a transition
the simplified DTD (noted DTDS). On the other tree.
436
This rule avoids the redundancy of elements in IDREF

a transition tree and identifies the elements shared }
}
by several trees as one transition tree.
{
R4. Each element directly or transitively con- In the transition tree construction algorithm,
taining its own declaration is the root of a the function AnnotateNode(e, #) marks any sub-
transition tree. element that is the root of a transition tree with the
symbol #. This annotation is borrowed from the
This rule treats the case of recursive DTD. concept of foreign keys in relational databases; it
Informally, if an element refers to itself in its is useful to link transition trees of the same DTDS
declaration (directly or indirectly), then this ele- and to construct parameter hierarchies. Note that
ment contains a non PCDATA element, and thus, with this annotation, the constructed transition
in accordance with rule R2, this element should trees will have at most a depth of four. Such a
be the root of a transition tree. limited depth accelerates the traversal during the
For our running example (Figure 2), the above DM schema construction step. On the other hand,
four rules identifies eight roots: e-Ticket,Bookings, the function MarkArcType(E, a) annotates the
Consumer, Room, Buy, Concert, City and State. arc from E to a with the type (ID or IDREF) if
the attribute is a unique identifier or an attribute
Transition Tree Construction that points to an ID attribute. Figure 5 shows the
transition trees constructed by this algorithm for
For each identified root, this step constructs a the e-Ticket DTD example.
corresponding transition tree. This requires a
fine scan of the simplified DTDS by applying the transition tree enrichment
algorithm Create_tree (E, DTDS) starting from a
root E (Hachaichi, Feki, & Ben-Abdallah, 2008-b): To build a relational database schema from transi-
tion trees, we need the attribute types (e.g., number,
Algorithm Create_tree (E, DTDS)
date…). However, such typing information is
// E is the current node in DTDS
{ totally absent in the DTD and XML documents.
1.foreach element e in the declara- In fact, a DTD schema declares all data as strings.
tion of Edo { To assign a data type to the attributes and
1.1 AddChildNode(E, e) PCDATA elements in a transition tree, we query a
//add a child e to the node E sample set of XML documents valid with respect
1.2 if ( e is deter-
to the source DTD. For each leaf not marked with
mined as a root) then //e identi-
fied by rules 1 to 4 # (i.e., each attribute and PCDATA element), we
1.2.1 AnnotateNode(e, #) consult the data contained in its corresponding
1.3 elseif (e contains XML tag. Then, we determine a type by scanning
other elements or attributes) then its text value and cast it into one of three appro-
1.3.1 CreateTree(e, priate types: date, number or string. To assist us
DTDS)
with this analysis of the XML documents, there
}
2. for each attribute a in the dec- are several semi-structured query languages for
laration of Edo { XML documents, including XML-QL, Lorel,
2.1 AddChildNode(E, a) UnQL, XQL (from Microsoft), and XQuery (from
2.2 MarkArcType(E, a) W3C). All these languages have the notion of path
//mark arc from E to a by ID or
437
Figure 5. Transition trees for the e-Ticket DTD
expressions to navigate the nested structures of example (Figure 5) adds the data types to the leaf
XML documents. The application of the enrich- nodes not annotated with the symbol # as shown
ment step on the transition trees of our running in Figure 6.
Figure 6. Transition trees of Figure 5 enriched with data types
438
relational schema Generation 4.2 ID_found = False;

for the xml source 4.3 For each child a of n
do {
4.3.1 Add_attribute(a, Rn);
In this stage, the typed transition trees are trans- 4.3.2 If (arc(n, a) = ID)
formed into relational schemas based on the links then {
among the transition trees. This transformation is 4.3.2.1 Mark_
conducted through the algorithm XML2R which PK({a}, Rn);
uses the following notation: 4.3.2.2 ID_found =
True;
}
• Refn: the set of nodes that can be foreign 4.3.4 If (arc (n, a) = IDREF)
keys in the relation Rn (constructed on then
node n) 4.3.4.1 Mark_FK(a,
• Build_R (n): a function that builds on node Rn)
n a relation called Rn }4.4 If ( ! ID_found) then
4.4.1 ADD_ID(Rn)
• ID (Rn): a function that returns the primary
} // end for line 2
key of a relation Rn 5. For each non terminal node n ∉
• Add_attribute (a, Rn): Adds attribute aas a NTdo {
columnto the relation Rn 5.1 Rn = Build_R (n);
• Mark_PK (pk, Rn): Marks the pk list of at- 5.2 Refn = ϕ ;
tributesas the primary key of the relation 5.3 ID_Found = False;
5.4 For each child a of n
Rn
do {
• Mark_FK (a, Rn): Marks the a attributeas a 5.4.1 If (a is a leaf marked
foreign key in the relation Rn with #) then {
• ADD_ID (Rn): adds an artificial primary 5.4.1.1 Add_attri-
key to the relation Rn. The primary key bute (ID(Ra), Rn);
name is nID: the concatenation of node 5.4.1.2 Mark_FK
(ID(Ra), Rn);
name n to the string ‘ID’.
5.4.1.3 Refn= Refn ∪
ID(Ra);
}
Algorithm XML2R( ) 5.4.2 else {
5.4.2.1 Add_attri-
1. NR = the set of transition bute (a, Rn);
trees reduced to roots, i.e., with- 5.4.2.2 If (arc(n,
out any descendant. a) = ID) then {
2. NT= theset of nodes all of Mark_
whose children are leaves not an- PK({a}, Rn);
notated with #.3. For each node n ∈ ID_found =
NRdo { True;
3.1 Rn = Build_R (n); }
3.2 Add_attribute(n, Rn); 5.4.3 If (arc (a, n) =
3.3 Mark_PK({n}, Rn); IDREF) then
}
4. For each node n ∈ NTdo { // n Mark_FK({a}, Rn);
has only leave children not anno- }
tated with # }
4.1 Rn = Build_R (n); 5.5 If (! ID_found) then
439
5.5.1 If COUNT (Refn) ≥ 2 the tables issued from this step may be re-
then ferred to by relations created in step 5. For
Mark_PK(Refn, Rn);
our example, this step produces the tables
else
ADD_ID (Rn); RoomTypes, RoomFacilities, County and
} //end of algorithm Singer of the Figure 7.
• Step 5 deals with tables referencing other
This transformation algorithm examines the tables. For each such table, this step creates
set of all transition trees and operates according a table with a primary key either the node
to the following five steps: annotated with ID, or the concatenation
of its foreign keys when their number ex-
• Step 1 finds the transition trees composed ceeds one, otherwise an artificial attribute
of a single node. is added as a primary key.
• Step 2 finds theset of nodes all of whose
children are leaves not annotated with #. Figure 7 shows the relational schema generated
• Step 3 transforms each transition tree re- by applying the XML2R algorithm on the transi-
duced to its root r into a single-column tation trees of the e-Ticket DTD. In this schema,
ble with key r. This transformation avoids primary keys are underlined and, foreign keys
the lost of degenerated dimensions and are followed by the sharp sign (#) and the name
ensures that dimensions can be potentially of the referenced relation.
built on these nodes.
• Step 4 builds a table Rn for each node n relational schema Integration
all of whose children are leaves not an-
notated with #. The columns of Rn are the At this stage, we have produced one relational
children of n. If a child a of n a leading schema from the relational database and a second
arc labeled with ID, then it becomes the one from the DTD/XML documents. However,
primary key of Rn; if no such a child ex- these two schemas represent one “virtually single”
ists, then an artificial primary key is gener- data source used to load the DM. Thus, they
ated via the ADD_ID function. Note that must be integrated to represent conceptually one,
Figure 7. Relational schema derived from e-Ticket DTD
440
coherent database. In other words, any semantic into the same model, schema correspondence that
heterogeneity must be resolved at this stage. resolves naming conflicts; and schema fusion that
Quite existing works treat semantic heteroge- produces a global schema by replacing equivalent
neity in relational databases, cf., (Bright, Hurson, schemas, regrouping intersecting schemas and
& Pakzad, 1994), (Sheth, & Larson, 1990), (Ceri, collecting independent schemas. In our method,
Widom, 1993) , (Hull,1997) (Zhang, & Yang, the pre-integration step is treated through our pre-
2008) and XML document storage as relational treatment step, which leaves us with the schema
databases, cf., (Ceri, Fraternali, & Paraboschi, correspondence and fusion steps.
2000), (Deutsch, Fernandez, Suciu, 1999), (Lee, To illustrate the integration step, let us revisit
& Chu, 2000), (Schmidt, Kersten, Windhouwer, our running example (Figures 3 and 7) which re-
& Waas, 2000), (Kappel, Kapsammer, & Retsch- quires only the fusion step. The integrated schema
itzegger, 2001). In addition, selected schema of Figure 8 is produced by applying the following
heterogeneity issues were treated in the context fusion operations:
of XML and relational database integration, cf.,
(Kappel, Kapsammer, Retschitzegger, 2000). • Import the relations Payments,
These works, used in the context of database model PaymentMethods and RoomBands from
transformation, can be also used to resolve the hotel room booking relational schema.
semantic heterogeneity within our design method. • Import the relations SingerBuy and Concert
Overall, the works on relational DB schema from the e-Ticket relational schema.
integration proceed in three steps: Pre-integration • Import the relations Room, RoomTypes,
where the different source models are transformed RoomBands, RoomFacilities, Bookings,
Figure 8. Integrated relational schema issued from e-Ticket and hotel room booking
441
Customer, County, State and City from e- identified, i.e., as a relationship if it originally
Ticket and hotel room booking schemas and models a relationship and as an entity if it models
integrate them using the union operator. an entity. Correctness is less trivial than the two
first properties and it is not satisfied in the two
relAtIon clAssIFIcAtIon following situations: 1) when the primary key of
a relationship is not the concatenation of all its
In current data-driven DW/DM development foreign keys, that is, this primary key can be an
methods, entities and relationships are the key- artificial attribute such as a sequential number; or
stone of the conceptual design. More precisely, 2) when the primary key of a relationship is the
dimensions are often built from entities and date concatenation of attributes coming from empty
dimensions from temporal attributes; whereas entities. Such attributes are never foreign keys
facts are mainly built from n-ary relationships since an empty entity (i.e., entity reduced to its
linking entities and, rarely on entities. However, key) never transforms into a relation. For more
this distinction is not explicitly present in the re- details about empty entities, the reader is referred to
lational model where one single concept is used (Feki, & Hachaichi, 2007-b) where an illustrative
to model both entities and relationships, namely example is given. Table 1 shows the classification
the relational table (or table for short). of the relations presented in Figure 8.
Hence, in order to define a design process that
correctly identifies facts and dimensions from a
relational schema (i.e., a set of relational tables) we dm scHemA constructIon
must first determine the conceptual class of each
table. To do so, we perform a reverse engineering This third design step builds DM schemas mod-
task by precisely examining the structure of the eled as stars through the application of a set of
tables in the sources. In fact, this leads to a scan extraction rules defined for each multidimen-
of the set of attributes composing the primary and
foreign keys of the source tables. Table 1. Source relations classified into entities
We have presented in our previous work and relationships
(Feki, & Hachaichi, 2007-a) how to partition a
set of tables, issued from an operational infor- Relational Tables Conceptual Class
mation system, into two subsets: A subset of Room Entity
tables modeling relationships and another subset RoomTypes Entity
modeling entities. Briefly, a table representing a RoomBands Entity
relationship is characterized by its primary key RoomFacilities Entity
being composed of one or several foreign keys, Payments Entity
whereas a table representing an entity generally PaymentMethods Entity
has its primary key not containing foreign keys. Bookings Relationship
This classification should well form the two sub-
Customer Entity
sets of tables by satisfying three basic properties:
County Entity
Disjointness, Completeness and Correctness. The
State Entity
first property imposes that the two subsets share
City Entity
no common table; this ensures that each table is
Singer Entity
uniquely classified. The completeness property
Concert Entity
ensures that every table has a class. The correctness
Buy Relationship
property recommends that each table is correctly
442
sional concept. Our rules have the merit to be facts without measures, which are considered as
independent of the semantics of the data source. infrequent facts.
In addition, they keep track of the origin (table
name, column name, data type and length…) of measure Identification
each multidimensional concept in the generated
DM schemas. This traceability is fundamental as A fact contains a finite set of measures. In most
we intend to help automating the generation of cases, measures serve to compute summarized
ETL (Extract Transform and Load) procedures to results (e.g., total amount of sales by month and
load the designed DM. year, by mart…) using aggregate functions; hence,
measures have numeric values. Therefore, we ex-
Fact Identification tract measures mainly from fact-tables (i.e., table
on which the fact is built). Furthermore, rarely
To identify facts, we exploit our previous table few additional measures could be extracted not
classification and build a set of facts of first- from a fact-table itself but from tables parallel to
relevance level from tables representing relation- it. Below, we informally explain how we extract
ships and then, a set of facts of second-relevance measures in each of these cases.
level issued from tables representing entities.
This distinction in analysis relevance levels is Measure Identified from a Fact-Table
very useful; it assists the DW designer during the
selection from several generated facts those facts To construct a significant set of candidate measures
that have a higher analysis potentiality. for a fact F-T built on a table T, we exclude key-
In fact, in all DW design approaches it has attributes from the set of numeric attributes of T
been unanimously accepted that a business ac- because keys are generally artificial and redundant
tivity (e.g., Sales, Bill) is generally modeled as data, and they do not trace/record the enterprise
a relationship. This observation incites to limit business activity. Moreover, we have shown in
the construction of facts mainly on relationships (Feki, & Hachaichi, 2007-b) that we must exclude
(Golfarelli, Maio, & Rizz, 1998), (Cabibbo, L., & from T its “non key attributes belonging to other
Torlone, R. 1998), (Feki, Nabli, Ben-Abdallah, & tables” because these attributes really represent
Gargouri, 2008) and rarely on entities (Moody , & keys issued from empty entities. Table 2 shows all
Kortnik, 2000), (Phipps, & Davis, 2002), (Feki, measures extracted from each extracted fact-table.
Nabli, Ben-Abdallah, & Gargouri, 2008). On the
other hand, in practice, not all relationships are
useful to build facts; so we limit the set of facts at
the first-relevance level to those containing a non
Table 2. Facts and measures for the integrated
key numeric attribute. For the integrated schema
schema issued from hotel room booking and e-
of our running example e-Ticket and hotel room
Ticket of Figure 8
booking, we obtain the facts depicted in Table 2
where a fact built on a table T is conventionally Fact Relevance Measure
named F-T. level
To complete this fact identification step, we F-Room Second Price
consider that each table representing an entity and F-Payments Second PaymentAmount
containing a numeric, non key attribute is a fact at F-Bookings First
TotalPayementDueAmount
TotalPayementDueDate
the second-relevance level (cf., Table 2). Note that
F-Buy First TotalPayementB
the numeric non key attribute condition excludes
443
Measure Identified from Parallel Tables must intervene to define the necessary aggregation
functions which are semantics dependent.
As mentioned above, a second origin of measures
is parallel tables. We adapted the definition of dimension Identification
parallel tables from the concept of parallel re-
lationships which is specific to the E/R model A dimension is generally made up of a finite set
(Seba, 2003). In an E/R diagram, a relationship of attributes that define various levels of details
R1 connected to m entities is said to be parallel (hierarchies), whereas others are less significant
to a relationship R2 connected to n entities (m≤n) but used, for instance, to label results or to restrict
if the m entities linked to R1 are also linked to data processed in queries. These latter are called
R2. By analogy, we define the concept of parallel weak (or non dimensional) attributes. The set of
tables as follows: candidate dimensions for a given fact can be built
Let T1 and T2 be two relationship tables such either on tables modeling entities or attributes.
as T1 and T2 are connected to m and n (m≤ n) Given a fact F-T (i.e., fact build on table T), we
tables respectively; T1 is said to be parallel to T2 consider every table T1 that represents an entity
(noted T1//T2) if and only if the primary key of and that is directly referred by the table T as a
T1 is included in or equal to the primary key of candidate dimension for F-T. Conventionally, the
T2 (Feki & Hachaichi, 2007-c). Note that, in this name of this dimension is D-T1 and its identifier
definition, T1 and T2 are assumed to be relation- is the primary key of T1.
ships because entities could not be parallel; this In addition to dimensions built on tables, we
optimizes the search of parallel tables (Feki, & can build a dimension both on an attribute of a
Hachaichi, 2007-c). In our running example, there special data type (Boolean, temporal) as well as
are no parallel tables. on an attribute issued from an empty entity. Such
Let T1 and T2 be already identified as two fact- a dimension is known in data warehousing as a
tables and let T1 be parallel to T2. The fact F-T1 degenerated dimension. For instance, a Boolean
(built on T1) can receive other measures coming column splits data-rows of its table into two
from the fact F-T2 (built on T2) by aggregating subsets; thus, such an attributecan be an axis of
measures of F-T2 before moving them to F-T1. analysis. In practice, a degenerated dimension is
Since the measures in F-T2 are more detailed than integrated inside the fact.
those in F-T1 (i.e., F-T2 has n-m>0 dimensions A Boolean column b pertinent to a fact-table
more than F-T1), then to move them to F-T1, they T produces for T a candidate, degenerated dimen-
must be aggregated on the set of their uncommon sion named D-b and identified by b. For instance,
dimensions. Note that if the dimension set of a Gender column in a Client database table can
F-T1 is equal to the dimension set of F-T2, then build a dimension D-Gender.
the set of uncommon dimensions between F-T1 Furthermore, the data warehouse community
and F-T2 is empty; therefore, T1 is parallel to T2 assumes a DW as achronological collection of data
and reciprocally. Consequently, the measures of (Kimball, Reeves, Ross, & Thornthwaite, 1998).
both facts F-T1 and F-T2 have the same granular- Consequently, the time dimensionappears in all
ity level and could be seen as two halves of the data warehouses. For this reason, we propose to
same fact; hence, we recommend merging them build dimensionson temporalattributes as follows:
into a single fact conventionally called F-R1-R2. A temporal attribute (date or time) belonging to
In our design method,aggregated measures as a fact-table T timestamps the occurrences of the
well as dimensions used in their calculation are fact built on T; it generates a candidate dimension
automatically identified. However, the designer for T where it is the identifier. For the relational
444
Table 3. Dimensions for the extracted facts Table 4. Parameters for the dimensions of Table
3
Fact Dimension Identifier
D-RoomTypes RoomTypeID Hierarchy parameters
Dimension
F-Room D-RoomBands RoomBandID (From finest to coarsest)
D-Facilities RoomFaciliyID CityID StateID CountyID
D-Customer
F-Pay- D-Customer CustomerID CustomerDOB
ments D-PaymentMethods PaymentMethodID RoomTypeID
D-Customer CustomerID D-Room RoomBandID
D-DateBookingMade DateBookingMade RoomFaciliyID
F-Book- D-TimebookingMade TimebookingMade SingerID

D-Concert
ings D-Room RoomID CityID StateID CountyID
D-BookedStartDate BookedStartDate
D-BookedEndDate BookedEndDate
D-Customer CustomerID b) a Boolean or temporal attribute belonging
D-Concert CustomerID to a dimension table d; or
F-Buy
D-BuyDate BuyDate c) a non (primary or foreign)-key attribute be-
D-ConcertDate ConcertDate longing to a dimension-table d and to other
tables.
schema issued from e-Ticket and Hotel room The recursive application of the above two
booking, the above rules produce the dimensions steps on the tables obtained in step a) produces
shown in Table 3. parameters of level higher than two. Table 4 pres-
ents the hierarchy parameters of each dimension
Hierarchy Identification in Table 3.
A parameter may functionally determine some
The attributes of a dimension are organized into attributes within its origin table; these attributes
one or several hierarchies. These attributes are describe the parameter and, therefore, are called
ordered from the finest towards the coarsest descriptive (also non-dimensional or weak) attri-
granularity. In order to distinguish them, we call butes. Descriptive attributes for a parameter p are
these attributes parameters. In addition, any hier- non-key textual or numerical attributes belonging
archy of a dimension d has the identifier of d as to a table supplying p and not belonging to other
its finest parameter (i.e., at the level one) already tables. Among these attributes, those textual are
extracted with d. The remaining parameters (i.e., more significant than numerical ones (Feki, &
those of level higher than one) forming candidate Hachaichi, 2007-a). Table 5 presents for each
hierarchies for d will be extracted in two steps. parameter of Table 4 its associated descriptive
First, we extract the parameters located im- attributes.
mediately after the dimension identifier; each ob-
tained parameter constitutes a hierarchy. Secondly,
for each one, we extract its successor parameters. cAse toolset
Thus, a parameter of level two is either:
To support our design method, we have imple-
a) the primary key of a table of class Entity mented the CAME (“Conception Assistée de
directly referred by a dimension-table d; Magasins et d’Entrepôts de données”) case toolset.
445
Table 5. Descriptive attributes for the parameters

CAME carries out the design of conceptual DM
of Table 4
schemes starting from either the relational database
Hierarchy parameters Descriptive attributes schemas of an operational database, or from a set
RoomTypeID TypeDesc
of XML documents compliant to a given DTD.
RoomBandID BandDesc
Its main functions cover our DM design method
RoomFaciliyID FacilityDesc
steps: 1) Acquisition and pretreatment of a DTD
using a DTD parser and the XQuery language to
RoomID Price, Floor, AdditionalNotes
extract typing information; 2) Conversion of XML
PaymentMethodID PaymentMethod
structure to relational scheme; 3) Acquisition of
CountyID CountyName
the relational schema; 4) Schema integration by
StateID StateName
applying the fusion approach presented in (Hull,
CityID CityName
1997); 5) Conceptual class identification; 6)
CustomerForenames
DM Conceptual design whose result can be seen
CustomerSurnames
both in a tabular format (as illustrated in Figure 9
CustomerHomePhone
CustomerID through the running example) and in a graphical
CustomerWorkPhone representation (Figure 10); and 7) Adaptation of
CustomerMobilePhone the obtained DM when the designer adjusts the
CustomerEmail constructed DM schemas to the analytical require-
SingerID
SingerForenames ments of decisional users. In this step, CAME is
SingerSurnames linked to our case tool MPI-EDITOR (Ben Abdal-
ConcertID ConcertName lah, Feki, & Ben-Abdallah, 2006) which allows the
designer to graphically manipulate the built DM
Figure 9. Candidate DM schema for the integrated schema
446
Figure 10. GUI for DM schema adaptation
schemas and adapt them to produce well-formed analytical subjects that could be automatically
schemas (Figure 10). extracted from their data sources. In addition, it
guarantees that the extracted subjects are load-
able fromthe enterprise information system and/
conclusIon or the XML source documents. Furthermore, it
keeps track of the origin of each component of the
In this chapter, we have presented a bottom-up/ generated DM schema; this traceability informa-
data-driven design method for DM schemas from tion is vital for the definition of ETL procedures.
two types of sources: a relational database and We are currently extending our design method
XML documents compliant to a given DTD. Our to deal with other types of data sources and, in
method operates in three automatic steps (Data particular, XML documents with XML schemas
source pretreatment, relation classification, and and object databases. Furthermore, we are ex-
DM schema construction) followed by a manual amining how to integrate adjusted/validated DM
step for DM schema adaptation. It exploits the schemas to design a DW schema loadable from
recent schema/DTD version of a data source to all the considered sources.
automatically apply a set of rules that extract all
candidate facts with their measures, dimensions
and hierarchies. In addition, being automatic, our reFerences
design method is supported by a CASE toolset
that allowed us to evaluate it through several Ben Abdallah, M., Feki, J., & Ben-Abdallah, H.
examples. In this paper, we illustrated the method (2006). MPI-EDITOR : Un outil de spécification
through an e-Ticket DTD used by an online broker de besoins OLAP par réutilisation logique de
and a relational database modeling a hotel room patrons multidimensionnels. In Proceedings of
booking system. the Atelier des Systèmes Décisionnels (ASD’06),
Our design method assists decision makers in Agadir, Morocco.
defining their analytical needs by proposing all
447
Böhnlein, M., & Ulbrich-vom Ende, A. (1999). Codd, E. F. (1970). A relational model of data for
Deriving initial data warehouse structures from large data banks. ACM Communications, 13(6),
the conceptual data models of the underlying 377–387. doi:10.1145/362384.362685
operational information systems. In Proceedings
Databasedev.co.uk. (2008). Sample data models
of the 2nd ACM international workshop on Data
for relational database design. Retrieved July
warehousing and OLAP, Kansas City, Missouri
30, 2008, from http://www.databasedev.co.uk/
(pp. 15-21).
data_models.html
Bonifati, A., Cattaneo, F., Ceri, S., Fuggetta, A.,
Deutsch, A., Fernandez, M., & Suciu, D. (1999).
& Paraboschi, S. (2001). Designing data marts for
Storing semi structured data in relations. In Pro-
data warehouse. ACM Transactions on Software
ceedings of the Workshop on Query Processing
Engineering and Methodology, 10, 452–483.
for Semi structured Data and Non-Standard Data
doi:10.1145/384189.384190
Formats.
Boufares, F., & Hamdoun, S. (2005). Integration
Feki, J., & Ben-Abdallah, H. (2007). Multidimen-
techniques to build a data warehouse using het-
sional pattern construction and logical reuse for
erogeneous data sources. Journal of Computer
the design of data marts. International Review on
Science, 48-55.
Computers and Software, 2(2), 124–134.
Bright, M. W., Hurson, A. R., & Pakzad, S. (1994).
Feki, J., & Hachaichi, Y. (2007). Du relationnel
Automated resolution of semantic heterogene-
au multidimensionnel: Conception de magasins
ity in multidatabases. [TODS]. ACM Transac-
de données. In Proceedings of the Revue des Nou-
tions on Database Systems, 19(2), 212–253.
velles Technologies de l’Information: Entrepôts
doi:10.1145/176567.176569
de données et analyse en ligne (EDA 2007) (Vol.
Bruckner, R., List, B., & Schiefer, J. (2001). De- B-3, pp.5-19).
veloping requirements for data warehouse systems
Feki, J., & Hachaichi, Y. (2007). Conception as-
with use cases. In Proceedings of the 7th Americas
sistée de MD: Une démarche et un outil. Journal
Conf. on Information Systems (pp. 329-335).
of Decision Systems, 16(3), 303–333. doi:10.3166/
Cabibbo, L., & Torlone, R. (1998). A logical jds.16.303-333
approach to multidimensional databases. In Pro-
Feki, J., & Hachaichi, Y. (2007). Constellation dis-
ceedings of the Conference on Extended Database
covery from OLTP parallel-relations. In Proceed-
Technology, Valencia, Spain (pp. 187-197).
ings of the 8th International Arab Conference on
Ceri, S., Fraternali, P., & Paraboschi, S. (2000). Information Technology ACIT 07, Lattakia, Syria.
XML: Current developments and future challenges
Feki, J., Nabli, A., Ben-Abdallah, H., & Gar-
for the database community. In Proceedings of
gouri, F. (2008). An automatic data warehouse
the 7th Int. Conf. on Extending Database Technol-
conceptual design approach. In Encyclopedia of
ogy (EDBT), (LNCS 1777). Berlin, Germany:
data warehousing and mining (2nd ed.). Hershey,
Springer.
PA: IGI Global.
Ceri, S., & Widom, J. (1993). Managing semantic
Giorgini, P., Rizzi, S., & Maddalena, G. (2005).
heterogeneity with production rules and persistent
Goal-oriented requirement analysis for data ware-
queues source. In Proceedings of the 19th Inter-
house design. In Proceedings of the ACM Eighth
national Conference on Very Large Data Bases
International Workshop on Data Warehousing and
(pp. 108-119).
OLAP, Bremen, Germany (pp 47-56).
448
Golfarelli, M., Maio, D., & Rizzi, S. (1998). Kappel, G., Kapsammer, E., Rausch-Schott, S.,
Conceptual design of data warehouses from E/R & Retschitzegger, W. (2000). X-ray - towards
schemas. In Proceedings of the Conference on integrating XML and relational database systems.
System Sciences, Kona, Hawaii. Washington, DC, In Proceedings of the 19th Int. Conf. on Concep-
USA: IEEE Computer Society. tual Modeling (ER), Salt Lake City, USA (LNCS
1920). Berlin, Germany: Springer.
Golfarelli, M., Rizzi, S., & Vrdoljak, B. (2001).
Data warehouse design from XML sources. In Kappel, G., Kapsammer, E., & Retschitzegger, W.
Proceedings of the Fourth ACM International (2001). XML and relational database systems - a
Workshop on Data Warehousing and OLAP At- comparison of concepts. In Proceedings of the
lanta, GA, USA (pp. 40-47). International Conference on Internet Computing
(1) (pp. 199-205).
Hachaichi, Y., & Feki, J. (2007). Patron multidi-
mensionnel et MDA pour les entrepôts de données. Kimball, R., Reeves, L., Ross, M., & Thornthwaite,
In Proceedings of the 2nd Workshop on Decisional W. (1998). The data warehouse lifecycle toolkit.
Systems, Sousse-Tunisia. New York: John Wiley & Sons.
Hachaichi, Y., Feki, J., & Ben-Abdallah, H. (2008). Lee, D., & Chu, W. (2000). Constraints-preserving
XML source preparation for building data ware- transformation from XML document type defini-
houses. In Proceedings of the International Con- tion to relational schema. In Proceedings of the
ference on Enterprise Information Systems and 19th Int. Conf. on Conceptual Modeling (ER), Salt
Web Technologies EISWT-08, Orlando, Florida, Lake City, USA (LNCS 1920). Berlin, Germany:
USA (pp. 61-67). Springer.
Hachaichi, Y., Feki, J., & Ben-Abdallah, H. (2008). List, B., Bruckner, R. M., Machacze, K., &
Du XML au multidimensionnel: Conception de Schiefer, J. (2002). A comparison of data ware-
magasins de données. In Proceedings of the 4èmes house development methodologies case study
journées francophones sur les Entrepôts de Don- of the process warehouse. In Proceedings of the
nées et l’Analyse en ligne (EDA 2008), Toulouse, International Conference on Database and Expert
RNTI, Toulouse, France (Vol. B-4. pp. 45-59). Systems Applications DEXA
Hull, R. (1997). Managing semantic heteroge- Mazón, J.-N., & Trujillo, J. (2008). An MDA
neity in databases: A theoretical prospective. approach for the development of data ware-
In Proceedings of the sixteenth ACM SIGACT- houses. Decision Support Systems, 45(1), 41–58.
SIGMOD-SIGART symposium on Principles of doi:10.1016/j.dss.2006.12.003
database systems, Tucson, AZ, USA (pp. 51-61).
Moody, D., & Kortnik, M. (2000). From enterprise
Inmon, W. H. (1996). Building the data warehouse. models to dimensional models: A methodology for
New York: John Wiley & Sons. data warehouse and data mart design. In Proceed-
ings of the DMDW’00, Suede.
Jensen, M., Møller, T., & Pedersen, T. B. (2001).
Specifying OLAP cubes on XML data. Journal Ouaret, Z., Bellatreche, L., & Boussaid, O. (2007).
of Intelligent Information Systems. XUML star: Conception d’un entrepôt de données
XML. In Proceedings of the Atelier des Systèmes
d’Information Décisionnels, Sousse, Tunisie (pp.
19-20).
449
Paim, F. R. S., & Castro, J. B. (2003). DWARF: An Schmidt, A. R., Kersten, M. L., Windhouwer, M.
approach for requirements definition and manage- A., & Waas, F. (2000). Efficient relational storage
ment of data warehouse systems. In Proceedings and retrieval of XML documents. In Proceed-
of the Int. Conf. on Requirements Engineering, ings of the Workshop on the Web and Databases
Monterey Bay, CA. (WebDB), Dallas, USA.
Phipps, C., & Davis, K. (2002). Automating data Schneider, M. (2003). Well-formed data ware-
warehouse conceptual schema design and evalu- house structures. In Proceedings of the 5th Inter-
ation. In Proceedings of the 4th Int. Workshop on national Workshop at VLDB’03 on Design and
Design and Management of Data Warehouses Management of Data Warehouses (DMDW’2003),
(Vol. 58, pp. 23-32). Berlin, Germany.
Prakash, N., & Gosain, A. (2003). Requirements Seba, D. (2003). Merise - concepts et mise en
driven data warehouse development. In Proceed- œuvre. France: Eni.
ings of the 15th Conference on Advanced Infor-
Shanmugasundarma, J., Tufte, K., He, G., Zhang,
mation Systems Engineering Short Paper Proc.,
C., DeWitt, D., & Naughton, J. (1999). Relational
Velden, Austria.
database for querying XML documents: Limitation
Prat, N., Akoka, J., & Comyn-Wattiau, I. (2006). and opportunities. Proceedings of the 25th VLDB
A UML-based data warehouse design method. Conferences, Scotland.
Decision Support Systems, 42, 1449–1473.
Sheth, A. P., & Larson, J. A. (1990). Feder-
doi:10.1016/j.dss.2005.12.001
ated database systems for managing distrib-
Rusu, L. I., Rahayu, W., & Taniar, D. (2004). On uted, heterogeneous, and autonoumous data-
data cleaning in building XML data warehouses. bases. ACM Computing Surveys, 22(3), 183–236.
In Proceedings of the 6th Intl. Conference on doi:10.1145/96602.96604
Information Integration and Web-based Applica-
Vrdoljak, B., Banek, M., & Rizzi, S. (2003). De-
tions & Services (iiWAS2004), Jakarta, Indonesia
signing Web warehouses from XML schema. In
(pp. 797-807).
Proceedings of the 5th International Conference
Rusu, L. I., Rahayu, W., & Taniar, D. (2005). A Data Warehousing and Knowledge Discovery:
methodology for building XML data warehouses. DaWak, Prague Czech.
International Journal of Data Warehousing and
Widom, J. (1999). Data management for XML
Mining, 1(2), 67–92.
- research directions. IEEE Data Engineering
Sahuguet, A. (2000). Everything you ever wanted Bulletin, Special Issue on XML, 22(3).
to know about DTDs, but were afraid to ask. In
Wikipedia Encyclopedia. (2008). Database. Re-
Proceedings of the International Workshop on the
trieved August 1, 2008, from http://en.wikipedia.
Web and Databases WebDB 2000 (pp. 171-183).
org/wiki/Database
Salem, A., Ghozzi, F., & Ben-Abdallah, H. (2008).
World Wide Web Consortium XML Schema.
Multi-dimensional modeling - formal specifica-
(2008). W3C candidate recommendation. Re-
tion and verification of the hierarchy concept. In
trieved August 1, 2008, from http://www.w3.org/
. Proceedings of the ICEIS, 2008(1), 317–322.
XML/Schema.html
450
Yan, M. H., & Ada, W. C. F. (2001). From XML to responding to that entity, and then it removes the
relational database. In Proceedings of the CEUR corresponding ENTITY declaration, and apply
Workshop. Flattening, Reduction and Grouping transforma-
tions.
Zhang, L., & Yang, X. (2008). An approach to
eXtensible Markup Language (XML): XML
semantic annotation for metadata in relational
is a general-purpose specification for creating
databases. In Proceedings of the International
custom markup languages. It is classified as an
Symposiums on Information Processing (ISIP)
extensible language because it allows the user to
(pp. 635-639).
define the mark-up elements. The main purpose
of XML is to aid information systems in sharing
data, especially via the Internet.
key terms And deFInItIons Model Integration: Model integration produc-
es a single model that combines two or more input
Data Mart: Data marts are analytical data models. The produced model can be represented
stores designed to focus on specific business in the same definitional formalism as the input
functions for a specific community within an models (or in one of the definitional formalisms
organization. A data mart is designed according used by the heterogeneous input models). The
to a specific model, namely the multidimensional expression of the new model must be formally
model that highlights the axes of data analyses. correct within the definitional formalism used.
Data marts are often derived from subsets of a Relational Data Model: The relational data
data warehouse data, though in the bottom-up model was introduced by E.F. Codd in 1970. It
data warehouse design methodology the data is particularly well suited for modeling business
warehouse is created from the union of data marts. data. In this model, data are organized in tables.
Document Type Definition (DTD): A DTD The set of names of the columns is called the
defines the tags and attributes that can be used in “schema” of the table. The relational model is
an XML document. It indicates which tags can the model in most common use today.
appear within other tags. XML documents are Star Schema: The star schema (sometimes
described using a subset of DTD which imposes a referenced as star join schema) is the simplest
number of restrictions on the document’s structure, model of multidimensional schema. The star
as required per the XML standard. schema consists of a few facts (possibly just one)
DTD Simplification: The simplification of referencing any number of dimensions.
a DTD removes empty elements, replaces each
reference to an ENTITY type with the text cor-
This work was previously published in Data Warehousing Design and Advanced Engineering Applications: Methods for Com-
plex Construction, edited by Ladjel Bellatreche, pp. 55-80, copyright 2010 by Information Science Reference (an imprint of
IGI Global).
451
452
Chapter 2.8
Migrating Legacy
Information Systems to Web
Services Architecture
Shing-Han Li
Tatung University, Taiwan
Shi-Ming Huang
National Chung Cheng University, Taiwan
David C. Yen
Miami University, USA
Cheng-Chun Chang
National Chung Cheng University, Taiwan
AbstrAct aforementioned architecture, enterprise informa-

tion systems tend to be more flexible and agile
The lifecycle of information system (IS) became to fit into the capricious business environment,
relatively shorter compared with earlier days as a and thus, be easier to integrate with additional
result of information technology (IT) revolution applications. In other words, it is indeed an es-
and advancement. It is tremendous difficult for sential requirement for an enterprise to establish
an old architecture to catch up with the dynamic such a system to further improve corporation’s
changes occurred in the market. To match with productivity and operational efficiency. In specific,
the fast pace of challenges, enterprises have to the requirement is simply to migrate the legacy
use the technology/concept of information system systems to be SOA architecture. However, it is a
reengineering (ISR) to preserve the value of their trade-off between the value of legacy systems and
legacy systems. Consequently, web services- the compatibility with SOA to decide whether this
based systems with Service-Oriented Architecture alternative is a feasible one. The purpose of this
(SOA) are widely accepted as one of the possible manuscript is to propose a migrating solution to
solutions for an enterprise information system convert the architecture of the legacy system into
to retain/keep its old legacy systems. Using this SOA with a systematic approach. This proposed
methodology is different from the traditional
DOI: 10.4018/978-1-60566-172-8.ch015 object-oriented approaches, which migrates the
Migrating Legacy Information Systems to Web Services Architecture
system to have a services-oriented focus without are two approaches available to migrate the
incorporating general objected-oriented (OO) legacy system to the Web services architecture
or functional oriented feature. In this study, a (Vanston, 2005). The first approach is the legacy
case study and information capacity theory were externalization approach. This approach is usu-
employed to verify/validate that this approach is ally the main alternative available on the current
indeed an effective and a practicable one. market. It generally uses strategic or pointed
forms, along with new types of interface display,
to develop the integrated products (such as “Web
IntrodutIon Scraping”). The other approach is the component
encapsulation approach. This is another viable
Due to the dynamic advancement of information alternative to utilize the component standard
technology (IT), the life cycle of the informa- technology like Common Object Request Broker
tion system (IS) is greatly reduced to a certain Architecture (CORBA) (OMG, 1995; Vinoski,
extent. Generally speaking, the traditional legacy 1997), Component Object Model (COM) (Mi-
information systems possess such undesirable crosoft, 2007), or Enterprise Java Beans (EJB)
characteristics as latency of information, poor (Sun, 2007) to encapsulate the legacy system into
reach, inflexibility, and higher cost of maintenance. the components, and then translate them into a
Furthermore, the traditional system architectures Web Services standard. Ultimately, this second
such as centralized and client/server are frequently approach is migrated to the component-based
incompatible with the requirements and specifi- and transaction-oriented framework (such as IBM
cations which exist in today’s business environ- WebSphere and BEA WebLogic) (Liu, Fekete, &
ment. To be more specific, the legacy information Gorton, 2005; Waguespack & Schiano, 2004).
systems have these aforementioned shortcomings, Both of the aforementioned approaches may not
which have prevented the businesses and/or or- be a bad way for the legacy system to migrate into
ganizations to react/respond dynamically to the the equivalent Web services standards. However,
rapid challenges as they should. Consequently, they normally utilize the hard-cording technique
enterprises have a strong need to utilize the tech- to implement the interface with the correspond-
nology of information system reengineering (ISR) ing standard (Brereton & Budgen, 2000; Kwan
to preserve the value of their legacy systems. & Li, 1999; McArthur, Saiedian, & Zand, 2002).
In this situation, enterprises or software com- Being a traditional structure program, the system
panies are always in a dilemma of redeveloping/ normally has a shorter life cycle and lacks scal-
redesigning their legacy systems to include the ability, feasibility, and reusability. Further, it
newer Web services components (Bouguettaya, would be much more difficult to maintain in the
Malik, Rezgui, & Korff, 2006; Chen, Zhou, & future. On the other hand, if a company is applying
Zhang, 2006; Kim, Sengupta, Fox, & Dalkilic, the component encapsulation approach without
2007). Discarding and redeveloping the existing incorporating appropriate component migrating
systems not only wastes the money allocated for methods, the system still has these aforementioned
software investments, but also causes organiza- shortcomings (Rahayu, Chang, Dillon, & Taniar,
tions to lose competitive advantages to meet 2000). Unfortunately, most alternatives adopted
numerous unanticipated contingencies and/or now by enterprises and/or businesses do not use
uncertainties. the proper component migrating method.
Based on prior study (Ommering, 2005), the Many related studies (Erickson, Lyytinen,
system migration will be one of the best ways to & Siau, 2005; Fong, Karlapalem, Li, & Kwan,
reengineer a legacy system. Traditionally, there 1999; Gall, Klosch, & Mittermeir, 1995; Kwan
453
& Li, 1999; Sang, Follen, Kim, & Lopez, 2002) services-oriented or functional-oriented ones.
have presented methods that can be utilized to Additionally, this approach can be employed to
systematically reengineer the legacy system develop a system which will be more flexible and
into the Object-Oriented (OO) or the distributed adaptable to fit better to the constantly-changing
system. However, the Web Services architecture business environment. Furthermore, this proposed
by nature is different from a general distributed approach will no doubt make the conversion
system. The core concept of Web Service is a process easier to integrate with other additional
Service-Oriented Architecture (SOA) (Huang, applications.
Hung, Yen, Li, & Wu, 2006; Stal, 2002). In The remainder of this article is organized as
the SOA environment, resources in a network follows: The second section provides a brief over-
are made available as an independent service view of some legacy systems’ reengineering and
that can be accessed without any knowledge of Web services approaches. The proposed legacy
the underlying platform implementation (Erl, system’s migrating approach and the implementa-
2005). Web services can certainly rely on a Web- tion of a prototype will be discussed in the third
services composition language (WSCL) such as section. The fourth section contains the system
the Business Process Execution Language for implementation using simulation and a real case
Web Services (BPEL4WS) (IBM, BEA Systems, study. A comparison of the proposed approach with
Microsoft, SAP AG, & Siebel Systems, 2002) to others is provided in the fifth section. Finally, the
transform an existing Web service into a new type sixth section concludes this manuscript.
of Web service by employing well-defined process
modeling constructs (Chen, Hsu, & Mehta, 2003;
Curbera, Khalef, Mukhi, Tai & Weerawarana, lIterAture revIeW
2003), enterprise architects believe that SOA can
help businesses respond more quickly and cost- Information Flow for
effectively to fast-changing market conditions business Process (IFbP)
(Sutor, 2006). This style of architecture, in fact,
promotes reusability at the macro level (service) Business Process Management (BPM) is one of
rather than the micro levels (objects). By doing the basic elements of Web services architecture
so, it can greatly simplify interconnection to and (Basu & Kumar, 2002). It can be decomposed of
usage of existing IT (legacy) assets (Carroll & two major phases—process design and diagnosis.
Calvo, 2004). The process diagnosis phase will discover an entire
The purpose of this study is to propose a meth- picture of the business process for an enterprise,
odology which utilizes the existing data design also known as the AS-IS Model. The exploration
of an information system to migrate the legacy of AS-IS Model is a very time-consuming and
system to an SOA system. The benefits include experience-oriented task. As a result, an enterprise
the following items: First, this approach has the has to spend a lot of time and pay huge labor costs
advantage of reengineering the legacy system during the process diagnosis phase. Besides, it is
to various system components from a technical difficult for the process designers to transform
aspect, and use the Web services composition one process model to another equivalent one.
language (WSCL) to translate the existing model There are certain gaps among different process-
into the new Web services architecture. Unlike designing methods. Thus, the study of Shi-Ming
traditional object-oriented approaches, this pro- Huang and Fu-Yun Yu (2004) investigates a
posed methodology migrates the old systems to novel methodology for business process discov-
454
ery based on information flow, called IFBP. This Indirect Migration

IFBP methodology includes the following three
phases: transformation, integration, and conver- Gall et al. (1995) proposed an approach, which
sion phases. The input of IFBP is actually dataflow re-architect the old procedural software to an
diagrams (DFD) for the existing information object-oriented architecture. The transformation
systems. The output is an Event-Driven Process process was developed to identify potential objects
Chain (EPC) diagram for enterprise process flow. in the procedural source codes in order to enable the
utilization of object-oriented concepts for future,
Information system reengineering related software maintenance activities. This ap-
proach was not directly performed on the source-
There are several direct and indirect system mi- code level, but instead, different representations
grating approaches. were developed out of the procedural program on
higher levels of abstraction (e.g., structure charts,
Direct Migration data flow diagrams, entity-relationship diagrams,
and application models), which represent differ-
Sang et al. (2002) presented an approach to inte- ent kinds of program information. Additional
grate legacy applications written in FORTRAN application-domain knowledge was introduced
into a distributed object-based framework. FOR- by human experts to support the program trans-
TRAN codes are modified as little as possible when formation to enable several higher-level decisions
being decomposed into modules and wrapped during the development process.
as objects. They used object-oriented technique Kwan and Li (1999) proposed a methodology to
such as C++ objects to encapsulate individual reengineer those previously-developed relational
engine components as well as the CORBA, and database systems to OO database systems. Their
implement a wrapper generator, which takes the approach is based on the input of: (1) Extended
FORTRAN applications as input and generates Entity Relationship model (EER) that provides
the C++ wrapper files and interface definition rules for structuring data; (2) Data Dictionary
language file. that provides static data semantics; and (3) DFD
Serrano, Montes, and Carver (1999) presented that provides dynamic data semantics. This ap-
a semiautomatic, evolutionary migration method- proach captures the existing database design, uses
ology that produces an object-based distributed knowledge in OO modeling, and then represents
system for legacy systems. They first used ISA them by means of production rules, to guild the
(Identification of Subsystems based on Associa- pattern extraction algorithm that is applied to
tions), a design recovery and subsystem classi- perform the data mining process to identify the
fication technique to produce a data-cohesive “data dependency of a process to an object”. The
hierarchical subsystem decomposition of the existing Data Dictionary, DFD, and EER model
subject system. Second, they adapted the sub- are all useful and hence, are needed for capturing
systems to develop the object-oriented paradigm. the existing database design to recover the hidden
Third, they wrapped up and defined interfaces of dynamic semantics.
the subsystems in order to define components. Huang et al. (2006) proposed a methodology
Finally, middleware technologies for distributed that focused on how to migrate legacy systems
systems were used to implement the communica- with a well-structured OO analysis to ensure the
tion between components. quality of a reengineered component-based sys-
tem. This research adopted the well-structured
object-oriented analysis to improve the quality
455
of reengineered systems. The result of the reen- are linked together, they can inter-operate to ac-
gineered system will be a Web-enabled system. complish an even higher level of business function
The proposed migration approach discusses how to (Herzum & Sims, 1998; Jin, Urban, & Dietrich,
process a well-structured object-oriented analysis. 2006; Lee, Pipino, Strong, & Wang, 2004; Vitha-
The research considered the following four fac- rana & Jain; 2000; Zhao & Siau, 2007). The scope
tors—(1) multi-value attributes, (2) inheritance of a component discussed in this article will include
relationships, (3) functional dependency, and the class and its related interfaces together as a
(4) object behavior to ensure the quality of a component. By doing so, it can preserve the simple
reengineered component-based system. Further, execution function of a component from getting
their study considered the migration from three over-complicated. Services are self-describing,
aspects—data, process, and user interface to make open components that can support rapid, low-cost
the applied reengineering process more complete. composition of distributed applications (Papazo-
The comparison of all aforementioned reengi- glou & Georgakopoulos, 2003). Service-Oriented
neering approaches is shown in Table 1, which is components attempt to fulfill users’ requirement.
shown below. In this comparison table, it is noted Consequently, the service providers’ responsibility
that the approach proposed by Huang et al. (2006) is mainly to design/develop the most adaptable
could present a more semantic legacy system with service processes and components for different
a higher quality of components. users. The users just enjoy the content and quality
of the provided service, but do not care about who
business component and service the true service provider is and how to acquire
component definition the service.
Some definitions of service argue that they can
Business components typically emulate a specific be implemented by components (Sprott, 2002).
business entity, such as a customer, an order, or However, in a complex environment, a service
an account. When sets of coherent components can actually include several components (Perrey
Table 1. Comparison of the reengineering approaches
Approach Sang Serrano Gall Huang

Kwan (1999)
Indicator (2002) (1999) (1995) (2006)
Proposed Time 2002 1999 1995 1997 2002
Direct Migra- Direct Migra- Indirect Indirect Migra- Indirect
Approach
tion tion Migration tion Migration
How to find objects codes codes ERD ERD ERD
How to find methods codes Codes Codes DFD DFD
Is component-based Yes Yes No No Yes
User Interface Consideration No No No No Yes
Quality
of
Object- Aggregation Consideration No No No No Yes
Oriented
Model
Inheritance Consideration No No No No Yes
Object Behavior Consideration Yes Yes Yes Yes Yes
Multi-value attributes consid-
No No No No Yes
eration
456
& Lycett, 2003). Therefore, this article considers research reengineers the legacy system to various
that the service is comprised of a number of system system components from the technical aspect,
components, which could simultaneously interact analyzes the model of the legacy system to attain
and integrate with each other. Enterprise has a the business processes from the business aspect,
tendency to keep distance from the composition of integrates the system components and the business
too many small and trivial system components and processes together, and then analyzes the service
hence has a strong reservation about the operation based on processes. Finally, according to this
and management of each other while applying the proposed integrated model, it translates the
services. For this reason, this research defines the model to the Web services composition language
service to be composed of system components to build a Web services architecture. The follow-
from the perspective of users’ requirements. A ing sections discussed below describe each of the
service-oriented feature should be an extension five steps of this proposed approach.
of the component-oriented one, and should be
aimed at satisfying the user’s situation. Step 1: Reengineer
System Components
mIGrAtInG metHodoloGy To migrate the system to a loosely-coupled ar-

chitecture, the OO reengineering approach could
By nature, Web services are different from lega- be utilized. In the comparison table (Table 1) dis-
cy systems in terms of architecture and the degree cussed earlier, the Huang et al. (2006) approach
of coupling. To migrate the system from a tightly- could be utilized to represent more semantics of
coupled one or from a loosely-coupled one, the the legacy system with higher-quality components.
best strategy is to apply OO reengineering. It is The Huang et al. (2006) approach, as visualized
noted that the complete Web services architecture in Figure 2, is summarized in the following five
generally utilizes the Web services composition
language (WSCL) to construct the service com-
ponents. Most available WSCL are designed based Figure 1. Migration methodology
on business process. From the business perspec-
tive, most approaches available today merely
apply business process to represent the legacy
system. To this end, this article will perceive that
the business process is one of the main elements
of Web services. From the above discussion, this
research analyzes the legacy system from two
aspects: technical and business. This approach,
as visualized in Figure 1, is summarized in five
subsequent steps. Traditionally the system design-
ers use the tools and techniques such as ERD,
DFD, and user interfaces to describe the entire
architecture and then design/develop the informa-
tion system. These aforementioned models have
been applied to analyze the business process of
the legacy system and then have been reengineered
to the corresponding system components. This
457
Figure 2. Object-oriented migration methodology (Source: Huang, Hung, Yen, Li & Wu, 2006)
steps. This study extends Huang’s methodology • Extended rule 2: In step 3 of Huang’s ap-
to reengineering system components. proach, this article performs a further anal-
The first activity is to employ the model dia- ysis between the functions and external
grams of a legacy system such as DFD, ERD, and entities (or data store). When a dataflow is
system interface as different inputs. There are two connected between two functions, it actu-
extended rules associated with this step and can ally indicates the directional flow of the ar-
be discussed as follows: row from one function to another one. It
is very similar to the control flow used in
• Extended rule 1: In step 1 of Huang’s ap- EPC.
proach, the fact represented from the data
stores and external entities of a DFD can In this step, the authors regard every dataflow
be translated to classes and attributes. In as a method, especially the dataflow between one
some cases, the DFD itself may also imply function and another. Since the aforementioned
a comprehensive system. For this reason, data flow implies that the IS controls the flow of
the DFD can consequently be translated to data in the system, there should be a corresponding
be a system class. method of a system class. An illustrative example
is provided in Figure 3.
458
Figure 3. The DFD is translated into several classes
Table 2. The conversion methodology for IFBP

After the completion of this step, we can get
the output—component specifications. Steps Description DFD EPC
Step 1 DFD(Function)maptoEPC(Function)
Step 2: Analyze the Business Process Step 2 DFD(Data Flow) map to EPC(Event)
Using EPC(Control Flow) con-
Due to tight integration with the system compo- Step 3 nect between EPC(Function) and
nents in step 1, this study analyzes business pro- EPC(Event)
cess from a system perspective. This step applies Using logical symbol (AND, OR,
Step 4 XOR) combine more than one
and extends the “Information Flow for Business EPC(Event) with EPC(Function)
Process” (IFBP) (Huang & Yu, 2004) method to DFD(External Entity) map to
Step 5
analyze the business process. Table 2 illustrates EPC(Information Object)
the idea of this conversion methodology for IFBP. Step 6 DFD(Data store) map to EPC(File)
Again, the first activity is to use DFD of the DFD(‘Data Flow’ connect with ‘Data
Step 7 Store’) map to EPC(Information
legacy system as an input. There are three ex- flow)
tended rules in this step; they are discussed below
and are also shown in Figure 4:
flow is used to connect the function and
• Extended rule 1: The dataflow between
the event, and the information flow is em-
one function and another should be trans-
ployed to connect the function and the ex-
lated to the event element and the associ-
ternal entity.
ated control flow of an EPC. Further, the
• Extended rule 3: This rule is similar to the
function can be connected with various
extended rule 2. The dataflow connected
events by using the control flow.
from a data store or an external entity to
• Extended rule 2: The dataflow connect-
a function should be translated to an event
ed from a function to a Data Store or an
and a control flow. It differs from the afore-
external entity should be translated to be
mentioned extended rule 2 in the direction
an event and a control flow. The control
459
Figure 4. The extended rules for IFBP
of the control flow and the information

flow.
Figure 5. Integrating the system components and
business processes
By applying these three rules, we can then get
the output—business process (EPC).
Step 3: Integrate System Components

and Business Process
In this step, this study integrates the system com-

ponent and business process. The first activity is
to use component specification and EPC as inputs.
In this step, the authors will integrate processes
from their original sources-DFD. The dataflow in
DFD is translated to be a method of a class in step
1 and an event in step 2, so that it can be integrated
later based on the same source. Afterwards, these
aforementioned DFD can be shown in an EPC
model, which is noted as the methods of classes.
Figure 5 demonstrates the situation to integrate
the system components and business processes.
Finally, in this step, we can get the output—the
integrated EPC model.
460
Step 4: Analyze Services 6 and Situation 6, A and B are two sepa-

rate services that do not interact with each
This study considers that a service is composed other.
of system components from the perspective of • Situation 7: Two external entities input
the users who are employing the service. First different functions. As presented in Figure
of all, using the integrated EPC model as inputs, 6 and Situation 7, A and B are two sepa-
this research utilized the integrated EPC model in rate services that do not interact with each
step 3 to analyze system services from the users’ other. Finally, we can get the outputs - ser-
perspective. Enterprises using services to apply vice component specification and services-
the interaction between the external entity and oriented EPC model.
other elements in the EPC model. For simplicity,
this study analyzes all possible interactive situa- After translating the related process to service
tions between two entities in the EPC model, and components, the service components can be pre-
then classifies them into seven possible situations sented by the EPC event to repaint the EPC
introduced as follows. model, as shown in Figure 7.
• Situation 1: There is no entity in the pro- Step 5: Translate to Web Services

cess. Analyzer should use low-level DFD Standard
until it locates the interaction between the
entities. The analyzer can translate the After the EPC model has been built, the services
whole process into a service. components and the services-oriented process
• Situation 2: The external entity only in- need to be translated into a suitable Web ser-
puts information into one function. The vices standard. This study translates it to business
analyzer can translate the whole process process execution language for Web services
into a service. (BPEL4WS), which is one of the Web services
• Situation 3: Only one function inputs in-
formation to the external entity. Again, the
analyzer can translate the whole process Figure 6. Services analysis
into a service.
• Situation 4: The external entity inputs in-
formation into the process, and then the
process outputs information to another ex-
ternal entity. As presented in Figure 6 and
Situation 4, the whole process provides the
service for A and B, so the analyzer can
translate the whole process into a service.
• Situation 5: The external entity outputs
the process, and then the process inputs
information into another external entity.
As presented in Figure 6 and Situation 5,
A and B are two separate services that do
interact with each other.
• Situation 6: Two external entities input
different functions. As presented in Figure
461
Figure 7. Translation of services to process
Figure 8. System architecture

composition languages. The user applies the trans-
lated Web services components and BPEL4WS
process model to build the Web Service Archi-
tecture-based system. In this step, the research
does not evaluate which component technology
should be used to encapsulate the component, but
translates the EPC model to BPEL4WS by using
the approach of Huang and Yu (2004). Since a
services component includes many associated
system components, it is required to encapsulate
the system component, process, and interface into
a new component.
cAse study Figure 9. The system snapshot of the prototype
As shown in Figure 8, the Prototype system in-

cludes three layers (i.e., Interaction layer, Transla-
tion layer, and Repository layer). Figure 9 shows
the system snapshot of the prototype.
The users input the legacy system information
(DFD, ERD, user interface) in the prototype
system, and then the components translator ana-
lyzes this information and stores the component
into a Metadata database. The users can get the
information flow by the IFBP translator and build
the service components as well as services-ori-
ented process by using the WS analyzer.
To validate this prototype, this article presents
a case study of the reengineering of an Accounts
462
Figure 10. The DFD of accounts receivable system

of components from the original data flow of
DFD.
In Step 2, the business process of this case
study is analyzed from the DFD model of the
legacy system—Accounts Receivable System.
The output is the EPC model as shown in Figure
11.
In Step 3, the business process and system
components are integrated, and the methods of
system components are employed onto the busi-
ness process components. It is shown in Figure
12.
In Step 4, the services for the integrated
model of this case study were analyzed, and the
services-oriented architecture was consequently
built. It is shown in Figure 13. The prototype
implementation combined seven components and
eleven methods into two services, as shown in
Figure 14.
In Step 5, services components and services-
oriented processes need to be translated into a
Receivable System (ARS). By employing experts’
suitable Web services standard. The users can then
help, the ERD of the database and DFD are de-
edit some information about the services compo-
scribed in Figure 10.
nents and translate it into business process execu-
According to Step 1, reengineering system
tion language for Web services (BPEL4WS).
component, the seven reengineering components,
Figure 15 demonstrates the system snapshot of
and eleven reengineered methods are shown on
this translation and the result.
Table 3. This case study named various methods
Table 3. Components and methods of ARS
Methods Sources Objectives Type Components (class)

 Receivable Receipt Receivable Data Account Detail Insert, Update, Delete Account Detail
 Account Data Sale Month Balance Receivable Data Insert, Update, Delete Receivable Data
 Receivable Data Receivable Data Write-off Select Receivable Data
 Write-off Write-off Receivable Data Insert, Update, Delete Receivable Data
 Sale Data Sale Data Sale Month Balance Select Sale Data
 Write-off Write-off Account Age List Insert, Update, Delete Account Age List
 Export Sheet Sale Month Balance Make Check Sheet Function to Function ARS
 Receivable Data Receivable Write-off Function to Function ARS
 Export Sheet Storehouse Receivable External Entity to Function Storehouse
 Write-off Receivable Customer External Entity to Function Customer
Loan Customer Receivable External Entity to Function Customer
463
Figure 11. The EPC of accounts receivable system
Figure 12. Integrated model
464
Figure 13. Service-oriented model
Figure 14. Service-oriented model
465
Figure 15. System snapshot of translation to Web services standard
In this case study, the legacy accounts receiv- (SOA) is widely adopted as a solution to reengineer
able system (ARS) is a stand-alone system. In enterprise ISs. Using this architecture, the system
order to provide more flexibility, improve the will be more flexible and adaptable to fit to the
corporation’s productivity, and enhance the ca- dramatically-changing business environment,
pability to fit into the capricious business environ- and hence, make it easier to integrate additional
ment, the Web services architecture is chosen. By applications. Obviously, an enterprise will need
using this proposed migration methodology, the to have great synergy to establish such a system.
original seven components and eleven methods The purpose of this research is to propose a mi-
are, in fact, combined into two services. The grating solution to translate the architecture of
proposed approach uses component-based tech- the legacy system to the SOA with a systematic
nology and builds the system based on business approach. This methodology is different from tra-
process. For this reason, it is more suitable for ditional object-oriented approaches (see Table 4),
business process and management. This study which migrates the system to be services-oriented
analyzes service components architecture based without applying general objected-oriented (OO)
on user interaction in the business process, so it or functional oriented features. In this study, the
will be more closely matched with the users’ re- system architecture is implemented according to
quirements. the prototyping development discussed above.
The contribution of this research can be sum-
marized as follows. First of all, this study uses a
conclusIon systematic approach to explore the service pro-
vided by a legacy system. In specific, this research
The advancement of information technology is uses a systematic approach to extract business
rapid. It is hard for an old architecture to keep up flows from the DFD diagram of a legacy system,
with the changes in the current market. Enterprises and analyzes the service provided by the system
need to use the technology of information system from a service perspective. It is indeed a reason-
reengineering (ISR) to preserve the value of their able fit for today’s Web Service environment.
existing legacy systems. Currently, a Web services- Second, this study has tightly coupled the analyzed
based system with Service-Oriented Architecture services and the rebuilt components, which will
466
Table 4 Comparison between the proposed methodology and other similar products and methodologies
Feature Support Support

Product/ Principle different Flexibility Reusability business
methodology platform process
Utilize the screen-scraping technology;
Cape Connect getting the legacy system interface defini-
Normal Low Low No
http://www.capeclear.com/products/ tions as input and generates the appropri-
ate WSDL files
Software AG EntireX Employ wrapping technology translate
High Normal Normal No
http://www.softwareag.com/Corporate/products/entirex/default.asp legacy system to Web services
IBM CICS Transaction Gateway Delivers J2EE™ standards-based access to
Low Low Low No
http://www-306.ibm.com/software/htp/cics/ctg/ CICS applications
EXTES Xuras Utilize the wrapping technology to trans-
http://www.beacon-it.co.jp/products/pro_serv/eai/xuras/index.shtml late legacy system into Web services
Integrate business processes by mapping
GoXML Transform Server the complex data formats such as EDI,
Normal Low Low No
http://www.goxml.com/features.php SWIFT, COBOL, CSV, flat text, XML,
XBRL
Xbridge Host Data Connect Enhance the Web-services’ capabilities of
Low Low Low No
http://www.xbridgesystems.com/ the OS/390 mainframe data access product
IONA Orbix Translate program to COBRA object and
http://www.iona.com/products/orbix/ then convert to Web services
Use forward engineering technique to
define business domain and use reverse
engineering technique to understand the

Yes
function of a legacy system;
(only focus
BALES (Heuvel, Hillegersberg, & Papazoglou, 2002) use of the Component Definition Lan- Low Low Low
on functions
guage (CDL) to specify both legacy and
reference)
business objects in order to facilitate a
search for matching objects and param-
eters of Web-services with legacy objects
Use a systematic approach
to explore the service
Yes
provided by a legacy
(using ERD, DFD, user
This study system, integrate system High High High
interface to understand
component and business
the business process)
process, and generates the
appropriate WSDL files
467
be easily extracted by the legacy system. Conse- gained considerable progress in translating source
quently, this proposed approach will create a more code to information flow such as DFD (Benedusi,
compact system, which may increase the opera- Cimitile, & De Carlini, 1989; Lakhotia, 1995) or
tional efficiency. Third, the proposed methodol- translating into ERD diagram from database (Com-
ogy presented in this study can help locate the puter Associates, Inc., 2006; SyBase, Inc., 2007).
shared semantic between system components and This study placed focus on the conversion
business processes, and use a systematic method methodology, and paid limited attention to the
to integrate the system component and business techniques to develop system components, such
process tightly by carefully analyzing the process. as EJB, CORBA, or COM+. Practically, differ-
Furthermore, this method can be employed to ent composition techniques may have different
build a methodology to translate the traditional ways to implement the Web Services. To this end,
structured analysis to a corresponding service- this study lacks any discussion of the different
oriented framework. This study analyzed the DFD, implementation alternatives to design/develop
ERD, and system user interface to develop busi- the Web Services.
ness processes and system components. In addi- The proposed methodology is tailored to IS’
tion, this study analyzed service to construct a assistance to various business operations. It works
systematic approach, which can be utilized to better in situations such as having complicated
translate the model diagram of a legacy system processes and external transactions. As a result,
into a services-oriented EPC model. The specifi- it does not have sufficient capability in dealing
cations of the Web Service components can be with a simple process or some applications without
easily built in this case, which in turn can help process. Furthermore, other system programming
users to construct Web Service components more and firmware that require specific hardware can-
efficiently. In addition, the IS reengineering tech- not be easily applied with this study.
nique can be utilized to build the prototype based Future work needs to be done towards a full
on the proposed Web Services framework. To this automatic reengineering process to eliminate
end, this study built a prototype to validate this the unexpected human factors and/or possible
proposed methodology. This prototype can guide errors. Further study should also be investigated
users to reengineer the legacy system into a Web to determine the right component implementa-
Service framework. As a result, this research can tion technique. In addition, it is also required to
be a valuable reference for other future studies. enhance the participation of domain knowledge
This study adopted a Rule-Based translation experts. In the future, additional benefit may
methodology, which can be used to improve not include the development of new applications by
only the quality of the design/development, but composing these software components found in
also to enhance the correctness of the original data. the reengineering process in order to accelerate
Some ambiguous or vague wordings presented the development time.
in the original DFD can definitely influence the
quality of a translated process. It is obvious that
every phase should be properly revised, so that AcknoWledGment
the output would better fit the actual practice
performed in the businesses and/or industries. The National Science Council, Taiwan, under
In practice, some organizations can either Grant No. NSC95-2524-S-194-004-, has sup-
have incomplete DFDs or even have no DFDs to ported the work presented in this article. We
perform the system/service design. This study can- greatly appreciate its financial support and en-
not be applied to such organizations. Fortunately, couragement.
techniques such as IS reverse engineering have
468
reFerences Erickson, J., Lyytinen, K., & Siau, K. (2005).

Agile Modeling, Agile Software Development,
Basu, A., & Kumar, A. (2002). Research com- and Extreme Programming: The State of Research.
mentary: Workflow management issues in e- Journal of Database Management, 16(4), 88–100.
business. Information Systems Research, 13(1),
1–14. doi:10.1287/isre.13.1.1.94 Erl, T. (2005). Service-oriented architecture:
Concepts, technology, and design. Upper Saddle
Benedusi, P., Cimitile, A., & De Carlini, U. River, NJ: Prentice Hall.
(1989). A reverse engineering methodology to
reconstruct hierarchical data flow diagrams for Fong, J., Karlapalem, K., Li, Q., & Kwan, I. (1999).
software maintenance. Proceedings of the IEEE Methodology of schema integration for new da-
International Conference on Software Mainte- tabase applications: A practitioner‘s approach.
nance (pp. 180-189). Journal of Database Management, 10(1), 3–18.
Bouguettaya, A., Malik, Z., Rezgui, A., & Korff, L. Gall, H., Klosch, R., & Mittermeir, R. (1995).
(2006). A Scalable Middleware for Web Databases. Object-oriented re-architecturing. Proceedings of
Journal of Database Management, 17(4), 20–46. the 5th European Software Engineering Confer-
ence (ESEC ‘95).
Brereton, P., & Budgen, D. (2000). Component-
based systems: A classification of issues. IEEE Herzum, P., & Sims, O. (1998). The business
Computer, 33(11), 54–62. component approach. Proceedings of OOPSLA’98
Business Object Workshop IV.
Carroll, N. L., & Calvo, R. A. (2004, July 5).
Querying data from distributed heterogeneous Heuvel, W. V. D., Hillegersberg, J. V., & Papazo-
database systems through Web services. Proceed- glou, M. (2002). A methodology to support Web-
ings of the Tenth Australian World Wide Web services development using legacy systems. IFIP
Conference (AUSWEB 04). Conference Proceedings; Vol. 231, Proceedings
of the IFIP TC8 / WG8.1 Working Conference on
Chen, Q., Hsu, M., & Mehta, V. (2003). How Engineering Information Systems in the Internet
public conversation management integrated with Context (pp. 81-103).
local business process management. Proceed-
ings of the IEEE International Conference on Huang, S. M., Hung, S. Y., Yen, D., Li, S. H., &
E-Commerce, CEC, 2003, 199–206. doi:10.1109/ Wu, C. J. (2006). Enterprise application system
COEC.2003.1210250 reengineering: A business component approach.
Journal of Database Management, 17(3), 66–91.
Chen, Y., Zhou, L., & Zhang, D. (2006). On-
tology-Supported Web Service Composition: Huang, S. M., & Yu, F. Y. (2004). IFBP: A meth-
An Approach to Service-Oriented Knowledge odology for business process discovery based
Management in Corporate Services. Journal of on information flow. Journal of International
Database Management, 17(1), 67–84. Management, 11(3), 55–78.
Computer Associates, Inc. (2006, October 6). IBM. BEA Systems, Microsoft, SAPAG, & Siebel
AllFusion Erwin Data Modeler. Systems (2002, July 30). Business Process Ex-
ecution Language for Web Services, Version 1.1.
Curbera, F., Khalef, R., Mukhi, N., Tai, S., &
Weerawarana, S. (2003, October). The next step
in Web services. Communications of the ACM,
46(10), 29–34. doi:10.1145/944217.944234
469
Jin, Y., Urban, S. D., & Dietrich, S. W. (2006). Papazoglou, M. P., & Georgakopoulos, D. (2003).
Extending the OBJECTIVE Benchmark for Evalu- Service-oriented computing. Communications of
ation of Active Rules in a Distributed Component the ACM, 46, 25–28. doi:10.1145/944217.944233
Integration Environment. Journal of Database
Perrey, R., & Lycett, M. (2003, January). Service-
Management, 17(4), 47–69.
oriented architecture. Proceedings of the 2003
Kim, H. M., Sengupta, A., Fox, M. S., & Dalkilic, Symposium on Applications and the Internet
M. (2007). A measurement ontology generalizable Workshops (pp. 27-31).
for emerging domain applications on the semantic
Rahayu, J. W., Chang, E., Dillon, T. S., & Ta-
Web. Journal of Database Management, 18(1),
niar, D. (2000). A methodology for transforming
20–42.
inheritance relationships in an object-oriented
Kwan, I., & Li, Q. (1999). A hybrid approach conceptual model to relational tables. Informa-
to convert relational schema to object-oriented tion and Software Technology, 42(8), 571–592.
schema. International Journal of Information doi:10.1016/S0950-5849(00)00103-8
Science, 117, 201–241.
Sang, J., Follen, G., Kim, C., & Lopez, I. (2002).
Lakhotia, A. (1995, February). Wolf: A tool to Development of CORBA-based engineering ap-
recover dataflow oriented design from source plications from legacy Fortran programs. Informa-
code. Proceedings of the Fifth Annual Workshop tion and Software Technology, 44(3), 175–184.
on Systems Reengineering Technology. doi:10.1016/S0950-5849(02)00005-8
Lee, Y. W., Pipino, L., Strong, D. M., & Wang, Serrano, M. A., Montes, D. O., & Carver, D. L.
R. Y. (2004). Process-embedded data integrity. (1999). Evolutionary migration of legacy systems
Journal of Database Management, 15(1), 87–103. to an object-based distributed environment. Pro-
ceedings of the IEEE International Conference
Liu, Y., Fekete, A., & Gorton, I. (2005). Design-
on Software Maintenance (ICSM’99) (pp. 86-95).
level performance prediction of component-based
applications. IEEE Transactions on Software Sprott, D. (2002). Service-oriented process mat-
Engineering, 31(11), 928–934. doi:10.1109/ ters. CBDi Forum Newsletter.
TSE.2005.127
Stal, M. (2002). Web services: Beyond compo-
McArthur, K., Saiedian, H., & Zand, M. (2002). nent-based computing association for computing
An evaluation of the impact of component–based machinery. Communications of the ACM, 45(10),
architectures on software reusability. Informa- 71–77. doi:10.1145/570907.570934
tion and Software Technology, 44(6), 351–359.
Sun - Java EE. (2007, April 16). Enterprise Ja-
doi:10.1016/S0950-5849(02)00020-4
vaBeans Technology.
Microsoft (2007, April 16). COM: Component
Sutor, B. (2006, May 21). Open standards vs. open
object model technologies.
source: How to think about software, standards,
OMG. (1995). Common Object Request Broker and service-oriented architecture at the beginning
Architecture. of the 21st century.
Ommering, R. V. (2005). Software reuse in SyBase, Inc. (2007, April 16). Sybase Power
product populations. IEEE Transactions on Soft- Designer Redefining Enterprise Modeling.
ware Engineering, 31(7), 537–544. doi:10.1109/
TSE.2005.84
470
Vanston, M. (2005, August 21). Integrating legacy Waguespack, L., & Schiano, W. T. (2004).
systems with Web services. The Meta Group Inc. Component-based IS architecture. Information
Systems Management, 21(3), 53–60. doi:10.120
Vinoski, S. (1997, February). CORBA. Integrating
1/1078/44432.21.3.20040601/82477.8
diverse applications within distributed hetero-
geneous environments. IEEE Communications Zhao, L., & Siau, K. (2007). Information media-
Magazine, 14(2), 46–55. doi:10.1109/35.565655 tion using metamodels—An approach using XML
and common warehouse metamodel. Journal of
Vitharana, P., & Jain, H. (2000). Research issues
Database Management, 18(3), 69–82.
in testing business components. Information
& Management, 37(6), 297–309. doi:10.1016/
S0378-7206(99)00056-7
This work was previously published in Advanced Principles for Improving Database Design, Systems Modeling, and Software
Development, edited by Keng Siau and John Erickson, pp. 282-306, copyright 2009 by Information Science Reference (an
471
472
Chapter 2.9
EIS for Consumers Classification
and Support Decision Making
in a Power Utility Database
Juan Ignacio Guerrero Alonso
University of Seville, Spain
Carlos León de Mora

Félix Biscarri Triviño

Iñigo Monedero Goicoechea

Jesús Biscarri Triviño

Rocío Millán
AbstrAct proposed. This EIS implements several technolo-

gies. Data Warehousing and Data Mining are two
The increasing of the storage system capacity and technologies which can analyze automatically
the reduction of the access time have allowed the corporative databases. Integration of these two
development of new technologies which have technologies is proposed by the present work to-
afforded solutions for the automatic treatment of gether with a rule based expert system to classify
great databases. In this chapter a methodology to the utility consumption through the information
create Enterprise Information Systems which are stored in corporative databases.
capable of using all information available about
customers is proposed. As example of utilization
of this methodology, an Enterprise Information IntroductIon
System for classification of customer problems is
Enterprise Information Systems (EIS) are applica-
DOI: 10.4018/978-1-61520-625-4.ch008 tions that provide high quality services by means
EIS for Consumers Classification and Support Decision Making in a Power Utility Database
of a treatment of great volumes of information. • Non-technical losses. This type of inci-

Frequently, these processes include artificial in- dents represents, faults and/or manipula-
telligence methods or any knowledge-discovery tions on the installation that induce the to-
technology. tal or partial absence or modification of the
Enterprise Information Systems can integrate consumption on the company side. If the
any technology that helps in the information treat- company cannot control the consumption
ment, in this way turn into Integrated Systems correctly it is not possible to invoice the
consisting of several modules that work jointly utility and, therefore, an economic loss is
to solve a certain problem. produced.
The great quantity of methodologies and tech-
nologies that have appeared for EIS development, Nowadays, companies have predictive sys-
have allowed the proliferation in many markets. tems of technical losses that work with a very
This situation has provoked the diversification of low mistake percentage; because normally they
the EIS, depending on the goal that they search are based on physical and climatic calculations.
for and on how EIS comes close to it. On the contrary, the non technical losses are very
In this paper, an Enterprise Information Sys- difficult to detect and control. Normally, the more
tem that integrates knowledge to help the human common non-technical are:
experts in the making decision, called Decision
Support System (DSS) is proposed. This kind of • Anomalies. They are characterized itself
systems is very useful for the utilities distribution by breakdowns or mistakes by the com-
companies. This kind of companies has several pany installation technical personnel or by
similar characteristics. For example, the con- deterioration of the client facilities.
sumption in water, power or gas utility is hardly • Frauds. They are inadequate manipulations
controlled. The company installs measure equip- realized by the clients in their installation,
ments to register the client consumption and, in with the objective to modify for their own
some case, it adds control equipments to avoid profit the energy that is registered on the
the overloads. Normally, these equipments are meter.
property of utility company and its manipulation
without company authorization is illegal. In most of the references (see ‘Overview and
In order to show the proposed DSS generic fraud detection’ section), this detection type is
methodology, an example of its application is realized treating the client’s consumption and
showed in the case of a power utility. This DSS more characteristics, such the economic sector
example try to help in the non-technical loss clas- and the geographic location. Nevertheless, on the
sification process. corporate databases there exists a lot of informa-
Mainly, the utilities present two classes of tion that includes:
incidents:
• Client information.
• Technical losses. These losses are pro- • Contract information.
duced in distribution stage. In the power • Client facilities technical specifications.
distribution companies, they correspond • Results and commentaries realized by the
with energy losses: company inspectors and technicians.
◦ Wire warming (Joule Effect).
◦ Distribution facility blemishes. According to the company, it is possible that
◦ Natural reasons. more information exists.
473
The joint treatment of all this information al- Most of the studies realized till now use a lot
lows the detection and classification procedures of information related to the client consumption
of non-technical losses. This treatment demands and the economic activity. The present paper raises
the utilization of diverse technologies to adapt the the need to use all the available information about
procedures to the information type that is used. At the client because the utility company distribu-
the same time, it is necessary to design integration tion commercial systems have more information
methods to construct the Information System. available, not only the information relating to
In the present chapter a general revision will consumption. Due to this need, it will be necessary
be done about the Information System checking to first perform an available information study
each of the following points: and to use the inspector knowledge to determine
what information is interesting. Once decided it is
• Objective of the present paper. necessary to determine what technology is most
• The form in which companies realize the adequate to extract the knowledge necessary for
activities related with client consumption the analysis of each information type.
control. The company inspector and expert knowledge
• The methodology adopted by the includes from the structure and information contained
Information System development. in the company commercial system, up to the proce-
• The information management methods dures and devices of the different existing facilities.
that companies have. The available information in utility distribution
• The Information System basic architec- companies has similarity characteristics and it is
ture joined with the integration limits that possible to apply the same techniques to treat the
are established, followed by all module available information.
descriptions.
• Information System verification and vali-
dation methods. overvIeW on FrAud detectIon
• Presentation of experimental results
obtained. Normally, the existing references limit themselves
to the information treatment about the consump-
tion.
obJectIves There exist three precedents directly related to
the topic that is shown in this chapter:
The EIS main objective consists of the accom-
plishment of an exhaustive client analysis, using ▪ F. Biscarri et al. (Biscarri, 2008) proposed
the most available information. This EIS proposes different artificial intelligence techniques
a classification of the different clients analyzed, and statistical methods for non-technical
depending on the found incidents, in this way, it losses detection. These techniques apply
supports the inspector as a Decision Support Sys- various methods that allow the detection
tem on using the knowledge acquired to realize a of anomalous consumption patterns.
classification, depending on the obtained results. ▪ J.R. Galván et al. (Galván, 1998) proposed
To obtain this goal, the EIS must take advantage a technology based on radial basis neural
of all the client information available and apply the networks, using only the consumption evo-
inspector knowledge to this information, in search lution in monthly periods. Also it uses the
of any anomaly or fraud that could mask a problem economic sector indirectly, since is the test
in the client installation (non-technical loss). only uses agricultural clients.
474
▪ José E. Cabral (Cabral, 2004; Cabral, 2006) • Telemeasuring. This type needs that in
proposed the application of rough sets to the client installation there be placed
classify categorized attribute values. This MODEMs that communicate the measur-
one is the only found offer capable of taking ing of telematic form to the distribution
advantage of a lot of information from the company.
databases, this system works with dicretized • Manual measuring. This type needs that
information, therefore, it is necessary to a company employee must visit each and
design discretization processes in the infor- every of the registers to take the measure
mation that possesses continuous values. manually.
Another one of the technologies used often The telemeasuring equipment provides a more
for the fraud detection is the forecasting using exhaustive control of the measurements. Never-
artificial neural networks (ANN) with different theless, the manual measuring is normally taken
architectures combined with other methods like: monthly or bi-monthly, which is what normally
data mining (Wheeler, 2000), fuzzy logic (Lau, happens in clients of low tension. This complicates
2008) and temporal series (Azadeh, 2007). This the client consumption analysis, because it reduces
application has a wide representation nowadays the consumption information and it is on this that
since it is used both for energy and for finance the present paper has been tested.
(Hand, 2001; Kou, 2004) and telecommunications An activity that the inspectors and technical
(Daskalaki, 2003). staff realize of the distribution companies are de-
The forecasting load subject is much more fined by the companies by means of procedures.
extensive since it includes short (Hobbs, 1998), In these procedures there are defined actions that
medium (Gavrilas, 2001) and long (Padmakumari, the inspector or technician must do according to
1999) forecasting methods, but it is used for client the goal that they pursue.
sets and not for detecting frauds in specific certain When inspectors or technicians visit the client
client. In other articles there appears the possibil- facilities they must perform the actions that are
ity of adding climatic parameters to increase the specified in a procedure, having ended it, must
forecasting efficiency (Shakouri, 2007). register it in the company commercial system and
the results. Normally, the inspectors or technicians
have the possibility of adding a commentary or
utIlIty dIstrIbutIon comPAnIes some observation. This information provides a
great advantage on the methods and technologies
All utility distribution companies have measure that only use the client consumption information,
equipment installation and information about since they give very important information that
client consumption. Concretely, in specific normally would not be available in the numeric
case of power utilities there are three types of information on the company databases.
samples. These samples depend on installed When an inspector finds a fraud or anomaly in
tension (low, medium or high). Normally, the the client installation this must be communicated
low tension client consumption is more diffi- to the company, which should create a process that
cult to control. Power utilities have million of stores all the information relating to the fraud or
clients with low tension contracts. To control anomaly from its detection until it is corrected.
the consumption of all clients there are mainly These processes are very important in the research
two methods: because of it can be used to determine the effi-
ciency of the automatic methods.
475
Figure 1. Methodology diagram
All these characteristics are normally common adapts to the needs of the activity that is going
in all existing utilities distribution companies who to be used. These Datamarts do not allow it to
have a large number of clients. be employed with online information, but it is
updated in short predefined or even incremental
periods. Normally, the update processes also are
busIness mAnAGement realized during periods of inactivity.
InFormAtIon In addition, also periodically, to perform mas-
sive reviews on a set of clients who satisfy series
All the utility distribution companies store of conditions, so it is necessary to realize batch
great quantities of information about clients. processes that are executed on nightly schedules
Normally these type of companies need a great and initiate the necessary procedures. This one
infrastructure to support and to manage all this is the expedient main source for the companies,
information. At the same time, the infrastructure since normally they are created by inspectors who
allows the company personnel to add, to modify search a certain pattern.
or to eliminate information on the company
database. These type of processes are the most
common and are more numerous than realized metHodoloGy
on the company database.
The companies cannot interrupt these processes The methodology used for this kind of EIS must
because it might provoke an economic loss. Nev- consider the existence of two types of processes:
ertheless, it is necessary to do a series of processes research and development processes.
that need a bigger load on the database, it is mainly This methodology could be used for imple-
performed during the night or at weekends, when menting generic EIS to analysis any kind of infor-
client administrative activities are not necessary. mation. In this chapter, this methodology is used
Another solution why they choose the com- to make a system for treat available information
panies is Datamarts creation. These represent of utility distribution companies in order to make
small images of the original database so that it a classification of losses.
476
In the figure 1 the evolutionary methodology done in the extraction phase, by means of
type proposed is shown. Later, each of the phases expert validation. The verification tests
is described briefly: are designed in the knowledge extraction
stage, establishing the evaluation criteria
• Identification. In this phase, the familiar- with the company. The different technol-
ization by the vocabulary and aspects re- ogy integration tests will allow the fitting
lated to the problem subject is made. the different technology processes of the
• Knowledge extraction. This is a research different technologies, in agreement with
phase. It consists of two stages, which can the evaluation criteria established, it will
be realized in parallel: be carried out at the same time as the pro-
◦ Knowledge Acquisiton. The research- cesses of verification, since the unification
ers obtain the necessary knowledge of the different technologies for the correct
to perform the consumers’ analysis functioning of the EIS is necessary.
using the meetings knowledge ac- • Implantation and Maintenance. It is nec-
quisitions with the experts. Also, essary to make a implantation and main-
one seeks to perform a design of the tenance plans that allows to the company
group of tests and the validation cri- assimilate the new EIS.
teria for EIS.
◦ Information Review. The researchers This methodology has been created from the
familiarize themselves with the avail- existing generic methodologies of each one of the
able information on the databases, most common technologies, trying to combine the
which determine the usable informa- different activities accomplishment of develop-
tion for the solution. ment and investigation. In this type of projects in
• Information Preprocessing. In this stage the which competitive companies have been involved,
preprocessing of the available information it is necessary to establish additional processes
in the database is realized in order to facili- that allow a control investigation, establishing
tate the adaptation of the same one and to criteria that determines when a development is
allow the design and structure of the nec- necessary and when an investigation phase has
essary databases. In addition, it establishes been satisfactorily completed.
the design of preprocessing techniques that The example that is shown in this paper pres-
can be used to solve the problem. ents the application of this methodology. EIS for
• Knowledge Assimilation. It has two re- classification losses in power utilities is made.
search stages: This system could be applied to other utilities.
◦ Knowledge Formalization. It estab-
lishes the design and structure of
knowledge base. InteGrAtIon
◦ Information Modelling. Modelling
algorithms to the preprocessed infor- In general, the integration of different modules is
mation are applied. necessary to solve some questions:
• Implementation. The development and
codification process of the knowledge base • The output format of each module for inte-
and modeling process. grating must verify a set of conditions to al-
• Validation, Verification and Integration. low the utilization inside a solid integration.
The acquired knowledge validation is To solve this problem there are two options:
477
◦ The design of translation intermedi- complex, because if heuristic search

ate processes is one of the most used processes (typical in artificial intelli-
options, since it allows supporting a gence) are used it is very complicated
modular design. The disadvantage of to determine the execution time.
this module types is that in applica- ◦ Asynchronous modules. In this case it
tions that use artificial intelligence, would be necessary to establish a few
they provoke the need to modify the minimal conditions of functioning for
translation intermediate module to each of the modules, so that the tak-
integrate a new module. Another dis- ing of decisions could be done with-
advantage of this solution is the in- out needing the modules to be done
crease of the execution time. completely. Logically, the results will
◦ The design of modules that obtain the be more efficient if the modules have
results in the necessary format. This performed their jobs.
option has the main disadvantage of ◦ Sequential modules. An execution
forcing design specific modules. order is established among the differ-
• Selecting how one is going to direct ent modules. This is the most com-
the Information System execution. mon, used in artificial intelligence
Traditionally, two options existed: central- processes.
ized or distributed. With the use of artificial
intelligence technologies, modules that use These questions raise a series of limits that
the knowledge extracted by other mod- force taking certain decisions in the design and
ules to obtain new knowledge or as help in development of each module.
the decision taking can be designed. This In some proposed sample aspects, combined
raises a distributed control system at every solutions have been chosen, for example, only
module that is employed independently, in certain occasions are they going to use in-
but with the existence of an implicit rela- termediate processes (middleware) because
tion, because it is the set of all modules that normally the utilization of fixed modules will
allows the decision making process. be more useful.
• Determining the way in which the In the interaction with the user, it proposes
Information System will communicate the utilization of reports with graphs to present
with the user. In this sense it is necessary the information related to the client with a pos-
to define a series of questions: sible conclusion. This conclusion is designated
◦ How is one going to present the infor- by means of the client classifications in different
mation to the user? categories that identify the problem related to
◦ What information is one going to the client. With this information the expert can
present to the user? determine if he/she agrees or not with the clas-
◦ In what levels can the user interact? sification, performing the System fixing in case
• Determining the synchronization way of it is necessary.
the different modules: In execution, initially the sequential way has
◦ Synchronous modules. It is necessary been chosen, but thanks to the adopted structures
to perform the coordination of the and the used technologies it is possible with small
different modules according to exe- modifications to realize a System capable of work-
cution times. In artificial intelligence ing in an asynchronous form and of allowing the
problems this solution type is very work with incremental load.
478
Figure 2. Generic architecture
In this kind of projects is necessary to use the develops, determines which is the most
best applications for each technology. But, in suitable information presenting form.
some case, it will be necessary to use only one • The company infrastructure. The
application, that integrates all technologies. This Information System must be measured
application is oriented to information treatment correctly to allow the implantation in the
of great volume. Normally, this condition is as- company infrastructure.
sociated to a project that is limited to one or more
applications These conditions are the same for all utilities
distribution companies, because they use very
similar information and have closed volume of
bAsIc ArcHItecture clients.
To determine the influence of these three fac-
The Information System basic architecture is tors it is necessary to do a knowledge extraction
based mainly on: phase, selecting information parts that can contrib-
ute something in the classification and detection
• The type of information that is going to process of non-technical losses.
treat. The nature of the information to treat In this database types there exists informa-
will determine what type of technology is tion of different natures: numerical, date and
most adapted to analyze it, since the same alphanumeric. For each of them, it is necessary
methods for numerical and alphanumeric to investigate the procedure or technology most
information directly cannot be used. adapted for the extraction and treatment. In this
• The way the information is presented to respect, the typical information in the utility dis-
the user. The activities that the employee tribution companies are:
479
• Client and contract data. The realized studies on large amount of infor-
• Facility information. mation sets are not useful, due to the long time
• Consumption information. process needed to get results. In addition, the com-
• Inspector documentation and panies need many resources for the information
commentaries. maintenance normal job and new client updates.
All these reasons force the decrease of the study
Once the necessary information is identified sample, allow the extraction in inactivity periods
the procedures for the treatment and analysis (night schedule) and the execution reduction time
could be designed. of the different studies. The previous information
So the different modules that compound this extraction is used creating offline databases with
architecture are shown in Figure 2. all the necessary information, although it is not
The process begins with the different source updated, it allows the acceleration of the investi-
information existence, that in some cases, will gation and depends on the commercial databases
be necessary to apply preprocessing methods, working locally.
because of possible problems existing in the
information getting.
Information System needs user interface to ProPosed ArcHItecture
show the getting of results, this module is essential,
due to DSS Enterprise Information System need The proposed architecture for utility distribu-
some interaction form with the user. In the same tion company client analysis has the following
form, the possibility in which an administrator can modules:
modify or update the system, in some exceptional
times must exist. This interface gives access to • Data Warehousing for the treatment of
each of the modules that make part of the EIS. great volume of information available in
Each one of these modules (1, 2, 3, …) per- power utility companies databases. These
forms a treatment on the information that allows techniques improve the state and quality of
adding new data to the knowledge database. The information, verify the coherence, the in-
goal of each one is to help on the fitting phase and tegrity and several kind of errors (format
on adaptation to different problems, decreasing errors, incorrect values, etc.).
the maximum possible interaction by the admin- • Data Mining for making the power con-
istrator. This information is treated by the DSS sumption studies; the trends and ranges of
module in the decision process according to the consumption are established by means of
stored information. statistical techniques. The studies of con-
sumption ranges are carried out through
the application of a statistic study which
InFormAtIon PreProcessInG searches normal pattern behaviour, but tak-
ing into consideration a series of criterium
The System has to use the necessary information which allow the distinction of one type of
to perform the analysis. The different develop- consumptions from the other.
ment and investigation phases are intended to • Text Mining for analyzing the documenta-
determine what information is used in the analysis tion of inspections made in electrical in-
and what information has be made in order that stallation clients. Initially, this module is
the Information System uses it. made based on experience, using concept
480
Figure 3. Integration method
extraction processes on documentation and rule based expert system

inspection commentaries of client facili-
ties. These concepts are organized in sev- The Information System is mainly structured using
eral categories that identify several events as a basis a rule based expert system. Each one of
in client facilities. the rules is used to analyze a selected aspect about
• Auxiliary tools realize the follow-up of the the client; in this way there exists seven groups of
client who presents certain characteristics. rules that try to check all the available information
aspect: consumption, contract, installation, etc.
The Rule Based Expert System is used for To access quickly the information and to allow
coordinating the results of the different modules the analysis of large client volumes, Data Ware-
and to take the decision about the client’s final housing techniques are used. This system uses the
classification. The expert system uses some rule Ralph Kimball (Kimball, 2002) point of view to
sets classified in 7 different groups according create a fact table in which the analysis subject
to they function. The rules have a structure IF- objective is sorted, from it, the client contracts
THEN-ELSE. who want to analyze it is extracted. Around this
These 7 rule groups allow implementing the table work the expert system applies the rules
integration between the modules and the decision that allow clustering the client according to the
system (in this case, the expert system). As is problem the client presents.
shown in Figure 3, the modules allow extracting
knowledge, that it generates the antecedent of the data mining
rules in a dynamic way.
This EIS has 135 static rules. But client analysis The information system is used like an additional
may apply round 500 rules, adding the rules with classification system, complementing the previous
dynamic antecedent. studies that are realized, using mainly the con-
sumption anomalous pattern detection by means
of statistical techniques, neural networks and
artificial intelligence. The anomalous consump-
tion patterns are varied in range. Due to this, it
481
is necessary to establish additional criterions to but it does allow use of information in most of
allow the analysis of large quantity of information. the groups:
The Data Mining process in this EIS is limited
to the implementation of a series of statistical 1. Discretized values of the powers to enable
techniques for combined rates and trends of that within each group there is a greater
normal consumption. This module proposes the representation.
idea of a study of normal consumption patterns of 2. The study on the largest possible number of
customers. This operation can not be performed customers, it also may have increased statis-
directly. The methods applied in this EIS began tical representation in each group, although
with a pre-filter that tries to remove clients who it requires a higher processing capacity.
have allegedly anomalous consumption. After the 3. Perform different groups, so that there are
removal with the statistical study of the resulting groups that contain only 4 of those specified
sample, of customer begins. characteristics, and others that use 5 or 6 of
It is necessary to do this study on the largest these features. This way, if the most stringent
possible number of customers since the avail- statistical entity is not sufficient, it can be
ability of more number of clients in the sample compared with less restrictive for a rough
is statistically representative. estimation of client normal consumption.
There is a problem applying this method on all
clients. For example, a customer who has a pub This study also provides additional information
with a capacity exceeding 15 kW on the coast, about the behaviour of each of the groups studied,
does not consume like a warehouse or another as it can be checked whether a certain group of
pub located in interior places. Thus arises the need customers presented a seasonal consumption or
for some kind of division that provides normal an irregular consumption.
patterns of consumption for each of the desired Moreover, this study along with the text min-
characteristics. ing, which will be described in the next section,
During the various investigations and tests, provides an inherent capability of automatic ad-
it has been concluded that there are a number of aptation to the characteristics of the sample to be
key information fields for determining acceptable analyzed, because if you study as many customers
consumption standard patterns: geographical lo- as possible, this will provide information on all
cation, economic activity, billing frequency, time possible cases analysable.
discrimination and contracted power. In addition, Currently, this study must be made on virtually
due to the different patterns that may be found all existing customers in the company, and it has
necessary to establish a temporary division in the provided enough information to analyze different
information for consumption patterns: total annual, types of customers, except in some groups where
seasonal and monthly. These groups provide us there is not enough information.
with divisions within which are defined normal
conditioned patterns by the client characteristics. text mining
The customer consumption comparison who pos-
sess the relevant groups with the same features will The purpose of the Text Mining process is to ana-
provide us an idea of client normal consumption. lyze the content of comments added by different
As can be inferred, the division of a sample can inspectors or alphanumeric fields. These provide
lead to groups that have no statistical significance. useful information in analyzing the consumption
Because of this it is necessary to carry out two of the clients as they provide additional data on the
processes, not to solve the problem completely, status of the facility and the inspector observations.
482
The techniques used in this process are reduced be missing if there is not a successful conversion
to the extraction and classification of concepts. of information.
The concepts are a set of one or more words that Validation is a procedure for determining
represent an idea, event, action or real object. In whether the information system does the work
this way, a concept can be anything from a simple for which it was created in a robust and effec-
word to a full term. The classification of these tive. You have to define, first, estimated limits
concepts has been done according to the experience of efficiency to be achieved and make different
of the inspectors and fuzzy techniques to detect prototypes, refining the information system to
similarities between different forms of spelling. equal or exceed the objectives.
The classification of these terms is based on The objectives for information systems are
problems that you want to identify. Thus, this defined in terms of time cost, robustness and effi-
information is used to verify the conclusions of ciency. In general, it is preferable to obtain systems
the analysis of customers. with a temporary low-cost, high efficiency and
Not all clients have this information. In such ruggedness, looking for a good balance, because
cases, the information system can provide all a very robust system is usually less efficient and
that is additional information on the client contemporary high cost.
sumption and different incidents identified in the In the sense of efficacy, validation of this
information. information system and other systems in the util-
As the implementation result of Text Mining ity distribution subject is quite complex. There
techniques, this yields a dictionary of concepts are mainly two methods to validate the different
to be added to the knowledge base information prototypes that are created:
system. This EIS process was particularly com-
plex because the initial classification was made • Perform analysis of data from a closed set
according to the knowledge provided by experts of customers and, through inspection, to
and inspectors, and has been a large volume of determine whether there is a non-technical
work of classification and verification. loss in the client facilities. This method is
quite common in some references (Cabral,
2004; Cabral, 2006). It presents several
verIFIcAtIon And vAlIdAtIon problems:
metHodoloGy ◦ The method is very inefficient, be-
cause the inspections need time,
The checking is whether the information system because he/she must visit them one
performs the analysis without losing information by one and complete the related
and without making mistakes in this process. This documentation.
requires reviewing the results of each module ◦ The amount of clients that can be
separately and Information System as a whole. reviewed is quite small, since more
Using the EIS cyclically in cases pre-designed, clients will take longer to get results.
we can fix the system so as not to lose information This decreases the probability of find-
and calculations are performed correctly. ing a client with a non-technical loss
The loss of information is one of the most reduction by decreasing the entire
common problems in systems that are created study.
by the merger of several modules. If not properly • Using two samples of the same clients that
designed and constructed, to two modules may are extracted in different moments of time.
483
This way you can see which clients have a tion about these clients and they are analyzed in
problem or if they are correct or have not the present Enterprise Information System. As a
been visited. The procedure is to analyze result of this analysis, reports and graphs for each
the information from the oldest sample and client are obtained.
compare the results with the new sample. It has been found that this Enterprise Informa-
This comparison is an investigation pro- tion System provides a complex filtering process to
cess and we must determine whether the prevent clients who are inspected are not actually
detected non-technical losses in the most non-technical losses. In this sense, the Enterprise
recent match what the analysis concludes. Information System provides a filter that removes
Moreover, the possibility of refining the usually between 20% and 70% of clients who
system using the client wrongly classified otherwise would be a wasted expenditure, since in
(as we have information on clients who most cases they do not have non-technical losses.
have not confirmed a non-technical loss). The reports of the Enterprise Information
However, this technique presents a prob- System allow the inspectors and researchers more
lem: it is not possible to establish any con- information about the client and on the problems
clusion on the remaining clients of which presented, as are just and necessary information,
there is no information on the new sample. using graphic information on a temporary basis
(consumption, etc.). Moreover, in these reports,
In the case of this sample of Enterprise Infor- the reasons why the client has been included in a
mation System the second option described has given category can be found.
been proposed, choosing a closed set of customers
in two separate moments of time.
This method has allowed the refinement of conclusIon
the Information System in order to achieve the
correct margins classifications proposed. This paper explores the expert system investiga-
There is an additional problem, which occurs tion in the utility consumption subject, and its
in the two methods and problems in the informa- combination with other technologies to enable
tion. For example, in some cases, companies have the design and construction of an Enterprise
taken measurement of the client facilities, in such Information System.
cases and where the problem is prolonged in time, Particularly, the research focuses on a very
may cause a non-technical loss. In fact, this may little exploited area: the automation of the avail-
mask a fraud. able information analysis for anomalies or fraud
The Enterprise Information System use within classification on utility companies.
the complete project validation procedure used is In the Enterprise Information System discussed
the first method described. in this paper, the main source of complexity was
the huge amount of knowledge required and the
variety of problems that may occur in the client
exPerImentAl results analysis.
The main tasks of investigation, which made
This system analyzes the information to poten- major contributions, focus on:
tial clients identified with non-technical losses
through statistical techniques, neural networks and • Methodology design for classification
artificial intelligence in a practical application in systems with different type of available
a particular company. Then you get all informa- information.
484
• Identification and classification of knowl- Daskalaki, S., Kopanas, I., Goudara, M., &
edge necessary. Avouris, N. (2003). Data Mining for decision
• Identification and classification of the anal- support on customer insolvency in telecom-
ysis cases of utility distribution client. munications business. European Journal of Op-
• Use of different information types in the erational Research, 145, 239–255. doi:10.1016/
analysis of clients. S0377-2217(02)00532-5
• Using different integration techniques.
Galván, J. R., Elices, A., Muñoz, A., Czernichow,
• Implementation of the Enterprise
T., & Sanz-Bobi, M. A. (1998, November). Sys-
Information System.
tem For Detection Of Abnormalities and Fraud
• Design and construction of an Enterprise
In Customer Consumption. In 12th Conference
Information System that automates their
on the Electric Power Supply Industry, Pattaya,
adaptation to the different test samples.
Thailand.
• Provides a Decision Support System
(DSS) making through analysis of reports Gavrilas, M., Ciutea, I., & Tanasa, C. (2001, June).
and graphs. Medium-term load forecasting with artificial
neural network models. CIRED2001, Conference
Publication No. 482.
reFerences
Hand, J. D. (2001). Prospecting for gems in credit
Azadeh, A., Ghaderi, S. F., & Sohrabkhani, S. card data. IMA Journal of Management Mathemat-
(2007). Forecasting electrical consumption by ics, 12, 172–200. doi:10.1093/imaman/12.2.173
integration of Neural Network, time series and Hobbs, B. F., Helman, U., Jitprapaikulsarn, S.,
ANOVA. Applied Mathematics and Computation, Konda, S., & Maratukulam, D. (1998). Artificial
186, 1753–1761. doi:10.1016/j.amc.2006.08.094 neural networks for short-term energy forecasting:
Biscarri, F., Monedero, I., León, C., Guerrero, J. accuracy and economic value. Neurocomputing,
I., Biscarri, J., & Millán, R. (2008, June). A data 23, 71–84. doi:10.1016/S0925-2312(98)00072-1
mining method base don the variability of the Kimball, R., & Ross, M. (2002). The Data Ware-
customer consumption: A special application on house Toolkit: The Complete Guide to Dimensional
electric utility companies. In Proceedings of the Modeling (2nd ed.). New York: John Wiley &
Tenth International Conference on Enterprise Sons Computer Publishing.
Information Systems, Volume Artificial Intel-
ligence and Decision Support System (AIDSS) Kou, Y., Lu, C., Sinvongwattana, S., & Huang, Y.
(pp. 370-374). Barcelona, Spain. (2004). Survey of Fraud Detection Techniques.
In Proceedings of the 2004 IEEE Intenational
Cabral, J., Pinto, J., Linares, K., & Pinto, A. (in Conference on Networking, Sensing & Control,
press). Methodology for fraud detection using (pp. 21-23), Taipei, Taiwan.
rough sets. IEEE International Conference on
Granular Computing. Lau, H. C. W., Cheng, E. N. M., Lee, C. K. M.,
& Ho, G. T. S. (2008). A fuzzy logic approach to
Cabral, J., Pinto, J. O. P., Gontijo, E., & Filho, J. forecast energy consumption change in a manufac-
Reis (2004, October). Fraud Detection In Electrical turing system. Expert Systems with Applications,
Energy Consumers Using Rough Sets. In IEEE 34, 1813–1824. doi:10.1016/j.eswa.2007.02.015
International Conference on System, Man and
Cybernetics, (Vol. 4, pp. 3625-3629).
485
Padmakumari, K., Mohandas, K. P., & Thiruven- Data Warehouse: This technology proposes
gadam, S. (1999). Long term distribution demand another way to organize the information in da-
forecasting using neuro fuzzy computations. Elec- tabases. The information is specially oriented to
trical Power and Energy Systems, 21, 315–322. the subject related with the problem. This idea
doi:10.1016/S0142-0615(98)00056-8 includes information redundancy but accelerates
the queries.
Shakouri, H., Nadimi, R., & Ghaderi, F. (2008). A
Datamart: It is small image of large database.
hybrid TSK-FR model to study short-term varia-
This image is specially oriented for one specific
tions of electricity demand versus the temperature
objective. Normally, the datamart is made with
changes. Expert Systems with Applications. doi:.
data warehouse technology.
doi:10.1016/j.eswa.2007.12.058
Expert Systems: The expert systems are pro-
Szkuta, B. R., Sanabria, L. A., & Dillon, T. S. grams that it is used for specific problem solution.
(1999, August). Electricity Price Short-Term These systems are made with the knowledge of
Forecasting using artificial neural networks. experts implemented in knowledge base.
IEEE Transactions on Power Systems, 14(3). Inspector: Normally, he (she) represents a
doi:10.1109/59.780895 person that is in company staff. He (she) is a
technician that visit the client facilities and is
Wheeler, R., & Aitken, S. (2000). Multiple Algo-
capable of manipulate the measure equipment
rithms for Fraud Detection. Review Knowledge-
and has the company authorization.
Based Systems, 13, 93–99. doi:10.1016/S0950-
Low Tension: Normally it is the set of clients
7051(00)00050-2
with tension less than 1000 Volts.
MODEM: MOdulator-DEModulator. This
electronic device is used for transmission of
key terms And deFInItIons information by means of modulation of signal.
Rule Based Expert Systems: Expert systems
Client Facilities: This represents from the whose have the knowledge base implemented by
finish of utility line distribution to client home. a set of rules.
Normally, in this facilities, there are measure and Text Mining: It is a technology born under
control equipment for client consumption. Data Mining. Recently, this technology became
Company Commercial System: The com- a great investigation field. The main objective
pany needs a database with the client information. of text mining is the automatic patter extraction
This database has an interface for interaction from unstructured information. The Text Mining
with user. techniques usually use natural language process-
Data Mining: It is a set of techniques based ing (NLP) for categorizing the information of
initially on statistical techniques. Actually, the unstructured information.
present techniques are based on other investiga-
tions related to artificial intelligence. The main
objective of data mining is the automatic pattern
extraction from data.
486
487
Chapter 2.10
An ERP Adoption Model
for Midsize Businesses
Fahd Alizai
Victoria University, Australia
Stephen Burgess
Victoria University, Australia
This chapter theorizes the development of a concep- The importance of midsize businesses has been
tual ERP adoption model, applicable to midsize busi- recognised in recent decades due to their role of
nesses. The general business factors associated with creating jobs, enhancement of global economic
ERP implementation along with the corresponding activity and most importantly higher growth rate,
organisational benefits are identified. This chapter regardless of their size (Rovere, L & Lebre, R (1996);
also highlights the constraints that confront midsize Acs 1990). To increase their production capabili-
businesses whilst implementing sophisticated ap- ties, they should be vigilant towards adoption of
plications. The needs for ERP adoption can occur the latest technology (Barad, M & Gien, D. 2001)
due to an attempt to be more competitive or due as use of Information Technology (IT) could result
to an external pressure from large businesses to in an increase in innovative activities, resulting in
adopt an ERP application. The proposed conceptual improved productivity and efficiency in business
model uses a strategic approach containing; ERP operations (Correa 1994). Therefore, it is appropri-
implementation processes, stages, factors & issues ate for midsize businesses to utilize their resources
associate with ERP adoption in midsize businesses. and adopt means of automated data transfer both
This research also focuses on identification of strat- internally and externally (Caillaud 2001). Business
egies in the organisational, people and technical applications such as ERP systems could provide a
domains that could be influential for ERP adoption. better way to execute business operations in an ef-
fective, organised and sophisticated way.
DOI: 10.4018/978-1-60566-892-5.ch009
An ERP Adoption Model for Midsize Businesses
The adoption of ERP applications in a modern availability, technology understanding and human
day organisation has been described as being one of resources constraints (Rao 2000).
the most innovative developments associated with Assuming the pending shift of ERP adoption
the IT sector (Al-Mashari 2002). ERP systems can to smaller-sized business entities, this chapter
be viewed as sophisticated business applications examines the relevant literature on ERP implemen-
that integrate major functions of different depart- tation as well as highlighting the characteristics
ments (Koch 2003) as its modules allow organiza- of midsize businesses to propose an adoption
tion to improve the functionality of its business model for implementing ERP systems in that
processes (Chung 1999). Hence, ERP software business sector.
modules have the ability; once implemented, to
integrate major activities across the organisa-
tional departments using one integrated software bAckGround
solution (Koch 2003). Amoako-Gyampah (2007)
suggests that ERP systems are integrated software erP systems
developed to handle multiple corporate functions,
allowing companies to synchronise activities, ERP applications were built primarily to integrate
eliminate multiple data sources with provision- different department functions and business pro-
ing of accurate and timely information, obtaining cesses to form a collaborative view of business
better communication among different units to operations in a single IT architecture (Klaus, H.,
meet expectations and reducing cost required to Rosemann, M, Gable, G. 2000). Modern day ERP
manage incompatible legacy systems. In effect, applications are business process centric, evolved
this can greatly assist organisations to carry out to address diverse aspects of corporate business
their operations in more effective and efficient requirements. One aspect of this evolution has
ways and allow the workforce to interact and col- been the ability of ERP systems to be a replace-
laborate in an information-enabled environment. ment for IT legacy systems that were developed in
ERP systems have been developed in the last different functional areas of the business. Another
two decades to replace the common legacy and aspect of ERP success has been the importance of
Material Requirement Planning (MRP) systems such systems to integrate the supply chain so as
that have traditionally been associated with larger to facilitate information flows across all business
enterprises. As the ERP market has evolved and areas— in effect allowing the large corporation to
matured, so have the related hardware and infra- be managed in real time (Turban 2006). The man-
structure technology. The cost of ERP solutions ner in which ERP has applied industry standards
have been reduced to the point where it has now to organisational business processes has also been
become viable for the midsize business sector to recognised as a significant ERP feature (Keller &
consider ERP implementation (Aberdeen 2006). Teufel 1998)— allowing a corporation to espouse
Arguably, the implementation of ERP systems in enterprise wide best practices.
midsize businesses could be viewed simply from Given the evolving nature of ERP systems,
the perspective of applying the success factors there are different point-of-views on how to ex-
already identified for larger businesses to a dif- plore ERP implementations. One view is to focus
ferent set of smaller entities. However, midsize on ERP as a product or commodity in terms of
businesses are unlike their larger business coun- software application (Klaus, H., Rosemann, M,
terparts. They have a diverse range of separate Gable, G. 2000) where ERP modules are integra-
adoption issues that need to be considered when tors of all business processes and data under one
it comes to ERP— issues such as limited finance inclusive umbrella. ERP systems are equipped
488
with features that embrace costing, finance, sales, ERP systems to integrate business functions does
contact management, customer relationship man- provide significant tangible and intangible benefits
agement and human resources issues (Rooney et al. (Sandoe, K. Corbitt G. Boykin R. 2001). The
2000). Arguably, each application area becomes a tangible benefits include reduction in employee
central focus in order to understand and facilitate numbers and inventory stock; improvements in
the ERP implementation process. It is important productivity, order management and timely deliv-
to note that getting a desired outcome from ERP eries – all of which can lead to increased profit-
applications can be difficult task due to related ability. The intangible benefits are associated with
constraints involved in its implementation and new/improved business processes; information
customisation (Marnewick, C. & Labuschagne, supply chain visibility; process standardization
L. 2005). Therefore, it is advisable to focus on and enhanced globalization opportunities.
long term business objectives associated with ERP systems also allow an organization to
ERP implementation to understand the complex become more ‘customer centric’, where more
nature of integration processes (Boubekri 2001). accurate/up to date information about customers
A strategic approach could also be a suitable by results in enhanced customer services (Rao 2000).
addressing business needs with respect to the A study conducted by Kennerley et al. (2001) iden-
organisational, technical and people (human) tifies benefits and shortcomings of ERP systems
aspects of ERP implementation. implementation in an organisation. The benefits
across four distinct areas were evaluated relat-
Some Benefits of ERP ing to; corporate organisation, operational plant,
functional divisions and the individual employees.
Turban et al. (2006) highlighted the internal and Kennerley et al. (2001) alludes to the benefits of
external nature of systems integration associated ERP systems implementation being:
with introduction of an ERP system. The internal
integration of systems allows different functional • Improved efficiency and control
areas of business to be presented under a ‘single • An ability to rationalize inventories
umbrella’ leading to many operational benefits. • Enhanced cross border capacity and
External integration benefits promote cross- optimization
collaboration and data exchanges between a firm’s • Increased leverage opportunities with
allied partners, enhancing significant business-to- suppliers
business (B2B) information exchanges as well • Improved resource and management
as improving partner relationship management planning
(PRM).
ERP systems standardize business operations Keller & Teufel (1998) describes the standard-
and automate business functions such as pro- ization imposed on business processes as being
duction, planning, manufacturing, purchasing, another benefit of ERP system implementation.
marketing and human resources into number of Arguably, standardization may come at the ex-
operational modules. These modules are integrated pense of business process flexibility, however
with each other, forming a relationship chain and business process standardization allows industry
can provide significant benefits across the enter- best practice to be adopted by a business with the
prise (Boubekri, 2001). Indeed the selection of commensurate benefits. The level of standardiza-
an ERP module is related to the factors such as tion resulting from the adoption of best practice
business attributes, specific operational needs and standards set by ERP packages might increase
the characteristics of the company. The ability of concerns regarding competitive advantage. For
489
successful implementations, existing business pany before ERP systems should be implemented.
knowledge must be translated into application Al-Mashari (2000) also indicates that successful
knowledge by mapping existing business pro- ERP implementation is directly related to organi-
cesses with ERP package embedded processes zational preparedness. Success could be defined
and defining new processes that should fit with as a favourable results or satisfactory outcomes
both new system and organisational needs (Van- in accordance with user expectations. Outcomes
daie 2008). of ERP system projects could be evaluated on
the basis of different factors, such as technical,
ERP System Implementation effectiveness and user experience related factors
(Wei 2008). According to Rao (2000) there is a
A conceptual model for ERP implementation certain level of competence that should be achieved
(the ‘4p’ model) was proposed by (Marnewick, that reflects organizational preparedness when
C. & Labuschagne, L. 2005), addressing four it comes to ERP system adoption— these levels
fundamental aspects of implementation. The related to areas associated with technical, human
model derived its structure from the well-known and management aspects of the organisation.
marketing ‘4Ps’ (people, product, process and A number of other approaches for ERP system
performance). The ERP areas associated with adoption exist. Wilhelm et al., (2000) indicated
4Ps entities are; certain traditional information systems modelling
methods that could be used to reduce the persistent
• People as the customers that represent or- cost of ongoing ERP implementation. As ERP is
ganisational requirements/mindset, defined as integrated business software, the mod-
• Product as software modules that are to be elling required for ERP implementation should
implemented across the business, detail the aspects relating to all abstraction layers
• Process as representing the project’s in integration management. The prime objective
change management issues should be progressing from upper to lower abstrac-
• Performance that is analogous to data tion levels such as enterprise modelling to final
flows associated with business process. coding with complete existing business process
information (Monnerat, Carvalho and Campos
Every aspect in this model has a direct or in- 2008). Edward et al, (2003), (drawing from the
direct impact on ERP implementation processes. work of Esteves & Pastor (1999)) uses the system
This includes identification of organizational life cycle model to explain six different stages
requirements, customization of selected software, of ERP systems adoption— such as adoption,
the installation and subsequent operations and decision-making, acquisition, implementation,
finally the important needs of system training use & maintenance, evolution and retirement.
for personnel. All various proposed levels are ERP has become a strategic survival instrument
important for ERP system adoption, allowing for businesses using information technology to
organisations to progress through implementa- conduct their operations. ERP implementation
tion processes that require all relevant factors to requires a huge investment and greater initiative
be considered (Marnewick, C. & Labuschagne, towards engaging resources such as time, money
L. 2005). and people (Yang, Wu and Tsai 2007). The use
Another approach for ERP implementation has of multi-factor business strategies (as identified)
been alluded to by Al-Mashari (2002) who sug- has been suggested as a suitable approach for
gests that an intense effort is required to highlight adoption or upgrade of an ERP system. Accord-
business and technological requirements of a coming to Aladwani (2001), a firm needs to identify
490
the various organizational, technical and people availability of adequate resources such as time,
strategies that could be used with the introduction money and skills to run business operations (Barad
of ERP systems. The organizational strategies et. al 2001). For instance, the literature suggests
include proper project management, recognition that the following affects the decision making
of organizational structure and business ideology, process of introducing latest IT applications in
change strategy development and deployment, midsize businesses: lack of resources, availability
appropriate managerial style and available com- of accurate information, lack of skilled labour
munication mechanisms. The technical strategies and management’s ability to adopt new change
address the technology challenges of ERP installa- (Rovere, L. & Lebre, R. 1996). An ERP imple-
tion and include gaining a thorough understanding mentation with a more strategic focus should be
of systems configuration, hardware complexity, of greater importance to senior management that
the capabilities of technical staff to handle pending has firm control over its IT operations. This would
challenges; and access to sufficient resources (time enhance management’s supervision of ERP, result-
and cost associated factors). The people strate- ing in better performance (along with improved
gies associate with ERP systems implementation operational and strategic control) (Ragowsky and
include the ability to identify and manage staff Gefen 2008). Another more strategic issue faced
attitudes towards change, inclusion and involve- by midsize businesses include their continuous
ment of all staff in the implementation process growth requirement and consistency to update
as well as an appropriate ERP training regime. their technological level to meet with the existing
Aladwani (2001) suggest that these strategies have technology standards (Rauof, 1998). The future
significant importance in the ERP implementation growth of midsize businesses depends upon the
process with adherence and use of these strategies use of advanced technologies for enhancement
reducing the likelihood of project failure. of their production capabilities. Use of the latest
technology can help enhance the production ca-
ERP in Midsize Business pabilities by producing good quality products at
cheaper cost and efficient delivery to its customers
ERP systems have been historically associated (Barad et. al 2001).
with implementation projects in large businesses. It has been suggested that information tech-
However; there has been a recent trend for midsize nology in general has created opportunities for
business to also consider adopting ERP systems. midsize businesses to be more competitive in
In this research an organization with 200-500 the marketplace (Rovere, L & Lebre, R (1996).
employees and/or an annual turnover of less than However, midsize enterprises, because of their
US$75 million is defined as a midsize business limited available resources can find it difficult
(Gefen et al. 2005, Yates 2004, APEC 2003, to improve IT support services (such as increas-
Duxbury et al. 2002). ing the number of educated IT professionals on
Midsize businesses are considered to be the staff and/or expanding their IT departments).
backbone of a country’s economy and play a vital In the midsize business arena several internal
role in economic development. They create job and external factors also can govern technol-
opportunities, accelerate economic revival and ogy adoption behaviour. Kennerley et al. (2001)
support industry to boost up economic progress identified internal factors such as lack of training
(Pramukti 2003). Midsize businesses are also vul- and insufficient information/documentation about
nerable and exposed to threats due to their size and IT systems as being problematic; external factors
operability (Sarbutts 2003). The risks associated were associated with the level of support provided
with midsize businesses could be related to the by implementation professionals and also the
491
nature of on-going technology upgrades. Accord- becoming a necessity in order to maintain relations
ing to Rao (2000), an ERP solution is expensive with larger enterprises (Rao 2000).
and some midsize companies may not be able to
afford them. Given this observation, Rao (2000) section summary
further indicates that information integration can
be a major motivational factor for midsize busi- Enterprise Resource Planning (ERP) has changed
nesses to implement ERP systems, allowing them the way of doing business by re-engineering and
to approach a level of business flexibility similar redesigning business processes in accordance with
to large enterprises. There is some suggestion that standardised business operations. The literature
smaller type businesses are unaware of the advan- highlights the main objective of having an ERP
tages of ERP technology and how the technology application implemented is to obtain and facilitate
has become necessary for global interaction— an best practice in the business operations. Evidence
issue that, if not addressed, may eventually push also suggests that midsize business play a vital
these businesses out of the market (BRW, 2002). role toward collective productivity of a nation.
In terms of successful ERP adoption, Lee Midsize Business often lacks leadership, strategic
(2000) highlights the concerns of small manu- vision and they mainly focus on day to day opera-
facturers, finding that the benefits associated tions. Implementing a new information system in
with ERP software had yet to be derived. Accord- a midsize business can be a cumbersome process
ing to Alison (2002), ERP systems users are not as there is not much information available for
technological experts and ERP software tends to these businesses to decide what could be a bet-
be less than user-friendly because of challenging ter solution for them. Midsize business are also
interfaces— these findings potentially posing a tend to be influenced by number of factors while
significant user training issue for resource limited selecting an information systems; these could
midsize businesses. With respect to ERP systems be relate to lack of resources such as knowledge
adoption several criteria have been proposed for and skill, availability of time and money. Midsize
small and midsize business to select and implement businesses have also adopted a cautions approach
an appropriate ERP system solution that includes towards ERP applications due to the lack of in-
affordability, supplier knowledge, local support, formation with strategic direction and associated
technical upgradeability and the availability of risks involved in ERP implementation.
the latest technology (Rao 2000).
According to Saccomano (2003) the initial
target market for ERP vendors was big companies erP ImPlementAtIon Issues
that could afford solutions costing millions of And mIdsIZe busIness
dollars at project start up. In recent times, many
multinational companies have restricted their There is considerable amount of evidence sug-
operations to partnering only those midsize com- gesting that companies face problems while
panies that are using compatible ERP software. implementing ERP applications. Millions of
Hence, it becomes essential for many midsize dollars are spent every year to purchase and
companies to adjust their business model and implement ERP products with problems relating
adopt ERP software that is compatible with the to customisation and resulting in over budgeted
large enterprises with which they deal (Rao 2000). delayed implementation (Martin 1998). The nature
Thus, midsize enterprises are increasingly finding of problems faced during ERP implementation is
themselves attracted to ERP solutions and their quite abnormal comparative to implementation of
associated benefits. Additionally, ERP systems are other IT products (Parr et. al. 2000). Some of the
492
important aspects in relation to ERP implementa- ERP Vendors & Midsize Businesses
tion are discussed below;
The benefits from ERP applications have been
erP Adoption realised by small manufacturing concerns (Lee
2000). As indicated earlier, the initial target market
Mostly large businesses have already adopted for ERP vendors was large enterprises that could af-
ERP applications to meet with their growing busi- ford to implement a costly application. Later, when
ness needs. Midsize businesses found themselves large enterprise market for ERP systems dried up
attracted to these applications due to their cost by the year 2000 (Saccomano 2003) ERP vendors
effectiveness and collaborative requirements to started focusing on the mid-market bracket. ERP
do business with larger enterprises (Klaus, H., application developing companies such as, SAP
Rosemann, M, Gable, G. 2000). and PeopleSoft have sought to increase their market
Some of the growth factors for ERP in mid- share in the mid-market bracket by boosting their of-
market bracket (midsize businesses) includes; con- fers and developing business specific applications.
tinuous industrialisation and its reliance on small PeopleSoft have also offered database, storage and
& midsize business, adoption of new technologies hardware bind options along with customer support
such as client server and availability of small & features and introducing self service portal with
medium business centric ERP applications and so features to access system availability, account, bill-
forth (Rao 2000). There is a general understanding ing and invoicing visibility (Ferguson, et al. 2004).
that ERP implementation is an expensive process ERP companies are offering extensive support
and midsize businesses cannot afford it but this with the help of their service partners (outsource
does not mean midsize businesses do not need to companies) in relation to business application strat-
have ERP applications. Information integration egy, implementation integration and optimisation
could be one of the major triggering points for services and so forth. Small and midsize business
midsize businesses to implement ERP applications centric packaging is another strategy adopted by
and achieve high levels of business flexibility with some of the ERP vendors to capture major market
their larger counterparts (Rao 2000). share such as, SAP Business by Design; an on
Another important aspect of ERP implementa- demand business solution for midsize businesses.
tion is to understand business needs and customise
ERP products to mould application(s) according ERP Implementation in
to existing business processes or altering busi- Midsize Business
ness function in accordance with ERP standards.
Research indicates that customisation in ERP It is not necessary that small & midsize businesses
application increases the risk of failure and cost should go for ‘high brand’and costly ERP products.
of the project significantly increases comparative They could consider other cheaper alternate ERP
to none customised implementation (Wilhelm et solutions that could serve their business needs (Lee
al. 2000). Higher levels of dissatisfaction amongst 2000). It is also important to note that selecting
ERP application users have been observed due an appropriate solution for small and midsize
to customisation and BPR (business process re- businesses can be difficult; depending upon their
engineering) related issues, impacting mainly on existing information technology management and
cost and duration of the project. ERP vendors have business needs (Wilhelm et al. 2000).
also admitted that generally a customer spends Rao (2000) presented criteria for midsize busi-
more to implement than to buy the software itself nesses to select an appropriate ERP application
(Wilhelm et al. 2000). consisting of following five points;
493
• Product affordability: A decision should There are some strategies suggested by Alad-
be made according to the affordability of a wani (2001) for improving ERP implementation
product and its price. processes. These strategies could be categorized
• Knowledge about supplier: An experi- as;
enced supplier should be selected with a
deep understanding of ERP implementa- • Organisational strategies comprising of
tion issues. project management, organizational struc-
• Domestic support: ERP applications are ture, change strategy development and
highly sophisticated requiring greater de- deployment, managerial style, ideology,
gree of hands on knowledge and expertise; communication and so forth.
it would be beneficial to choose a supplier • Technical strategies contain technical as-
who provides domestic/local support. pects of ERP implementation such as, ERP
• Technically upgradeable: Product with installation, configuration, complexity and
upgradeability features should be selected capable technical staff to handle the com-
that will allow the company to upgrade plexities, time - cost factors and so forth.
applications with changes in technology. • People strategies including management
A contract should be established with the and staff related issues towards change
vendor to provide annual upgrade software management, training and level of staff en-
support. gagement in a project.
• Latest Technology: An easily imple-
mentable product should be selected with As mentioned earlier, Marnewick et al., (2005)
user friendly interface with capability to ERP conceptual model that comprises of four
adopt any future modifications. It would components. These components are derived from
be better if product is designed on object marketing 4P’s model (People, Product, Process,
oriented technology and GUI interface. and Performance) and maps on ERP components;
People as Customers mindset, Product as Software,
According to Wilhelm et al., (2000), ERP Process as Change Management and Performance
affordability could be increased by reducing the as Process flow. Change management strategies
cost of implementation and increasing the user are vital to promote steps necessary for adapt-
acceptance. Certain modelling approaches could ability to change. Therefore, it is important to
be used to reduce the cost of implementation such identify factors that influence ERP user acceptance
as; (Wilhelm et al., 2000) (Bueno and Salmeron 2008). Similarly, every other
component also has an impact on ERP implemen-
• Use a reference model to select best prac- tation process (direct or indirect), starting from
tice case for implementation. identification of organisational requirements to
• Modelling techniques to be used while doc- customisation of software, installation to make
umenting requirement definition details. software operational and training for successful
• To make the business logic more under- adoption. These levels have significant importance
standable, the system requirements should and every organisation has to go through them.
be documented with help of conceptual
modelling methods.
• A conceptual model should be used as
starting point for system automation, con-
figuration and customisation (if required).
494
Barriers to Implement ERP manner. Communication strategies could be useful

in Midsize Business to educate prospective users about benefits of ERP
applications. In many cases ERP projects fails due
The adaptability of ERP applications is one of to lack of communication and if problems are ad-
the major constraints faced by businesses imple- dressed appropriately, positive benefits could be
menting ERP applications. Mostly ERP users entertained (Al-Mashari et. al., 2000).
are not their experts and they do not desire to be
due to the complex and non user friendly nature erP Implementation models
of these applications (Alison 2002). To increase
the adaptability of ERP applications, vendors Some researchers have categorised ERP imple-
introduced different integration techniques such mentation into stages and tried to standardize the
as Enterprise Application Integration (EAI) and processes for successful implementation. Bancroft
also improved application design and modules (1996), Ross (1998), Markus and Tanis (2000) and
to facilitate implementation handling capabilities Parr et al. (2000) proposed models of ERP imple-
of business, regardless of its size (Alison, 2002). mentation to obtain much deeper understanding
Gable (1999) identified some of the potential bar- of implementation processes and the purposed
riers that could cause implementation hazards to models could be used as an initiating point to
midsize businesses; create a similar model for midsize business.
• Lack of resources and less control over 1. Bancroft et al., (1998) developed a model
business operations. as result of a comprehensive study carried
• Managers/owner might be more influential out on ERP implementation in three anony-
towards strategic policy making issues and mous multinational companies and with
could make a biased choice consultations of 20 ERP practitioners. This
• The decision maker’s background could be model consists of five phases including; four
less or none technical and could result in pre-implementation phases (‘focus’, ‘as is’,
lesser understanding of technology and its ‘to be’, ‘construction and testing’) and one
implications on the business actual implementation phase (‘go live’). This
• Business might try to resolve sophisticat- model covers all major ERP implementation
ed technical issue with less technological activities, starting from ‘focus’ to ‘go live’
understanding. and is briefly described below;
◦ The Planning (focus) phase con-
Aladwani (2001) indicated some crucial issues sists of initial project activities such
in relation to ERP implementation that includes as, formation of steering commit-
mainly the possible resistance from staff toward tee, project team selection, project
adaptability of the product. If staff considers ERP guide development and project plan
applications as threat to their job, they will develop creation.
negative attitude towards it. The ERP literature ◦ The Analysis (as is) phase consists of
does not provide sufficient help to cope with this business process analysis, initial ERP
problem and it should be considered as a major system installation, business process
threat to ERP implementation. mapping on ERP functions and proj-
To overcome the possible resistance (to change) ect team training etc.
problem, management could engage and com- ◦ The Design (to be) phase includes,
municate with its employees in a more effective high level and detailed designing for
495
user acceptance, interactive proto- implementation of ERP systems in large size

typing with constant communication businesses. Markus et al. (2000) specified
with ERP users. four major phases in the implementation life
◦ The Construction (construction & cycle; Chartering, Project, Shakedown and
testing) phase consists of comprehen- Onwards & upwards.
sive configuration development, pop- ◦ The Chartering phase starts before
ulation of real data in test instance, Bancroft et al.’s (1998) focus and
interfaces building and testing, cre- Ross’ (1998) design phases. It com-
ation and testing of reports, system prises of decisions that lead to finan-
and user testing. cial approval of an ERP project and
◦ The Actual implementation phase (go it includes development of a business
live) includes, network building, in- case, package selection, identifica-
stallation of desktops and organising tion for the project team, budget and
the user training and support. schedule approval and so on.
2. Ross (1998) presented another model after ◦ The Project phase is similar to Ross’
analysing ERP implementations at 15 case implementation phase and it covers
study large organisations. This model com- all Bancroft’s model phases except
prises of five phases; Design, implementa- focus (‘as is’, ‘to be’, ‘construction &
tion, stabilization, continuous improvement testing’ and ‘actual implementation’
and transformation. phases). In this phase system config-
◦ The Design phase (which could be uration and rollout occurs and major
rephrased as planning) includes criti- activities such as, software configu-
cal guidelines and decisions made to- ration, system integration, testing,
wards ERP implementation. data conversion, training and roll-out
◦ The Implementation phase includes takes place.
several phases of Bancroft et al.’s ◦ The Shakedown phase refers to the
(1998) model such as; ‘as is’, ‘to be’, period when system is beginning to
‘construction & testing’ and actual operate normally by removing all
implementation (‘go live’). glitches and implementing standards.
◦ The Stabilisation phase comes af- ◦ The Onward and upwards phase is a
ter cut-over (final sign off) and if combination of Ross’ (1998) continu-
problems identified are fixed, conse- ous improvement and stabilization
quently improves the organisational phases. This phase refers to continu-
performance. ing maintenance, user support, up-
◦ The Continuous improvement phase grade or enhancements required by
includes any functionality added to ERP system and focuses on any fur-
the system. ther system extensions.
◦ Finally, the Transformation phase 4. Parr et al. (2000) Project Phase model (PPM)
covers achievement of maximum synthesizes previous models (Bancroft et
system flexibility up to organisational al. (1998), Ross (1998), Markus and Tanis
boundaries (system’s operability on (2000)) and includes the planning and post
every organisational level). implementation stages. The focus of this
3. Markus et al’s (2000) theory concentrates on model is on project implementation & factors
sequences of activities that lead to successful that influence a successful implementation at
496
each phase. Parr et al. (2000) indicated that additional details for user ac-
for an organisation it is important to have ceptance. It also includes inter-
significant amount of knowledge regarding active prototyping through con-
unsuccessful projects and an experienced stant communication with users.
“Champion” should be appointed with ▪ The Configuration & testing
well defined responsibilities. One large sub-phase includes, develop-
project should be partitioned into several ment of comprehensive configu-
sub-projects that can be identified as vanilla ration, real data population in
implementation. The PPM model consists test instance, building and test-
of three major phases; planning, project and ing interfaces, writing and test-
enhancement. ing reports and finally system
◦ The Planning phase comprises of se- and user testing.
lecting an ERP application, formation ▪ The Installation sub-phase in-
of steering committee, project scope cludes building networks, in-
determination and broad implemen- stallation of desktops & manag-
tation approach, selection of project ing user training and support.
team and determination of resources. (Last four sub-phases are simi-
◦ The Project phase includes a range of lar to the phases described in
activities from identification of ERP Bancroft et al. (1998) model)
modules to installation and cut-over. ◦ The Enhancement phase comprises
As the prime focus of this model is on of stages of system repair, extension
implementation, therefore, this phase and transformation and it may extend
has been divided into five sub-phases: over number of years. This phase en-
set-up, re-engineering, design, con- capsulates the Ross (1998)’s continu-
figuration and testing, installation. ous improvement and stabilization
▪ Setup comprises of project team phases and Markus et al. (2000) on-
selection and structuring with wards and upwards phases.
suitable mix of technical and
business expertise. The team(s) section summary
integration and reporting pro-
cesses are established and guid- ERP implementation has been described as unique
ing principles are established or and different from other software implementations
re-affirmed. due to its strategic impact over business. There
▪ Re-engineering comprises of are number of attempts made to produce an ef-
analysis of current business pro- fective model, providing an appropriate strategic
cess and to determine the level direction for large enterprises while implementing
of process engineering required. sophisticated business applications.
This phase also includes, instal- Based on existing research work, sequence of
lation of ERP application, map- events outlined as stages could be represented as
ping of business processes on follows, (See Figure 1)
ERP functions and training the
project teams.
▪ The Design sub-phase includes
high level designing with
497
Figure 1. Sequence of events outlined as stages
solutIons And post production application maintenance (Barad et.

recommendAtIons al 2001, Rao 2000, Gable 1999, Rovere et. al 1996).
Thus, the midsize business environment (with its
It cannot be assumed that ERP implementation limitations) is an important governing aspect of
by midsize business can directly use the existing research that is associated with ERP adoption and
ERP implementation frameworks that have tradi- needs to be part of a conceptual working model.
tionally been used during application assessment, Companies adopt precautionary measures
implementation and evaluation processes in large while implementing ERP applications and attempt
enterprises (Rao 2000). to mitigate associated risks. It is suggested that
The adoption of technology by midsize busi- midsize businesses should adopt an implementable
nesses tends to be influenced by a number of asso- strategy with proper planning and should resolve
ciated factors. These factors could be summarized related problems to increase project success rate.
as a lack of experience in adopting new technology Taylor (1999) discussed solution to nine chal-
and its implementation, access to decision making lenges faced by small and midsize business while
information and availability of general resources implementing an ERP application;
(i.e. skill, time & money). Midsize businesses also
face a number of other challenges during ERP • Scalable software that could meet future
implementation such as, selection of an effective IT growth requirements.
solution, cost of implementation and customisation, • Finding a best way to implement solutions
staff training, business process standardisation and with minimum cost.
498
• Realistic and achievable expectations from adoption by midsize businesses and their large
the application and implementation. business counterparts. It is important to adopt
• The correct level of resources should be al- a collaborative approach based upon existing
located to achieve maximum outcomes. research work to provide a road map (in form
• Reduce possible staff resistance by over- of ERP implementation models for large scale
coming the fear of change by consistent businesses) and also the strategic approach across
communication and staff engagement. organizational, technical and people domains and
• Mapping out key business processes to a their resource limitations for midsize businesses.
negotiable point where software could be This forms a strong base for the proposed ERP
implemented easily. adoption model described below.
• Data conversation (data reformation from
one application to another) should be per- An erP Adoption model
formed appropriately. for midsize business
• Avoid taking short cuts or quick fixes.
• Technical and hands on training must be A need to investigate ERP implementation is-
provided. sues in relation to their applicability in midsize
business is apparent. The literature outlines
Different research approaches have also been implementation models and strategies for large
used to examine and identify factors that are enterprises to have successful ERP implemen-
critical for successful implementation of ERP tation. This existing knowledge base could be
applications. For example, ERP implementation beneficial to develop a strategic ERP adoption
models (Bancroft (1996), Ross (1998), Markus and model for midsize businesses that should provide
Tanis (1999), Parr et al. (2000)) identify factors a workable solution for their ERP implementa-
associated with ERP implementation stages and tions. In the past, much of ERP research was
the degree of importance of each factor to every described as ‘factor research’ that mainly focused
implementation stage, as do the traditional system upon identifying factors or variables critical to
approach (Edward et al 2003) and marketing de- ERP implementation. More recent research fo-
rived 4Ps model (Marnewick, C. & Labuschagne, cus has been on processes that helps understand
L. 2005). Arguably these methods are reliant on ‘how’ an implementation takes place (Aladwani,
resource intense activities that are necessary in 2001). To take advantage of both perspectives, it
larger and change resistant organizations. Another is important to focus on an integrated approach to
approach to ERP research is to focus on business have a better understanding about issues relating
strategies that allow an understanding of ERP to ERP implementation. The link between factors
implementation as business progresses from one and stages is crucial to analyse the importance
implementation stage into another (Aladwani of different factors with the change in each
2001). The three core strategies (organisational, stage during ERP implementation (Markus et al.
technical and people strategies) could be crucial 2000). This will help to assess what factors are
for an organization to adopt ERP application and affecting which process during certain periods
indeed these three are tangibly identifiable within of time and what impact is seen on the process
the midsize business environment and less prob- itself. Parr et al’s (2000) Project phase model and
lematical to investigate if the study was to focus Markus et. als’ (2000) process theory are useful
purely on business processes. tools to conduct the factor impact analysis while
Clearly, there is a difference between issues developing an ERP adoption model for midsize
that need to be considered while examining ERP businesses.
499
The major focus of this research is to develop technical and people strategies identified by Al-
an ERP adoption model for midsize businesses dwani (2001). It also includes the midsize busi-
by critically evaluating the strategic factors and ness specific issues (Barad et. al 2001, Rao 2000,
issues with respect to different stages of imple- Gable 1999, Rovere et. al 1996) identification and
mentation. Given the various resource limita- their management to mitigate any risks associated
tions associated with midsize businesses and the with them. Thus this model adopts an integrated
potential challenges of ERP systems adoption, approach of identifying factors critical to ERP
this study is important in focusing on specific implementation along with the processes crucial
business sector (midsize) as a basis for proposing to every stage of ERP implementation.
a model. The resultant model will contribute to Table 1 provides a detailed view of issues in rela-
an increased understanding of implementation tion to implementation stages and strategies during
processes, factors, strategies and issues in relation the implementation process. The model is designed
to midsize business, enabling them to determine to identify key factors associated with ERP imple-
appropriate solution in accordance with their mentation processes. The intent is to adopt a best
operational needs. practice theoretical base approach by encapsulating
Figure 2 provides a ‘bird’s eye’ view of the existing literature to propose a strategic ERP adoption
complex relationship that exists between the model specifically designed to facilitate the needs of
project implementation phases and strategies midsize businesses. There is an enormous amount of
(organisational, people, technical) with issues research that has been conducted in relation to ERP
relating to midsize business. implementation in large enterprises; that is used to
This model is developed by identifying the help identify many of the ERP implementation issues
ERP implementation stages; defined in Parr et faced by midsize businesses. However, we are also
al.’s (2000) Project Phase model (PPM) (also introducing those factors that relate specifically to
presented byBancroft et al. (1998),Ross (1998), midsize businesses because they are different from
andMarkus and Tanis (2000)ERP implementation large size businesses. Midsize businesses are stra-
models) and the three major strategies impacting tegically fragile and economically less stable with
ERP implementation including organisational, limited operability.
Figure 2. ERP adoption model for midsize businesses
500
Table 1. ERP adoption model for midsize businesses – Detailed diagram

Factors
Stages
Activities Midsize Business
Bancroft et al.
Bancroft et al. (1998), Organisational People Technical (Barad et. al 2001, Rao,
(1998), Parr et
Parr et al. (2000) Aladwani (2001) Aladwani (2001) Aladwani (2001) 2000, Gable 1999, Rov-
al. (2000)
ere 1996)
* Business & Technol-
* Change strategies ogy Issues;
development; * Strategic Manage-
Pre-Planning
* Risk Manage- ment Issues;
ment; * Criteria of Selecting
an IS;
* ERP Application * Change strategies
Selection, development; * Training strate- * Accurate Informa-
* Project Scope determi- * Project manage- gies; * Time & Cost of tion;
Planning
nation, ment; * Change Manage- implementation * Limited Resources
* Project team selection, * Risk Manage- ment; (Time, Budget);
* Resource determination ment;
* Team Structure &
* Organizational
integration,
resources,
* Guiding principles, * Staff attitude to * ERP complexity;
* Organisational
Setup & * Business process change; * In house expertise; * Limited Resources
structure;
Re-engineer analysis, * Management * Cost of implemen- (Budget, Skill);
* Managerial style;
* Installation of ERP app, attitude; tation
* Organisational
* BP mapping,
Ideology;
* Team training
* High level designing * Organizational
* ERP complexity;
* Additional details resources;
* Staff Involve- * In house expertise; * Business & Technol-
System Design for user acceptance * Communication &
ment; * Cost of implemen- ogy Issues;
* Interactive prototyping * Coordination;
tation
* User Communication * Risk Monitoring;
* Comprehensive con-
figuration, * ERP instal-
* Information
* Real time data in Test lation aspects;
Configuration System Function * Staff Involve- * Limited Resources
instance, * In house expertise;
& Testing * Communication & ment; (Budget, Skill);
* Build test interfaces, * Cost of implemen-
* Coordination;
* Write & test reports, tation
* System & User testing
* Building Network, * Change strate- * Staff attitude to * Business & Technol-
* ERP implemen-
Installation & * Desktop installation, gies (Update); change (Update); ogy Issues (Update);
tation issues (Up-
Go live * User training, * Risk Management * Management at- * Strategic Management
date);
* System Support (Update); titude (Update); Issues (Update);
Now we shall discuss the proposed composition the interrelationship between these two modules
of the model and the importance of the strategic and to suggest activities, strategies and tasks to
mixture being proposed. execute the project efficiently. Midsize businesses
often lack leadership and strategic vision and they
the model mainly focus on day to day operations. Midsize
businesses also tend to be influenced by number of
The model is divided into two major dimensions, factors while selecting an information system and
ERP implementation stages and factors impact- are often limited by their lack of knowledge and
ing implementation, which are represented as a skill. Our model will provide midsize business a
matrix in Table 1. The objective is to underline broader picture of issues that they could encounter
501
during the ERP implementation processes and will and updated if necessary. A project manage-
assist them to have a controlled implementation. ment plan should be developed to scope the
project activities, project tasks should be
ERP Implementation Stages scheduled and resources should be identified
and allocated (including time and money).
As shown earlier, different researchers have identi- Accurate and timely information is very
fied planning, set up, engineering, system design, important for midsize businesses to execute
configuration, testing and installation as separate project plans in accordance with their desired
stages but we have consolidated these to align in expectations. Therefore it is important that
accordance with midsize implementation. Markus information should be accurate during entire
et al. (2000) identified ‘chartering’ as a crucial planning process.
stage that contains decision making processes 3. Setup & Re-engineer: To execute the project
leading up to selection of an ERP application. This effectively, it is important to structure the
has been reflected as a pre-planning phase in our project team with the correct mix of technical
model to highlight the need for activities that are and business professionals. As midsize busi-
important for decision making processes leading nesses lack resources, it is crucial for them to
up to selection of a suitable ERP application. decide whether they need to hire or acquire
the necessary skills. Midsize businesses
1. Pre-Planning: It is important for midsize should identify and reassess their available
businesses to perform comprehensive pre- resources (in-house expertise and money)
planning analysis of their existing financial to structure the project team according to
and operational performance indicators. At the standard required for ERP implementa-
the organisational level; strategic planning tion. The cost of implementation could be
for projects becomes vital when risks are significantly high if there is a need of cus-
high and resources are limited. All important tomisation in the application. Therefore, the
decisions leading to financial approval, the organisation’s ideology should be examined
development of a business case, gathering ap- to assess the staff and management’s attitude
propriate business, technical & architectural to change before taking any decisions. The
information should be obtained and shared guiding principles of the project should be
with the appropriate people for an informed identified and a business case analysis should
decision. Midsize businesses should assess be completed to underline the expectations.
the operational significance and collective The ERP application should be installed in
business benefits of the proposed applica- the development environment and business
tion before making any judgements. Change process mapping should take place with
management and risk management plans gap analysis. Internal team training should
should be developed to underline areas that occur to equip existing organisational staff
should be considered during implementation. with the appropriate skill levels. For midsize
2. Planning: This is the first official stage of businesses it would usually be wise to have
the project in which initial project activities right mix of in-house and third party technical
should be performed, such as the identifica- expertise to avoid any surprises in the ‘post
tion of key stakeholders and formation of a go-live’ phase.
governing body and project team selection 4. System Design: This is an important
(including hiring new staff). Change and risk stage in which higher level design should
management strategies should be revisited be completed and approved. Extensive
502
communication and coordination is required be documented, including change and risk

to address organisational expectations and management strategies, management of staff
users should be engaged consistently during attitudes to change, ERP implementation,
development process. Details in relation to business technology issues and so forth.
user acceptance should be captured and docu- The system support should be ongoing to
mented. Staff and management’s attitude to perform post production glitch analysis.
change should be examined and the change
management plan should be updated to cater
for resistance to change. ERP applications Future reseArcH
are complex in nature, therefore, associated
risks should be analysed and addressed by developing and refining an
developing a suitable risk mitigation plan. erP Adoption model
An initial application interactive prototype
should be completed to demonstrate applica- In order to refine the ERP adoption model for
tion functionality. This functionality should midsize businesses a research study containing
also be compared with the midsize business quantitative and qualitative data collection stages
expectations to ensure that it is addressing is proposed (Leedy et al. 1997). The data collection
the business and technology needs. focus in this study will primarily be based upon
5. Configuration & Testing: Once the inter- Australian midsize businesses; however it is also
active prototype is completed, its compre- applicable to midsize businesses around the world.
hensive configuration should be executed in Stage one of the methodology roadmap will use
accordance with the requirements identified surveys to provide an understanding of what is
in the design document. Real data should happening in midsize business market in relation
be populated in test instances for system to ERP adoption. The second stage of roadmap
testing, test interfaces should be developed utilises case studies that will help to further refine
and reports should be documented and tested the adoption model by determining the important
accordingly. During the entire testing pro- reasons as to way midsize businesses should adopt
cess, staff should be engaged and extensive ERP applications.
communication should be conducted at the
organisational level. Information system Stage One— Industry Survey
functions should be assessed and prospective
change should be coordinated. The project Stage one of research methodology explores the
budgetary estimates should be assessed and implementation stages with reference to overall
existing staff skill levels should be reas- strategies that are important while implementing
sessed. System and user testing should be an ERP application and also the midsize business
completed in this stage. specific issues. In this stage, activities to be per-
6. Installation & Go live: In this stage, all post formed in each implementation stage, impact of
testing activities should be executed, such as relating strategies and midsize business relating
building the production environment, build- issues, problems and benefits will be explored. The
ing the network (if required) and desktop survey instrument will concentrate on around 200
installation (if required). User training should midsize companies using ERP applications along
be completed and the system should go live with 200 non ERP using companies in Australia in
in the production environment. The lessons an endeavour to determine the various operational
learned from the implementation should & implementation factors, strategic issues associ-
503
ated with ERP implementation. The questionnaire A total of fifteen to eighteen businesses will be
will investigate the level of usage of ERP systems interviewed and they will be selected depending
in mid-market and specially examine; the strate- upon responses received and their ERP implemen-
gies that led to smooth implementation of ERP tation results. The results of second iteration of the
applications; selection of business solutions for model will be presented to ERP system experts to
a business type; issues and impact of proposed ascertain technological and managerial implica-
strategies during implementation process; fac- tions. The feedback gauged from ERP experts will
tors affecting the implementation processes and confirm the industry findings and potentially also
sequence of activities performed in each imple- identify new or undocumented strategies that are
mentation stage. directly related to the area of research.
The outcome of this survey will be used to The data received from different sources will
generate a first iteration or revised version of ERP be evaluated and result will be used in further
adoption model. refinements of ERP adoption model. The resultant
ERP adoption model derived from this two stage
Stage Two— Industry Case methodology will provide a road map for midsize
Studies & Expert Panel businesses, incorporating strategies that should
be considered in accordance with their situation,
The stage two of investigative methodology will background, financial situation and applicability
consist of multiple case studies. The respondents to implement ERP applications successfully.
of industry survey (stage one) will be asked to
state their interest in a follow up interview. The
purpose of this exercise is to conduct a qualitative conclusIon
analysis to examine each case in depth in order to
understand the individual experiences of midsize This chapter examines the impact of ERP imple-
businesses. This will provide an opportunity to mentation on midsize businesses by discussing
obtain first hand knowledge about organisation factors with reference to strategies and processes
and their understanding of issues in relation to that are important for ERP implementation in
ERP implementation in midsize businesses. The midsize businesses. There are many constraints
strategies adopted by these businesses to over- being identified in relation to ERP implementation,
come issues relating to their implementation of especially when there is a need of their customi-
ERP applications will also be discussed. This sation. Sometimes businesses need to customise
data collection stage will help to identify specific these applications to add or delete features to serve
strategies that these businesses have adopted to their business needs, hence it would be beneficial
help them address certain issues and problematic to identify business requirements and scope the
situations. The case studies will be categorised project objective before its initiation. The chapter
into three levels; also examines the nature of midsize businesses
and argues that ERP adoption is likely to be an
• Midsize businesses that have performed important consideration for these businesses in
ERP implementation, the near future. Midsize businesses are mainly
• Midsize companies that tried to implement dependent upon many internal (organisational)
an ERP application but ‘rolled back’ and and external (wider economic) influential factors.
• Midsize businesses that have not tried Arguably, some internal or external factors might
but considering ERP solution for their force midsize businesses to adopt ERP applica-
problems tions, not only to make them more competitive
504
but also due to pressures associated with their sational, people, organisational). The qualitative
larger counterparts. Therefore, it is important to stage of the roadmap involves the capture of
underline factors that could impact midsize busi- individual experiences of midsize businesses
nesses during sophisticated application adoption adoption through the case study method in order to
processes. gain first hand knowledge about the organisations
Implementation of an ERP system is different and the ERP strategies they implemented. Each
from any other software application due to its stage of the roadmap will allow the progressive
impact on business operations and requirement development and refinement of the proposed ERP
to facilitate business needs. There have been a adoption model.
number of attempts made by different research-
ers to produce an effective framework, enabling
businesses to have a strategic direction for ERP reFerences
implementation. ERP implementation models
were developed specifically to focus on the Aberdeen Group. (2006). ERP in the mid-market.
identification of large enterprise implementation Boston: Aberdeen Group, Inc.
requirements and activities/ stages that are crucial Al-Mashari, M. (2002). ERP Systems: A research
for their implementation. It would be beneficial to agenda. Industrial Management & Data Systems,
utilise existing knowledge base while developing 165–170. doi:10.1108/02635570210421354
a strategic model for ERP adoption in midsize
businesses. This model should focus on strategic Al-Mashari, M., & Zairi, M. (2000). Informa-
issues faced by midsize businesses providing tion and business process equality: The case of
guidelines that could help to mitigate associated SAP R/3 implementation. Electronic Journal on
risks to ERP implementation. Information Systems in Developing Countries, 2.
ERP adoption is also discussed from aspect Aladwani, A. M. (2001). Change management
of number of different implementation meth- strategies for successful ERP implementation.
odologies that are traditional, process focussed, Business Process Management Journal, 7(3),
marketing enabled, strategy or process oriented. 266–275. doi:10.1108/14637150110392764
It is argued that a strategic approach focusing on
organisational, technical and people area along- Alison, C. (2002, Dec). Works management.
side with midsize business factor analysis would HortonKir by, 55(12), 30–33.
be desirable while outlining the activities to be
Amoako-Gyampah, K. (2007). Perceived useful-
performed in each implementation stage. Hence, a
ness, user involvement and behavioural inten-
conceptual ERP adoption model would include a
tion: an empirical study of ERP implementation.
strategic approach to investigate and deliver a road
Computers in Human Behavior, 23, 1232–1248.
map to a limited midsize business environment.
doi:10.1016/j.chb.2004.12.002
To test and reform the proposed ERP adoption
model, a methodological roadmap is documented APEC Profile of SMEs. (2003). What is an SME?
that embraced quantitative and qualitative data Definitions and statistical issues. Journal of En-
collection stages and will be tested on Australian terprising Culture, 11(3), 173–183. doi:10.1142/
midsize business market. The quantitative stage S021849580300010X
will be associated with the capture of business
Bancroft, N., Seip, H., & Sprengel, A. (1998).
characteristics that detail the scope of ERP systems
Implementing SAP R/3 (2nd ed.). Greenwich:
adoption and identification of salient aspects of
Manning Publications.
strategy across the three areas of interest (organi-
505
Barad, M., & Gien, D. (2001). Linking improve- Ferguson, R.B. (2004). ERP targets the midmarket.
ment models to manufacturing strategies – A meth- eWeek, 21(6), 41.
odology for SMEs and other enterprises. Inter-
Gable, G., & Stewart, G. (1999). SAP R/3 imple-
national Journal of Production Research, 39(12),
mentation issues for small to medium enterprises.
2675–2695. doi:10.1080/002075400110051824
Americas Conference on Information Systems,
Boubekri, N. (2001). Technology Enablers for August 13-15, Milwaukee, WI.
supply chain management. Integrated Manu-
Gefen, D., & Ragowsky, A. (2005). A multi-level
facturing Systems, 12(6), 394–399. doi:10.1108/
approach to measuring the benefits of an ERP sys-
EUM0000000006104
tem in manufacturing firms. Information Systems
BRW. (2002, November). Fast 100 Issue. Cited by Management Journal, 22(1), 18–25. doi:10.1201
Business Technologies for SMEs, October 2003, /1078/44912.22.1.20051201/85735.3
Conference at Sydney.
Keller, G., & Teufel, T. (1998). SAP R/3, process
Bueno, S., & Salmeron, J. (2008). TAM-based oriented implementation. Harlow: Addison-
success modelling in ERP. Interacting with Wesley.
Computers, 20, 515–523. doi:10.1016/j.in-
Kennerley, M., & Neely, A. (2001). Enterprise
tcom.2008.08.003
resource planning: Analysing the impact. Inte-
Caillaud, E., & Passemard, C. (2001). CIM grated Manufacturing Systems, 12(2), 103–113.
and virtual enterprises: A case study in a doi:10.1108/09576060110384299
SME. International Journal of Computer
Klaus, H., Rosemann, M., & Gable, G. G. (2000).
Integrated Manufacturing, 14(2), 168–174.
What is ERP? Information Systems Frontiers,
doi:10.1080/09511920150216288
2(2), 141–162. doi:10.1023/A:1026543906354
Chung, S. H., & Synder, C. A. (1999). ERP ini-
Koch, C. (2008). The ABC of ERP. Enterprise
tiation- A historical perspective. Americas Con-
Resource Planning Research Center. Retrieved
ference on Information Systems, August 13-15,
from http://www.cio.com/research/erp/edit/erp-
Milwaukee, WI, 1999
basics.html
Correa, C. (1994). cited by Aladwani (2001).
Lee, T. T. (2000). Apt ERP alternatives. New
Change management strategies for successful ERP
Straits Times-Management Times.
implementation. Business Process Management
Journal, 7(3), 266–275. Leedy, P. D. (1997). Practical research – Plan-
ning and design (6th ed.). NJ: Prentice-Hall, Inc.
Duxbury, L., Decady, Y., & Tse, A. (2002). Adop-
tion and use of computer technology in Canadian Markus, M. L., Axline, S., Petrie, D., & Tanis, C.
small businesses: A comparative study. In Manag- (2000). Learning from adopters’ experiences with
ing Information Technology in Small Business: ERP: problems encountered and success achieved.
Challenges & Solutions (pp. 22-23). Hershey, Journal of Information Technology, 15, 245–265.
PA: Information Science Publishing. doi:10.1080/02683960010008944
Edward, W. N., Bernroider, N., & Tang, K. H. Markus, M. L., & Tanis, C. (2000). In R.W. Zmud
(2003). A preliminary empirical study of the dif- (Ed.), The enterprise systems experience – From
fusion of ERP systems in Austrian and British adoption to success, in framing the domains of IT
SMEs. Working Papers on Information Processing management: Projecting the future…… through
and Information Management. the past (pp. 173-207).
506
Marnewick, C., & Labuschagne, L. (2005). A Rovere, L., & Lebre, R. (1996). IT diffusion in
conceptual model for enterprise resource planning small and medium-sized enterprises: Elements
(ERP). Information Management & Computer for policy definition. Information Technology for
Security, 13(2). doi:10.1108/09685220510589325 Development, 7(4), 169–181.
Martin, M. (1998). An electronics firm will save Saccomano, A. (2003). ERP vendors consolidate.
big money by replacing six people…… not every Journal of Commerce, 4(24), 46.
company has been so lucky. Fortune, 137(2),
Sandoe, K., Corbitt, G., & Boykin, R. (2001).
149–151.
Enterprise Integration. New York: Wiley.
Monnerat, R., Carvalho, R., & Campos, R. (2008).
Sarbutts, N. (2003). Can SMEs ‘do’ CSR? A prac-
Enterprise systems modeling: The ERP5 devel-
titioner’s views of the ways small-and medium-
opment process. [Fortaleza, Ceara, Brazil]. SAC,
sized enterprises are able to manage reputation
08(March), 16–20.
through corporate social responsibility. Journal
Parr, A., & Shanks, G. A. (2000). Model of Communication Management, 7(4), 340–348.
of ERP project implementation. Journal doi:10.1108/13632540310807476
of Information Technology, 15, 289–303.
Taylor, J. (1999). Management Accounting.
doi:10.1080/02683960010009051
Turban, E., Leidner, D., Mclean, E., & Wetherbe,
Pramukti, S. (2003). Establishing synergy
J. (2006). Information Technology Manage-
between small companies and banks. JA-
ment: Transforming Organisations in the Digital
KARTA POST 06/03/2003. Accession Number:
Economy (5th ed.). New York: John Wiley & Sons.
2W81194803776, Business Source Premier
Vandaie, R. (2008). The role of organizational
Ragowsky, A., & Gefen, D. (2008). What Makes
knowledge management in successful ERP imple-
the Competitive Contribution of ERP Strategic.
mentation projects. Knowledge-Based Systems,
The Data Base for Advances in Information Sys-
21, 920–926. doi:10.1016/j.knosys.2008.04.001
tems, 39(2).
Wei, C. (2008). Evaluating the performance of
Rao, S. S. (2000). Enterprise resource planning:
an ERP system based on the knowledge of ERP
business needs and technologies. Industrial
implementation objectives. International Jour-
Management & Data Systems, 100(2), 81–88.
nal of Advanced Manufacturing Technology, 39,
doi:10.1108/02635570010286078
168–181. doi:10.1007/s00170-007-1189-3
Raouf, A. (1998). Development of operations
Wilhelm, S., & Habermann, F. (2000). Making
management in Pakistan. International Journal
ERP a success. Communications of the ACM,
of Operations & Production Management, 18(7),
43(4), 57–61. doi:10.1145/332051.332073
649–650. doi:10.1108/01443579810217602
Yang, J., Wu, C., & Tsai, C. (2007). Selection of
Rooney, C., & Bangert, C. (2000). Is an ERP Sys-
an ERP system for a construction firm in Taiwan:
tem Right for You? Adhesives Age, 43(9), 30–33.
A case study. Automation in Construction, 16,
Ross, J. W. (1998). The ERP revolution: Surviving 787–796. doi:10.1016/j.autcon.2007.02.001
versus thriving. Centre for Information Systems
Yates, I. (2004). 2004 Proved Successful for SAP
Research, Sloan School of Management.
Latin America. Caribbean Business,33(10).
This work was previously published in Enterprise Information Systems for Business Integration in SMEs: Technological, Or-
ganizational, and Social Dimensions, edited by Maria Manuela Cruz-Cunha, pp. 153-174, copyright 2010 by Business Science
Reference (an imprint of IGI Global).
507
508
Chapter 2.11
Developing and Customizing
Federated ERP Systems
Daniel Lübke
Leibniz Universität Hannover, Germany
Jorge Marx Gómez

University Oldenburg, Germany
AbstrAct The service composition is realized by a workflow

system that is also responsible for creating and
Small and Medium Enterprises (SMEs) are the managing the user interfaces and the data-flow. By
most important drivers in many economies. Due integrating only the Web services that are needed
to their flexibility and willingness to innovate they (possibly from third parties) the cost is reduced
can stand up to larger industry players. However, and the functionality can be scaled to the actual
SMEs – as every other company – need to further needs. However, not only a technical solution is
reduce costs and optimize their business in order needed but also the development process must be
to stay competitive. Larger enterprises utilize ERP tailored towards SMEs. Small companies cannot
systems and other IT support for reducing costs afford highly-skilled staff and often do not have
and time in their business processes. SMEs lack defined business processes.
behind because the introduction and maintenance
of ERP systems are too expensive, the return on
investment is achieved too late and the associ- IntroductIon
ated financial risks are too high. However, SMEs
would like to have IT support for their business. The business world is rapidly moving and
The research behind the Federated ERP System Small-to-Medium Size Enterprises (SMEs) are
(FERP) addresses the problems SMEs face with competing within this vibrant marketplace with
conventional ERP systems and offers reasonable their flexibility and ability to innovate. They are
and scalable IT support. This is done by decompos- an important part of the economy. For example,
ing the whole business logic of the ERP system according to the IfM Bonn (2008) SMEs in Ger-
into Web services, which are linked at run-time. many account for 38.3% of the overall turnover
Developing and Customizing Federated ERP Systems
Figure 1. Reference architecture of an FERP system
and employ 70.6% of all employees nationwide. In changed is worthless if no one knows what the
order to operate efficiently, SMEs need enterprise desired result is.
software, like ERP systems, for managing their Within this chapter we introduce the Feder-
business operations efficiently. However, ERP ated ERP System as a new architecture for ERP
systems impose high costs due to their expensive systems that are especially suited to SMEs. We
purchase, customizing costs and re-customizing describe the overall architectural ideas as well
costs whenever business processes are changed. as our implementation. In the second part we
Thus, business process changes that are neces- present a technique for deriving and discovering
sary to stay competitive become more costly as business processes from textual scenarios – so
before. called use cases known from the software engi-
This inevitably leads to the question how to neering domain.
make ERP systems better suited to SMEs in order
to make them more competitive in the long run.
The answer to this question is decomposed into FederAted erP systems
two parts. The first part is a new architecture for
such systems that can be introduced, operated, Problem Addressed
and maintained cheaper. The second part is en-
gaged with the question on how to come to (new) An ERP system is a standard software system
requirements for the ERP system based on the which provides functionality to integrate and
business processes. A system that can be flexibly automate the business practices associated with
509
the operation or production aspects of a company. The subsystems of the proposed architecture
The integration is based on a common data model are the following:
for all system components and extents to more
than one enterprise sector (see Robey et al., 2002; FERP Workflow System (FWfS)
Rautenstrauch et al., 2003).
However, there are some disadvantages as- The FWfS coordinates all business processes
sociated with conventional ERP systems. The which have to be described in an appropriate
main ones are: XML-based workflow language. A workflow in
this context is a plan of sequentially or parallelly
• In most cases not all of the installed com- chained functions as working steps. Each step
ponents are needed, represents an activity which leads to the creation
• high-end computer hardware is required to or utilization of business benefits. Workflows
run the system, and implicitly contain the business logic of the overall
• customization of ERP systems is very ex- system. The function types that can be contained in
pensive because product specific know-how a workflow in FERP systems are the following:
of experts is necessary.
• Model-based user interface functions, e.g.
Due to the expensive process of installation show, edit, select, control
and maintenance only large enterprises can afford • Database access functions, e.g. read, up-
complex ERP systems, which provide business date
logic for all sectors of the functional enterprise • Application tasks which are connected to
organization. Contrary to these aspects, FERP Web Service calls
systems allow the separation of local and remote
functions whereby no local resources are wasted FERP User System (FUS)
for unnecessary components. Furthermore, single
components are executable on small computers The FUS is the subsystem which implements func-
and due to decreasing complexity of the local tions for the visualization of graphical elements
system installation and maintenance costs sub- and coordinates interactions with end users. This
side, too. subsystem is able to generate user screens at run-
time. Screen descriptions, which have to comply
reference Architecture with the FERP UI standard, are transformed to an
end device-readable format, e.g. HTML in case
Figure 1 gives an overview of the reference ar- of web browsers.
chitecture of a Web Service-based FERP system.
The architecture consists of several subsystems, FERP Database System (FDS)
which are interconnected. Because one of the
main objective of an FERP system is to integrate The FDS is the subsystem which implements
business components of different vendors, all functions for the communication with the FERP
components have to comply with standards. In database. This subsystem is able to interpret XML
this approach these standards are described by structures which comply with the FERP data
using XML schema documents. In order to sepa- standard. The interface differentiates between
rate the three different layers of a typical layered two kinds of requests. Database update requests
architecture of conventional ERP systems, each contain object oriented representations of business
layer is assigned its own standard. entities as XML trees. Database read requests
510
Figure 2. Architecture of the prototype
contain X-Path or X-Query expressions specifying which are wrapped in a Web Service. The FWCS
portions of data to be extracted. In both cases the is the subsystem that provides the functionality
request parameters have to be transformed into for the invocation of Web Services. All possible
different types of request statements that vary types of FERP Web Services are specified by the
depending on the type of database management FERP WS standard. This standard contains XML
system (DBMS) that is used. Assuming the use schema definitions that describe Web Service op-
of a relational DBMS (RDBMS), the underlying erations as well as input and output messages. A
data model also has to comply with the FERP data Web Service references these types in its WSDL
standard, which means that the corresponding description. Furthermore this subsystem is able to
table structure has to reflect the XML-Schema search for Web Services, which are defined by a
specifications respectively. The java.net project unique identifier. This way it is possible that dif-
hyperjaxb21 provides a solution to generate SQL ferent Web Service providers implement the same
statements on the basis of XML schema defini- business component type as Web Service. Beside
tions. Another solution is the application of native the implementation of Web Service invocation
XML databases or XML-enabled RDBMS. and search functionality this subsystem is respon-
sible for the interpretation and consideration of
FERP Web Service Consumer non-functional parameters. Examples for those
System (FWCS) parameters are security policies, payment polices,
and Quality of Service (QoS) requirements on the
The business logic of FERP systems is encap- part of Web Service consumers.
sulated in so called FERP business components
511
Figure 3. Process model in YAWL as simplified example for the creation of a purchase order
FERP Web Service Provider System FERP Web Service Directory (FWD)
(FWPS)
The FWD provides an interface for the publica-
The FWPS is the subsystem which implements tion and the searching of FERP Web Services
functions for the provision of Web Services which based on the UDDI standard. The structure of
comply with the FERP WS Standard. The subsys- this registry leans on the FERP WS standard. In
tem includes a Web Server which is responsible for this standard Web Services are assigned to cat-
the interpretation of incoming and outgoing HTTP egories mirroring the predetermined functional
requests which in turn encapsulate SOAP requests. organization of enterprises.
The subsystem provides business components of
the FERP system as Web Services. Prototype development
A connection to the FERP Web Service Di-
rectory allows the publication of Web Services. The following paragraph briefly describes a first
Furthermore this subsystem is responsible for the implementation of the proposed reference archi-
negotiation of common communication policies tecture which is based on open source software
such as e.g. security protocols or usage fees with components. Figure 2 shows the architecture
the requesting client. of our prototype. For the implementation of the
FWfS we chose the workflow engine of the YAWL
project2. The FUS was implemented on the basis
512
Figure 4. Generation of a simple user interface for a customer record
of Apache Struts3. Our FDS is mainly based on the calculation of the total sum the Web Service
the API of the Hyperjaxb2 project which in turn returns the completed order as XML document.
uses JAXB4 and Hibernate5. jUDDI6 served as The next workflow task visualizes this XML
basis for implementation of the FWD. The FWCS document. After the user agreed the XML docu-
uses JAX-RPC7 (Java API for XML-based RPC) ment is transmitted to the FERP database system
which is provided by the SUN Developer Network which transforms it to an SQL-INSERT statement
(SDN). Our FWPS uses Apache AXIS8 as basis in the next workflow task.
for the provision of Web Services.
Figure 3 shows an example process model in user Interface Generation
YAWL
Every ERP system needs to be operated by users.
Tasks in our process definitions can be assigned In the end, they need to make decisions, retrieve
to one of the three function types: data or enter new records. While classical ERP
systems offer clients for personal computers only,
• Database communication (in figure 3 indi- now mobile devices, like handhelds and mobile
cated as DB-task) phones are emerging. Because of this situation
• End-user communication (in figure 3 indi- the Federated ERP system will face many types
cated as GUI-task) of clients. Furthermore, these clients need to be
• Web Service communication (in figure 3 easily updatable. For a simple process change it is
indicated as WS-task) not feasible to update hundreds of possibly mobile
or distributed computers. Thus the user interface
All other symbols comply with the graphi- must be managed on the server-side and must be
cal notation of YAWL. The example process platform-neutral.
model demonstrates a workflow for the creation Our approach for minimizing the effort needed
of a purchase order9. The example includes only to develop and customize the user interface is to
one Web Service call which is responsible for automatically generate the interfaces from the
the calculation of the total sum of a purchase business process descriptions. Much research has
order which consists of one or more order items. been done in the field of model-based user interface
Order items include a price and an amount. The (MB-UI), which aims to model user interfaces in
Web Service receives the whole order as XML the way program logic is modeled in UML. Re-
document without total sum. Having finished search in these fields has been going on for more
513
Figure 5. Refinement of the business process and user interface generation
than a decade. For example, (Paterno, 1999) gives strives to reduce the inherent complexity. This
an overview over the field of MB-UI. Numerous is especially important for being useful for the
design environments have been proposed as result targeted, non-expert audience.
of MB-UI. Each differs in the number and type of Because we assume the business process to be
models used (for a thorough overview the reader already modeled, the user interface is expressed by
is referred to (da Silva, 2000). However, most ap- stereotyping business functions. Four stereotypes
proaches share a common element: the task model. have been introduced:
Fortunately, this task model is easily related to
our approach: The business process model is in • Selection: The user shall select data from a
fact a task model on a very high abstraction level collection of possible choices. For example:
(see Traetteberg, 1999). Furthermore, the field Select product from a catalogue.
of MB-UI has matured. Especially insight into • Edit: The user shall edit some information
reasons for failure of some approaches has been object from the data model. For example,
beneficial for our research. Common mistakes edit order.
and problems concerning practical adoption of • Control: The user wants to explicitly in-
MB-UI techniques are listed by (Traetteberg voke some action. This is used to model
et. al., 2004): The biggest problem has been the navigational decisions. For example, “Accept
complexity of the introduced models. While order”.
complex and detailed models give the designer • User: The user has to do something by
the best level of control, such models are difficult himself, e.g. planning, comparing, etc.
to learn, time-consuming to design and hard to
maintain. Therefore, our approach particularly
514
These four actions can be attached to a busi- example, a connector for XForms – an XML
ness function and are visualized by small icons standard for describing input forms - is under
on the left-hand side. The annotated business development and generation of HTML pages is
processes are downloaded by the client software, possible as well for integration into intranet and
which generates user interfaces from these models portal applications. For further discussion on the
and sends the data and user decisions back to the topic of generation of user interfaces from EPC
server. This way, the user interface can be edited models (see Lüecke, 2005; Lübke et. al., 2006).
simply by installing new business process models
on the server. FerP as an soA Instance
For the generation of user interfaces, the data
types are used to look up matching editors. Be- Because of FERP system is based on Web Ser-
cause Web services are based on XML, the data vices and their orchestration, we can say that
types are represented by XML Schema definitions. it is a service-oriented architecture solution
XML Schema defines data types recursively: whereby all functions are available as services
Primitive types can be grouped to complex types. and unambiguously addressable. This system
Complex and primitive types can be grouped to works as follows:
new complex types and so on. Editors are created
by traversing this structure and look for matching 1. The network consists of service consuming
editors registered in the system. At least for each and service providing network nodes.
primitive type, like integers and strings, an editor 2. Each client, which provides an interface
is provided by the system. Therefore, a (possibly to an enterprise is called mandator and is
primitive) editor can be generated for each XML connected to the enterprise database.
Schema. Figure 4 shows a simple generated editor 3. The processing steps of a business process
for a customer record. are stored in the local database of a mandator
Figure 5 shows the hierarchical refinement as workflow. A workflow in this context is
of the example process with user interface ste- a plan of sequentially or in parallel chained
reotypes, and the resulting user interface using functions as working steps in the meaning
a custom editor. The client application shows the of activities which lead to the creation or
processes needing further action by the user, and utilization of business benefits.
the processes which are currently executed by 4. Finding a function within the P2P-network
someone else. This information is given on the means that a request which contains the
right hand side. function type must be send to all service
Since the user interface generation is based providing peers.
on the business process description, context in- 5. After receiving the responses to a function
formation can be given to the user. For example, type request the mandator must elect a net-
descriptions of the currently active business func- work node to be accessed.
tion can be displayed. In our prototype these are 6. A function call contains parameters as busi-
realized by giving tool-tip information. At this ness objects and other (primitive) values that
point it is even possible to integrate experience are delivered to the service providing net-
bases to facilitate the communication between work node. A business object in this context
developers, process designers and end users. is a snapshot of the enterprise database at
Since this approach is based on the process a particular time in a standardized format.
description only, it is possible to generate user Function calls can contain other function
interfaces for different target platforms. For calls.
515
7. A function returns a list of either directly they have to be performed. But even if SMEs
modified business objects or independent have documented business processes, these de-
values that are necessary for subsequent scriptions are typically not suited for software
business object updates (e.g. intermediate development or customization because technical
data). details are missing.
8. Returned business objects must be syn- Therefore, the knowledge of process partici-
chronized with the local database of the pants needs to be externalized into documented
mandator. business processes that are suited to be the basis
for software development and customization
projects because they lack sufficient detail.
busIness Process Within the FERP context, we propose a light-
modelInG In tHe FerP weight approach to elicitate the business processes
context and to generate explicit models from there. These
explicit business process models can be used to
The FERP system is targeted at SMEs. This pre-generate and develop the workflows in FERP
poses some additional challenges besides the systems that are the technical representation of
already outlined technical problems. Especially the business.
the gathering of requirements - and as their most The technique is constrained for this particular
important part business processes – has to be per- context as following:
formed before any implementation of an (F)ERP
system can start. • Easy to use: The technique must easy to use
However, in most SMEs the business processes because SME employees usually do not have
are not defined explicitly, but the organization extensive technical knowledge and no time
as a whole has tacit knowledge of the activities for learning complex techniques.
that are to be performed and the order in which • Easy to understand: When discussing results
it is important that all stakeholders can par-
Table 1.
Use Case #3: Thesis Supervisor hands out topic
Primary Actor Thesis Supervisor
Thesis Supervisor: wants to hand out topic easily and without much paperwork
Stakeholders Student: wants to receive topic quickly
Secretary: wants easy to use/read forms for completing registration
Minimal
A topic is only handed out once at a time
Guarantees
Successs Student knows topic
Guarantees Supervisor knows all needed adminstrative information of the student
Student has achieved at least 80% of credit points
Preconditions
Student has clearance from Academic Examination Office
Triggers Student wants to sign up for a topic
1. Supervisor checks whether the topic is still available or not
2. Supervisor reserves topic for student
Main Success
3. System updates list of current thesis
Scenario
4. Supervisor confirms thesis topic and student’s information to the Academic Examination Office
5. System sends confirmation to student, supervisor and Academic Examination Office
Extensions 1a If topic is not available anymore, then EXIT
516
ticipate and contribute. Therefore, the same of view of a single main actor. The most common
restrictions apply to understanding as they form for documenting use cases are tables as il-
do to usage. lustrated in table 1:
• Lightweight: The technique must quickly The table contains additional information
save time for associated employees. Not like preconditions and success conditions that
much effort may be spend as this would pose express goals and constraints from the business
undue costs and would hinder flexibility. and software point of view that are associated
• Offer basis for later development: The results with the use case.
must be usable by the customization team For our approach, we assume the use cases
later on. The smoother this transition is the to conform to the meta-model as illustrated in
better. figure 6.
The metamodel captures the properties of the
use cases as the basis tabular template. A use case consists primarily of
the main scenario that has a sequence of steps. Each
We use a use case-based approach for interview- step can be extended for operational sequences
ing the users and stakeholders in SMEs and that are not default. Each extension consists of a
documenting the results. new scenario. Each step within a scenario has an
Use Cases (Cockburn, 2005) are a technique actor, i.e. the role or person that is performing the
from the Requirements Engineering community. activity. A use case is written from the perspective
They represent possible scenarios from the point of a main actor that is primarily concerned with
Figure 6. Use case metamodel
cd: use case metamodel(uc)
*
Project usecaseset
+sets
inv: this.returnJump.
* scenario !=
+ triggers this.extensionScenario
* 1
+ preconditions +mainScenario scenario
* + extensionScenario
condition usecase 1
+minimal g.
*
+ success g 0..1
decomposition
+ condition 1
* 1..*
{ordered}
*
1 * + extended by
stakeholder Actor step 0..1 extension
+ actor
+ return jump
computer
Human Actor
system
517
the goals the use case has. The goals are repre- Because use cases conforming to the presented
sented by success guarantees. A use case starts metamodel must not contain complex control-
when the trigger applies and may be performed flow constructs, they only require a limited set of
iff all preconditions are met. workflow-patterns (van der Aalst, ter Hofstede,
Because use cases are nearly freely-written Kiepuszewski, & Barros, 2003). Following work-
text, they have the advantage that they are easily flow-patterns are needed within a use case:
comprehendible by all kinds of users and stake-
holders (Lübke, 2006). Furthermore, the tabular • Sequence: A sequence is used to order the
structure imposes a semi-formal format that can steps within a scenario,
be processed by computer programs later on. • Exclusive Choice: For attaching extensions
to a step, this type of split in the control-flow
mapping use cases to business is needed,
Processes • Simple Merge: When extension jump back,
the control-flow is merged with this type of
Use Cases are written from the perspective of merges.
the main actor only. However, a set of use cases
can represent a business process. Therefore, the However, it is necessary to join use cases. This
individual use cases need to be joined together is done by comparing the triggers, preconditions,
into one large process. and success guarantees. The use cases are ordered
The set of use cases and a use case itself can in a way that success guarantees satisfy triggers
represent a business process. The steps within and preconditions of the following use cases.
the use case can be seen as the activities of the Because a success guarantee can satisfy the
business process. The actors of the step become preconditions of more than one use case, and
the actors of the business activities. the precondition and trigger of a use case can be
Figure 7. Joining multiple use cases in EPC notation to a single EPC model
518
satisfied by more than one use case, following 5. The extension scenario is converted like the
workflow-patterns are needed: main scenario
° Extensions of extensions are handled
• Parallel Split: The control-flow of one use recursively
case is split to several use cases because more 6. Return jumps are realized with an XOR-
than one precondition can be satisfied, join
• Generalized Synchronized Merge: The pre- ° The join is introduced before the function
condition can be satisfied by one of many that is the jump target
use cases.
This algorithm is applied to every use case.
Because all business process languages In the following step, the use cases have to be
support these basic workflow-patterns, the gen- joined. Depending on different needs, several
eration can have any language as its target. We join strategies are available:
demonstrate this by generating EPCs that can be
the basis for composing the Web services of an • Large EPC: Generates a single, large EPC
FERP system. model from all use cases.
• Short EPC: Generates an EPC that has a
Generation of event-driven function for each use case. The details of
Process chains the use case are discarded and not displayed.
This type of model is well-suited for discuss-
For demonstrating our approach we chose Event- ing the ordering of use cases and the global
Driven Process Chains (EPCs). They are easy to control-flow.
understand and are therefore well-suited for fos- • Short EPC with hierarchical refinement:
tering communication between the stakeholders EPCs allow function to be detailed in
of FERP systems. other EPCs. This approach combines the
The generation of a single use case is done by advantages of the first two approaches by
applying following algorithm: generating a short EPC and placing a more
detailed EPC for the use case behind every
1. Preconditions and Triggers are converted function.
to events
° If there is more than one event, the events Common to all three approaches is the strat-
will be joined by an AND-join egy for merging the set of use cases to a single
2. Each Step of the main scenario is converted model. The events that have been generated by
to a function the algorithm above are unified, i.e. all events that
° Connected with simple okay-events have the same name are reduced to a single event.
3. Success guarantees are converted to end- Connectors for the control-flow are introduced
events accordingly. This is illustrated in figure 7.
° If there is more than one event, an AND-
split is introduced Advantages of the Use Case-
4. Extensions are introduced with an XOR- Centered Approach
split
° The extension condition becomes an By using textual use cases, it is possible to docu-
event ment and discover business processes. Due to the
use of plain, semi-formal text, they can be used
519
by all involved stakeholders. These stakeholders tive Models and Development Environments: A
do not need to learn a new notation nor do they Survey. In Palanque & Patern`o (Eds.), DSV-IS,
have difficulties while interpreting and validat- volume 1946 of Lecture Notes in Computer Sci-
ing the documented parts of business processes. ence (pp. 207–226). London: Springer.
Use Cases can guide interviews with isolated
IfM Bonn (2008). Schlüsselzahlen Deutschland
stakeholders. The generation generates a global
(Key Indicators Germany). Retrieved December
view by combining the use cases to a large busi-
28, 2008, from http://www.ifm-bonn.org/index.
ness process model. The business process model
php?id=99, 2008-12-28.
is the foundation for later ongoing development.
It can be readily used by developers. Therefore, Lübke, D., Lüecke, T., Schneider, K., & Marx
our approach satisfies the constraints outlined Gómez, J. (2006). Using Event-Driven Process
above nicely. Chains for Model-Driven Development of Busi-
ness Applications. In Nüttgens & Mendling (Eds.),
Proceedings of the XML4BPM 2006.
conclusIon And outlook
Lübke, D. (2006). Transformation of Use Cases
to EPC Models. In M. Nüttgens, F. Rump, & J.
Within this chapter we have presented the FERP
Mendling (Eds.), Proceedings of the EPK 2006.
architecture as an architecture for ERP systems
CEUR Proceedings Vol 224. http://ftp.informa-
that are well-suited for SMEs; such systems can
tik.rwth-aachen.de/Publications/CEUR-WS/
be flexibly altered, and are comparably cheap to
Vol-224/.
install and maintain. However, in order to know
what to change and what to install, business pro- Lüecke, T. (2005). Development of a Concept for
cesses need to be defined first. Within SMEs such Creating and Managing User Interfaces bound
processes need to be documented because usually to Business Processes. Master’s Thesis, Leibniz
no explicit business process documentation exists. Universität Hannover, Germany.
We proposed a use case-centered approach for
Paterno, F. (1999). Model-Based Design and
elicitating the business processes in a way that is
Evaluation of Interactive Applications. London,
comprehensibly by non-tech-savvy people.
United Kingdom: Springer-Verlag.
With the combination of these two parts, SMEs
can introduce and maintain their ERP systems Rautenstrauch, C., & Schulze, T. (2003). Infor-
and can stay competitive in the market. matik für Wirtschaftswissenschaftler und Wirt-
While the FERP architecture was developed for schaftsinformatiker, Berlin.
addressing requirements of SMEs, the architecture
Robey, D., Ross, J., & and Boudreau, M. (2002).
may be suitable for larger companies as well. The
Learning to implement enterprise systems: An
assessment to what extend the architecture can
exploratory study of the dialectics of change.
scale will be part of our future work.
Journal of Management Information Systems,
19(1), 17-46.
reFerences Trætteberg, H. (1999). Modelling Work. Workflow
and Task Modelling. In Vanderdonckt, & Puerta
Cockburn, A. (2005). Writing Effective Use Cases. (Eds.), CADUI (pp. 275-280). Kluwer.
Amsterdam: Addison-Wesley Longman.
Trætteberg, H., Molina, P. J., & Nunes, N. J. (2004).
da Silva, P. P. (2002). User Interface Declara- Making model-based UI design practical: usable
520
and open methods and tools. In Vanderdonckt, 2

http://yawlfoundation.org/
Nunes, & Rich (Eds.), Intelligent User Interfaces 3
http://sturts.apache.org/
(pp. 376–377). ACM. 4
http://jaxp.dev.java.net/
5
http://www.hibernate.org/
van der Aalst, W., ter Hofstede, A.H.M., Kie- 6
http://ws.apache.org/juddi/
puszewski, B., & Barros, A. P. (n.d.). Workflow 7
http://java.sun.com/webservices/jaxrpc/
Patterns. Journal of Distributed and Parallel 8
http://ws.apache.org/axis/
Databases, 3(14), 5-51. 9
In order to improve understandability the
process was simplified. Changes of entered
data and order items are not supported.
endnotes
1
Hyperjaxb2 – relational persistence for
JAXB objects: https://hyperjaxb2.dev.java.
net/ (last visit October 2006)
521
522
Chapter 2.12
Creation of a Process
Framework for Transitioning
to a Mobile Enterprise
Bhuvan Unhelkar
This chapter presents the creation of a process This chapter presents an approach to transitioning
framework that can be used by enterprises in to mobile enterprises. The earlier outline of this
order to transition to mobile enterprises. This approach was published by Unhelkar (2005) and it
framework facilitates adoption of mobile tech- contained three dimensions of a mobile enterprise
nologies by organizations in a strategic manner. A transition framework. Later, based on the research
mobile enterprise transition framework provides undertaken by the author, this transition framework
a process for transition that is based on the fac- was modified and extended to result in a four
tors that influence such transition. The Mobile dimensional framework. This framework is the
Enterprise Transition (MET) framework, outlined core discussion topic of this chapter. A complete
in this chapter, is based on the four dimensions of and in-depth discussion of this process framework
economy, technology, methodology, and sociol- also appears in Unhelkar (2008).
ogy. These four dimensions for MET have been Mobile technologies form the basis of the
identified based on an understanding of people, communications revolution that has resulted in
processes, and technologies. A research project elimination of physical connectivity for people,
undertaken by the author validates these four processes and things. This wireless connectivity
dimensions. has resulted in significant impact on the organiza-
tion of the business and its relationship with the
DOI: 10.4018/978-1-60566-156-8.ch006 customers. The ability of businesses and customers
Creation of a Process Framework for Transitioning to a Mobile Enterprise
to connect to each other ubiquitously-independent consIderInG tHe nAture

of time and location – using mobile technologies is oF mobIlIty In tHe
the core driver of this change. However, successful “met” FrAmeWork
changes in terms of adoption of mobile technolo-
gies and applications in an organization depend Electronic business transitions have been studied,
on a process framework. This chapter discusses amongst others, by Ginige et al (2001), Lan and Un-
a Mobile Enterprise Transitions framework for helkar (2005). However, the uniqueness of mobile
transitioning an organization to a mobile organi- technologies in terms of their impact on business
zation. The purpose of this MET framework is to has been discussed by Marmaridis and Unhelkar
provide guidance in terms of people, processes and (2005), Arunatileka and Unhelkar (2003), Godbole
technologies involved in successful transitioning and Unhelkar (2003), Lan and Unhelkar (2005),
of the enterprises. and Unhelkar (2008). These authors have focussed
A MET can be defined by extending and refin- on the specific nature of mobility as depicted in
ing an earlier definition of mobile transformation Figure 1. The inner square in Figure 1 indicates
given by Marmaridis and Unhelkar (2005) as land-based connectivity between enterprises,
“evolution of business practices through the adop- functional units and other fixed devices. This
tion of suitable mobile technologies and processes connectivity evolved from the initial centralized
resulting in pervasiveness.” This definition sug- connectivity of the mainframe, followed by the
gests that the MET will facilitate incorporation client-server connectivity and finally resulting
of mobile technologies in business processes that in the Internet connectivity (business to busi-
will result in pervasive business activities inde- ness - B2B and business to customer - B2C). The
pendent of location and time. The understanding Internet-based connectivity is further augmented
of MET, however, needs to be based on a firm by the XML (eXtensible Markup Language) to
understanding of how mobility is unique and how facilitate the Internet as a medium of computing,
it is different to land-based Internet connectivity. rather than merely as a means of communication.
However, as depicted by the outer square in Figure
1, the external wireless connectivity, by its very
nature, is between an individual and the business
or between two individuals. As correctly stressed
Figure 1. Mobility is personal in nature (based on Unhelkar, 2005)
523
by Elliott and Phillips (2004), a mobile phone is the “what”, “how” and “who” of the process. The
a far more personal device that is carried by an transition process framework thus reveals itself
individual as compared with a desktop personal into four major dimensions as shown in Figure 2
computer. (based on Unhelkar (2003a) and (2008)). These
This nature of wireless connectivity needs to dimensions of mobile transitions are, however,
be understood and incorporated in all dimensions not exclusive to each other. When applied in
of MET (this could be based on discussions such practice, they tend to overlap each other, result-
as Thai et.al (2003); the four dimensions are ing in a cohesive transition process. However,
discussed next). For example, economically, the separate understanding of each of these dimen-
cost of a mobile device has dropped and continues sions is helpful in creating the MET framework
to drop significantly making is obligatory for in the first place. Thus, this discussion starts with
businesses to consider mobility in order to access the theoretical framework and then discusses its
and serve the customer. Technically, it is essential application in practice. The four process dimen-
to consider the “individuality” of the mobile sions for a mobile enterprise transition, as shown
gadgets and their ability to be location-aware (see in Figure 2, can be understood as follows:
Adam and Katos (2005) for the resultant chal-
lenges of security and privacy of individuals that • The Economic dimension deals with the
is different to the challenges in corresponding core business driver of “why” to undertake
land-based connections). The activities and tasks the transition. Costs and competition have
of the methodological dimension should be cre- been the core business drivers for most
ated in a way that properly exploits the hand-held business decisions and they are also true in
gadgets; and socially, mobile transitions need to this case. Reduction in costs and increase
consider the impact of mobility on the socio- in competition encourage the business to
cultural fabric of the society and the correspond- undertake formal MET.
ing changing value systems like work ethics and • The Technical dimension of the mobile
social protocols. Thus, when businesses transitions transformation process considers “what”
from the land-based connectivity paradigm and technologies are used in the transforma-
incorporate wireless connectivity in their business tion, and “what” deliverables are produced
practices, they have to ensure that an individual at the end of the process. Examples of is-
is considered in all their process dimensions as
compared to when the business was organized
only around land-based workstations. Figure 2. Four dimensions of a mobile enterprise
transition framework (based on Unhelkar, 2008)
tHe Four dImensIons

oF A mobIle enterPrIse
trAnsItIon FrAmeWork
When an organization decides to incorporate

mobility in its business, it is a strategic decision
that is based on the primary question of “why” to
“mobilize”. This strategic decision is followed by
investigations into the technical, methodological
and social dimensions dealing respectively with
524
sues discussed in this dimension include undertaken by the business decision makers of the
devices/gadgets, programming, databases, organization. This decision to transition using the
networking, security and architecture of MET process framework can take the business to
mobile technologies participating in the a global mobile playing field and therefore it has
MET. to be taken carefully and seriously.
• The Methodological dimension of the pro-
cess deals primarily with the question of
“how” to – amongst other things- model tecHnIcAl dImensIon In
and design business processes, approach m-trAnsFormAtIon
methodologies and quality in software pro-
cesses and re-organize business structure This technical dimension of the MET framework is
and its relationship with customers and concerned with the question of “what” technology
employees. to use, and “what” deliverables are produced at the
• The Social dimension of the transformation end of the process. Thus, the technical dimension
process focuses on “who” is involved in and primarily includes the understanding and the ap-
influenced by the process. Typically these plication of the various mobile hardware devices
are the users of the business (e.g. custom- and gadgets, issues of GPS enabled gadgets (3G)
ers and employees) The discussion in this and wireless networking and security. The types
dimension deals with the effect of mobile of devices, their capacity and costs, and their us-
technologies on the socio-cultural aspects ability are of utmost importance in this dimension.
of people’s lives -especially the changing Mobile devices have developed well beyond the
working formats and work ethics– and or- ubiquitous mobile phone, and now include a wide
ganizational and social structures. range of devices like Personal Digital Assistants
(PDA) and wireless computers (typically laptops
The aforementioned four dimensions of MET with inbuilt wireless processors). Each device can
are now discussed in greater detail. have numerous features that need to be considered
when they are incorporated in the business pro-
cesses. For example, the ability of Wi-Fi enabled
economIc dImensIon In mobile gadgets to take photographs onsite and to
m-trAnsFormAtIon instantly transmit them to the business centre has
an invaluable application in processes that are
The question of “why” to “mobilize” provides the related; for example to insurance claims, medical
strategic reason for a business undertaking MET. emergencies and sports. Additional device related
The economic dimension considers the costs of issues include the ability of the devices to function
running the business and maps it with the cost of alone -standalone-, or with other wireless compo-
acquiring mobility. Furthermore, the economic nents; for example, some wireless PCMCIA cards
dimension is also concerned with the competition cannot be connected to the Internet and receive
and how they are putting pressure on the busi- SMS messages simultaneously.
ness. Furthermore, the economic dimension also Functionality of mobile devices are further
investigates the potential effect of mobility on augmented by mobile networks, which play a
customers as well as employees of the organiza- significant role in terms of the device’s abili-
tion in terms of providing efficient service to the ties to browse, locate and transmit information.
customers by conducting the business efficiently. Some examples of technological challenges in
Thus, “why” to “mobilize” is a strategic decision networking include the incorporation of wireless
525
broadband services, creation of local hotspots for content security) is a vital issue in the technical
services, integration of devices and their software dimension of MET.
with the existing organizational infrastructure, Databases have evolved to content manage-
combination of WLAN (Wireless Local Area ment systems (CMS) that are capable of storing
Network) and WWAN (Wireless Wide Area Net- audios, videos, photos, graphs and charts and many
work), satellite communications and even simple other content formats that could not be stored in a
issues like user access through VPN or dialup standard relational database. Technical dimension
connections. Mobile networking influences the of MET has to, therefore, consider issues like the
breadth of coverage (i.e. area being covered) and type and nature of content, location of contents
the depth of coverage (the amount of information influencing the speed and security of downloads,
being transmitted/received) based on the available duration of content before they are replaced or
bandwidth, speed of transmission, as well as its upgraded, synchronization of contents between the
reliability. At the individual level, the ability of mobile devices and back-end servers, and so on.
infrared and Bluetooth enabled devices and the A mobile process is made up of a diverse
potentials offered by Radio Frequency Identifica- range of heterogeneous mobile infrastructures
tion tags (RFID) are also increasingly playing an including the hardware, operating system, mes-
important role in mobile enterprise transitions. saging systems and databases –which all need to
Wireless connectivity has further provided be integrated to provide mobile functionalities.
opportunities for handheld devices to use com- Although most business applications today are
puting power of other handheld and stationery web-enabled, they also need to be integrated
devices, leading to the creation of wireless Grids further with mobile applications in order to make
(McKnight and Howison (2004), and Unhelkar them mobile-web-enabled applications. Using
(2004). Wireless grids facilitate creation of virtual SMS and MMS, creation of WiFi hotspots and
processors that can be used to deliver far more incorporation of seamless wireless broadband
functionalities than is evident today. For example, through a single ISP are some practical options
wireless grids can provide sufficient computing in such integrations. Some of the integration chal-
power for Mobile Internet Agents (as discussed lenges for mobile infrastructures include issues
by Subramanium et. al (2004)), enabling them to such as the movement of the users, mobile nodes,
perform advanced functions like searching and fluctuating demands on the infrastructure (notably
comparing different products and services on networks and databases), changing security needs
handheld devices. (depending on the type of information required)
Security (as discussed by Godbole (2003) in and reliability (ability to restart transaction from
the context of the Internet) is the next important any point in time).
part of the technical dimension. Mobility provides Finally, usability can also be considered as an
greater opportunity for unscrupulous ‘tapping’ important technical issue (in addition to being a
into the networks, siphoning off data, information sociological issue) in MET–especially as more
and even identities of the users as compared with and more functionality is being provided on small
land-based network. And because many mobile- screen mobile devices. Usability considerations,
enabled business processes (e.g. in healthcare so vividly depicted by Constantine and Lockwood
domain with ambulances) depend heavily on the (1999), need to be extended further when it comes
security of the mobile networks, protection of to mobile devices to ensure information is pro-
sensitive corporate data during transmission to vided succinctly to the users. Wireless Markup
and from a mobile devices (channel security and Language (WML) can play an important part in
526
removing unnecessary elements from the display fill out his/her time sheets would change with the
of information on mobile devices. advent of mobility in the organization. Or the way
in which a customer enquires about her balance in
a banking environment will change depending on
metHodoloGIcAl dImensIon mobility in the process. Organizational processes,
In m-trAnsFormAtIon primarily dealing with other businesses, but also
related to organizational structure and internal
The methodological dimension of MET primarily management, undergo change when mobility is
deals with the business processes, which can be incorporated in the business. Finally, as discussed
understood as the “manner in which a business by S’duk and Unhelkar (2005), an entire busi-
carries out its activities and tasks”. These activi- ness sector made up of a group or collaboration
ties and tasks deal with both external parties and of businesses could start interacting with each
internally with its employees and management. other through wireless applications and networks,
The modelling of these business processes is resulting in a need to reengineer processes that
complicated by the fact that the users (especially criss-cross an industrial sector. This may result in
customers) have an expanding array of different the need for industrial process reengineering (IPR).
types and models of mobile gadgets available to For example, with mobile connectivity, an airline
them. Not only do organizations making the transi- and a car hire company may need to change their
tion doesn’t have much control on these devices, business processes together to inform an individual
but they are also faced with the ever increasing of changes to flight timings on her hired mobile
expectations by the users that the m-businesses car dashboard; or a hospital, an insurance company
will support their specific gadgets. and a company providing road side assistance may
Whenever a business transitions to a mobile introduce new business processes that facilitates
business, three possibilities emerge with respect their collaboration to provide immediate care and
to its business processes: support to a mobile subscriber who may have met
with an accident. Similarly, advances in wireless
• Firstly, existing business processes are grids further increase the opportunities for an
re-engineered to incorporate mobility; the industrial segment (or a dynamic group) to create
re-engineered processes will be impacted entirely new business processes. For example,
by mobility in terms of location and time the creation of on-the-spot dynamic customer
independence for the users groups at shopping malls and sports venues was
• Secondly, totally new business processes not feasible without the mobile connectivity;
are brought in; as a result to the introduc- with this mobile connectivity businesses are
tion of a mobile technology, and would able to target a group of customers dynamically.
otherwise have not existed and Extending Kanter’s (2003) argument further, for
• Thirdly, some redundant and/or irrelevant future service growth mobile business processes
business processes are dropped and they need to continuously keep the user’s context in
do not make sense with the incorporation mind to be able to fully exploit the dynamicity
of mobility. offered by mobility.
Deshpande et.al (2004), in the Device Indepen-
Each of the aforementioned process changes dent Web Engineering workshop, describe how,
impact the individual user, as well as the organiza- with continuing advances in mobile communica-
tion and even the business sector as a whole. For tion capabilities, and dramatic reduction in costs of
example, the manner in which an employee would mobile devices, the demand for Web access from
527
many different types of mobile devices has now discussed by Devereaus and Johansen (1994), now
gone up substantially. Incorporation of mobile needs to be further considered in the context of a
Internet access has resulted in a paradigm shift global-mobile society. This is so because when a
in the way businesses and individuals tend to use business enterprise undergoes a MET, it affects
the Internet. For example, customers, instead of the socio-cultural aspects of both individuals and
making a call using a mobile phone may use its the groups they are part of. For example, should
Internet capabilities to locate good consumer deals an individual answer the mobile phone provided
or find convenient service locations for their needs. by the company during his or her private time?
Mobile Internet has also resulted in evolution of Or should the mobile service provider be allowed
the customer relationship management (CRM) to send unsolicited promotional messages to its
systems into what can be called mobile CRM (m- subscribers? While mobility enables people to be
CRM) systems (Arunatileka and Unhelkar, 2003). productive anytime and anywhere, the need to
Reengineering of processes also needs to con- separate personal from official work and respon-
sider the devices that will facilitate the business sibilities are far greater in today’s mobile society.
processes. For example, for corporate solutions, we These social issues related to work ethics and
need to consider whether there are devices already behaviour in an Internet-based society, already
deployed? And if so, can they be reused or are new studied by Ranjbar and Unhelkar (2003), need
devices required to be provided for the business to be further extended and applied in the context
processes? Finally, there is also a need to keep of mobility in an MET.
the software development methodologies in mind The social advantages resulting from mobile
when the MET results in changes to the information technology infrastructures that may impact MET.
systems of the transitioning organization. These For example, in developing nations (and even
methodologies can help in standardization, formal otherwise), the infrastructure costs associated with
modelling of requirements, user-centered designs, land-based Internet connectivity are far higher
understanding the technology of mobile network than corresponding costs of setting up wireless
architectures, and issues in integrating mobile computing infrastructure. This results in an op-
software with the existing software (Godbole, portunity to reach people en masse through the
2003). These technical process considerations will relatively cheap mobile devices, and conducting
result in the implementation of good quality mobile business activities with them. The resulting change
applications and, as a result, would improve the in the social landscape of a country is an extremely
overall quality of service offered by the business interesting phenomena that needs to be studied
to its customers and users. under this third dimension of MET.
socIAl dImensIon In consIderInG mobIle busIness

m-trAnsFormAtIon Internet usAGe And levels
In tHe “met” FrAmeWork
Mobility has had a significant impact on the
quality of life of individuals and the society in Mobile Internet usage has been discussed by Un-
which they live. While the location-aware mo- helkar (2003b). This usage describes the increas-
bile connectivity has dramatically increased the ingly complex application of mobile internet by
ability of individuals to communicate, it has also businesses in informative, transactive, operative
produced challenges in terms of privacy and new and collaborative manners. The mapping of this
social protocols. The effect of globalization, as usage with the four dimensions of the MET frame-
528
Table 1. Mobility considerations in mobile Internet usage by business (based on Unhelkar, 2008)
M-Informative M-Transactive M-Operative M-Collaborative

Economic Costs; Profit Costs; Inventory; Trust;
(Why) Nuisance; Sharing; Alliance formation; Legal mandates
Technology Device Availability Networking – Internet connectivity; Intranet; Extranet; Group- Portals; Groupware;
(What) and Access Reliability; Security; ware; Standards and inter-
Reliability operability;
Methodology Personal Process Business Process Engineering (BPR) Organizational Policies; Industrial Process
(How) BPR Reengineering; Business
Collaboration
Sociology Privacy; Access Security, Confidence; Convenience; Security, Trust, Workplace Security, Trust, socio-
(Who) Regulations; Ethics; cultural issues
work is summarised in Table 1. The subsequent as dealing with the third-party credit/debit card
sections describe that usage further from the point organizations that facilitate online payments. Tech-
of view of understanding the MET. nically, devices such as a wireless enabled PDA
or laptop with necessary access and security on
mobile Informative layer the mobile network are required to conduct these
transactions. Methodologically, it will usually be
This is the usage of the Internet to merely provide a transaction with a user “known” to the business
information. As such, this is a one-way transfer (for example, a registered bank user conducting
of information requiring no security. Example of account transactions), requiring businesses to in-
mobile informative usage include broadcasting corporate upfront registration processes in dealing
of schedules, information of products, services with their customers. Socially, the convenience of
or places, and capitalizing on the common Short conducting two-way transactions will affect the
Messaging Service (SMS) feature of mobile individuals, but so will their confidence in the
gadgets. Technically availability and access to a systems being used for such transactions.
mobile device is important to enable provision
of information. However, methodologically, it is mobile operative layer
the individual’s process (or mode of usage of the
device) that influences how the information is This layer deals with moving the core internal
received. Socially, the information layer of mobile business processes (that are typically operational
Internet usage has the potential for degenerating in nature) on to the mobile internet. Common
into mobile SPAM and other unsolicited messag- examples of operative business processes that
ing that is a part of MET challenge. transition to the mobile internet include timesheets
and inventory. Thus, mobile technologies may
mobile transactive layer enable a manager using a simple mobile-enabled
gadget (phone, pager of PDA) to keep track of
During this stage of Internet usage, businesses are employees; or may enable her to keep a tab on the
more serious in conducting business on the Internet inventory and place a re-order at the right time.
than they are during the informative usage. The
transactive usage implies a “two-way” (or more)
communication between the business and the user,
resulting in sales of products and services as well
529
mobile collaborative layer practical experiences of the author, enactment of

MET pilot should consider at least the following:
This layer will result in numerous individuals as
well as businesses all collaborating to satisfy the • Planning the pilot over an entire end-to-
needs and demands of numerous other businesses. end chain of a small section of the business
Groupware and portals are the technical starting • Identifying the champions within the orga-
points for collaborations. However, methodologi- nization who can demonstrate and use the
cally, with the collaborative usage of the mobile output of MET in their activities
Internet, the MET will have to consider reengineer- • Creating a suite of performance metrics
ing a group of collaborating businesses as against and evaluating the results of the pilot for
reengineering only a single business. Creation of mobile enterprise transition through a suite
collaborating business clusters can be an interest- of metrics
ing study on its own, as discussed by Unhelkar • Ensuring proper starting and completion of
(2003b), and may be considered in greater detail the pilot by announcement and information
in MET. Socially, though, collaborative usage in- to stakeholders
troduces challenges in terms of ability of business • Properly time the transition to derive maxi-
partners to communicate, trust and work together mum benefit and cause minimum disrup-
to satisfy common goals. Sociology, rather than tion to normal functioning of the business
technology, and all the associated socio-cultural
issues is at the crux of the collaborative usage of
the mobile Internet. conclusIon And
Future dIrectIons
enActInG met This chapter provides the outline of a framework

for transitioning an enterprise to a mobile enter-
Describing a framework is not enough in practice. prise. This chapter outlines an orderly approach
There is also a need to work out the details of to mobile transformation make up of the four
enacting the framework. Thus, enactment of MET dimensions. The MET discussed here needs to be
is its practical implementation that would bring further augmented with appropriate people and
about mobile transition in a real enterprise. During tools to enable successful enactment of a mobile
enactment, elements within the four dimensions transition. However, this current outline of an MET
of the MET discussed here need to be worked out framework is laid down with an intention to give
in greater detail and carried out step-by-step. This directions to businesses incorporating mobility in
execution of MET requires practical project plan- provision of information, conducting of transac-
ning and project management. Detailed discussion tions with external businesses and customers, and
of the project management aspect of MET is out also internally in its operations.
of scope for this discussion. However, there is a
need to manage and contain the exposure to risks
to the project as well as to the business itself as a reFerences
result of MET. A well known approach to reducing
risks in MET enactment is to apply it to a pilot Adam, C., & Katos, V. (2005, June). The ubiqui-
project. This pilot project can be created and en- tous mobile and location-awareness time bomb.
acted over a relatively small part of the business. Cutter IT Journal, 18(6), 20–26.
Based on suggestions by Brans (2003) and the
530
Arunatileka, D., & Unhelkar, B. (2003). Mobile Kanter, T. (2003, February). Going wireless, en-
Technologies, providing new possibilities in Cus- abling an adaptive and extensible environment.
tomer Relationship Management. Proceedings of Mobile Networks and Applications. ACM Press
5th International Information Technology Confer- New York, NY, USA, 8(1), 37–50.
ence, Colombo, Sri Lanka, December.
Lan, Y., & Unhelkar, B. (2005). Global Enterprise
Brans, P. (2003). Mobilize Your Enterprise: Transitions. Idea Group Publication (IGI press),
Achieving Competitive Advantage through Wire- Hershey, PA.
less Technology. Upper Saddle River, NJ: Hewlett-
Marmaridis, I. (Makis), & Unhelkar, B. (2005).
Packard, Pearson Education as Prentice Hall PTR.
Challenges in Mobile Transformations: A Re-
Constantine, L., & Lockwood, L. (1999). Software quirements modeling perspective for Small and
for Use: a Practical Guide to Models and Methods Medium Enterprises. Proceedings of International
of Usage-centered Design, Addison-Wesley. Also Conference on Mobile Business, ICMB, Sydney
see www.foruse.com
McKnight, L., & Howison, J. (2004). Wireless
Deshpande, Y., Murugesan, S., Unhelkar, B., & Grids: Distributed Resource Sharing by Mobile,
Arunatileka, D. (2004). Workshop on Device Nomadic, and Fixed Devices. IEEE Internet
Independent Web Engineering: Methodological Computing, Jul/Aug 2004 issue, http://dsonline.
Considerations and Challenges in Moving Web computer.org/0407/f/w4gei.htm (last accessed
Applications from Desk-top to Diverse Mobile 19th July, 2004)
Devices. Proceedings of the Device Independent
Ranjbar. M., & Unhelkar, B. (2003). Globalisation
Web Engineering Workshop, Munich.
and Its Impact on Telecommuting: An Australian
Devereaus, M., & Johansen, R. (1994, Global Perspective. Presented at IBIM03 - International
Work: Bridging Distance, Culture and Time, Business Information Management Conference
Jossey-Bass, 38-39 Elliott, G., & Phillips, N. (www.ibima.org), Cairo, Egypt.
(2004). Mobile Commerce and Wireless Comput-
S’duk. R., & Unhelkar, B. (2005). Web Services
ing Systems, Pearson/Addison-Wesley, Harlow,
Extending BPR to Industrial Process Reengineer-
England.
ing. Proceedings of International Resource Man-
Ginige, A., Murugesan, S., & Kazanis, P. (2001). agement Association (IRMA) Conference;http://
A Road Map for Successfully Transforming SMEs www.irma-international.org, San Diego, USA.
into E-Businesses. Cutter IT Journal, 14. 15th to 18th May.
Godbole, N. (2003). Mobile Computing: Security Subramanium, C., Kuppuswami, A., & Unhelkar,
Issues in Hand-held Devices. Paper presented at B. (2004). Relevance of State, Nature, Scale and
NASONES 2003 National Seminar on Network- Location of Business E-Transformation in Web
ing and e-Security by Computer Society of India. Services. Proceedings of the 2004 International
Symposium on Web Services and Applications
Godbole, N., & Unhelkar, B. (2003). Enhancing
(ISWS’04: June 21-24, 2004, Las Vegas, Nevada,
Quality of Mobile Applications through Model-
USA; http://www.world-academy-of-science.org)
ing. Proceedings of Computer Society of India’s
35th Convention, December, Indian Institute of
Technology, Delhi, India
531
Thai, B., Wan, Seneviratne, A., & Rakotoar- key terms And deFInItIons
ivelo, T. (2003). Integrated personal mobil-
ity architecture: a complete personal mobility Mobile Technologies: Are made up of wire-
solution. [ACM Press New York, NY, USA.]. less network, devices and contents. Mobile tech-
Mobile Networks and Applications, 8(1), 27–36. nologies are at the crux of the communications
.doi:10.1023/A:1021115610456 revolution.
Economic Dimension of MET: Describes the
Unhelkar, B. (2003a). Process Quality Assurance business reasons for undertaking transformation
for UML-based Projects. Boston, MA: Addison- and includes discussions on costs and competition.
Wesley. Technical Dimension of MET: Describes
Unhelkar, B. (2003b) Understanding Collabora- the technologies for transformation and include
tions and Clusters in the e-Business World. We-B devices/gadgets, programming, databases, net-
Conference, (www.we-bcentre.com; with Edith working, security and architecture.
Cowan University), Perth, 23-24 Nov. Methodological Dimension of MET: Deals
primarily with the question of “how” to – amongst
Unhelkar, B. (2004). Globalization with Mobility. other things- model and design business processes,
Presented at ADCOM 2004, 12th International approach methodologies and quality in software
Conference on Advanced Computing and Com- processes.
munications, Ahmedabad, India. Social Dimension of MET: Deals with “who”
Unhelkar, B. (2005). Transitioning to a Mobile is involved in and influenced by the transformation
Enterprise: A Three-Dimensional Framework. and typically it includes the users, customers and
Cutter IT Journal, 18(8). employees of the business.
Unhelkar, B. (2008). Mobile Enterprise Transition

and Management. New York, USA: Taylor and
Francis (Auerbach) Publications.
This work was previously published in Handbook of Research in Mobile Business, Second Edition: Technical, Methodological
and Social Perspectives, edited by Bhuvan Unhelkar, pp. 63-72, copyright 2009 by Information Science Reference (an imprint
of IGI Global).
532
533
Chapter 2.13
Development and Design
Methodologies in DWM
James Yao
John Wang
Qiyang Chen
June Lu
University of Houston – Victoria, USA
IntroductIon information systems. The topics on information

systems and DSS research diversifies. One of
Information systems were developed in early the major topics has been on how to get systems
1960s to process orders, billings, inventory design right.
controls, payrolls, and accounts payables. Soon As an active component of DSS, data ware-
information systems research began. Harry Stern housing became one of the most important devel-
started the “Information Systems in Management opments in the information systems field during
Science” column in Management Science journal the mid-to-late 1990s. It has been estimated that
to provide a forum for discussion beyond just re- about 95% of the Fortune 1000 companies either
search papers (Banker & Kauffman, 2004). Ackoff have a data warehouse in place or are planning
(1967) led the earliest research on management to develop one (Wixon & Watson, 2001). Data
information systems for decision-making purposes warehousing is a product of business need and
and published it in Management Science. Gorry technological advances. Since business envi-
and Scott Morton (1971) first used the term deci- ronment has become more global, competitive,
sion support systems (DSS) in a paper and con- complex, and volatile customer relationship
structed a framework for improving management management (CRM) and e-commerce initiatives
are creating requirements for large, integrated data
DOI: 10.4018/978-1-59904-843-7.ch029 repositories and advanced analytical capabilities.
Development and Design Methodologies in DWM
By using a data warehouse, companies can make ing applications (Hoffer, Prescott, & McFadden,
decisions about customer-specific strategies such 2007). A data warehouse schema is viewed as a
as customer profiling, customer segmentation, dimensional model (Ahmad, Azhar, & Lukauskis,
and cross-selling analysis (Cunningham, Song, 2004; Han & Kamber, 2006; Levene & Loizou,
& Chen, 2006). To analyze these large quantities 2003). It typically adopts either a star or snowflake
of data, data mining has been widely used to find schema and a subject-oriented database design
hidden patterns in the data and even discover (Han & Kamber, 2006). The schema design is the
knowledge from the collected data. Thus how to most critical to the design of a data warehouse.
design and develop a data warehouse and how to Many approaches and methodologies have
use data mining in the data warehouse development been proposed in the design and development
have become important issues for information of data warehouses. Two major data warehouse
systems designers and developers. design methodologies have been paid more at-
This article presents some of the currently tention. Inmon, Terdeman, and Imhoff (2000)
discussed development and design methodologies proposed the CIF architecture. This architecture,
in data warehousing and data mining, such as the in the design of the atomic-level data marts, uses
multidimensional model vs. relational entity- denormalized entity-relationship diagram (ERD)
relationship (ER) model, corporate information schema. Kimball (1996, 1997) proposed multidi-
factory (CIF) vs. multidimensional methodolo- mensional (MD) architecture. This architecture
gies, data-driven vs. metric-driven approaches, uses star schema at atomic-level data marts.
top-down vs. bottom-up design approaches, data Which architecture should an enterprise follow?
partitioning and parallel processing, materialized Is one better than the other? Currently, the most
view, data mining, and knowledge discovery in popular data model for data warehouse design is
database (KDD). the dimensional model (Bellatreche & Mohania,
2006; Han & Kamber, 2006). Some researchers
call this model the data-driven design model. Artz
bAckGround (2006) advocates the metric-driven view, which,
as another view of data warehouse design, begins
Data warehouse design is a lengthy, time- by identifying key business processes that need
consuming, and costly process. Any wrongly to be measured and tracked over time in order
calculated step can lead to a failure. Therefore, for the organization to function more efficiently.
researchers have placed important efforts to the There has always been the issue of top-down vs.
study of design and development related issues bottom-up approaches in the design of informa-
and methodologies. tion systems. The same is with a data warehouse
Data modeling for a data warehouse is different design. These have been puzzling questions for
from operational database, for example, online business intelligent architects and data warehouse
transaction processing (OLTP), data modeling. designers and developers. The next section will
An operational system is a system that is used to extend the discussion on issues related to data
run a business in real time, based on current data. warehouse and mining design and development
An OLTP system usually adopts ER modeling methodologies.
and application-oriented database design (Han
& Kamber, 2006). An information system, like a
data warehouse, is designed to support decision
making based on historical point-in-time and
prediction data for complex queries or data min-
534
desIGn And develoPment of smaller dimension tables (Ahmad et al., 2004;

metHodoloGIes Han & Kamber, 2006). The fact table is connected
with the dimension tables using many-to-one
data Warehouse data modeling relationships to ensure their hierarchy. The star
schema can provide fast response time allowing
Database design is typically divided into a four- database optimizers to work with simple database
stage process (Raisinghani, 2000). After require- structures in order to yield better execution plans.
ments are collected, conceptual design, logical The snowflake schema (see Figure 2) is a
design, and physical design follow. Of the four variation of the star schema model, in which all
stages, logical design is the key focal point of the dimensional information is stored in the third
database design process and most critical to the normal form, thereby further splitting the data
design of a database. In terms of an OLTP system into additional tables, while keeping fact table
design, it usually adopts an ER data model and structure the same. To take care of hierarchy, the
an application-oriented database design (Han & dimension tables are connected with sub-dimen-
Kamber, 2006). The majority of modern enter- sion tables using many-to-one relationships. The
prise information systems are built using the ER resulting schema graph forms a shape similar to
model (Raisinghani, 2000). The ER data model a snowflake (Ahmad et al., 2004; Han & Kamber,
is commonly used in relational database design, 2006). The snowflake schema can reduce redun-
where a database schema consists of a set of enti- dancy and save storage space. However, it can
ties and the relationship between them. The ER also reduce the effectiveness of browsing and the
model is used to demonstrate detailed relationships system performance may be adversely impacted.
between the data elements. It focuses on remov- Hence, the snowflake schema is not as popular
ing redundancy of data elements in the database. as star schema in data warehouse design (Han &
The schema is a database design containing the Kamber, 2006). In general, the star schema requires
logic and showing relationships between the data greater storage, but it is faster to process than the
organized in different relations (Ahmad et al., snowflake schema (Kroenke, 2004).
2004). Conversely, a data warehouse requires a The starflake schema (Ahmad et al., 2004),
concise, subject-oriented schema that facilitates also known as galaxy schema or fact constellation
online data analysis. A data warehouse schema is schema (Han & Kamber, 2006), is a combination
viewed as a dimensional model which is composed
of a central fact table and a set of surrounding
dimension tables, each corresponding to one of Figure 1. Example of a star schema (adapted from
the components or dimensions of the fact table Kroenke, 2004)
(Levene & Loizou, 2003). Dimensional models
are oriented toward a specific business process or
subject. This approach keeps the data elements
associated with the business process only one
join away. The most popular data model for a data
warehouse is the multidimensional model. Such
a model can exist in the form of a star schema, a
snowflake schema, or a starflake schema.
The star schema (see Figure 1) is the simplest
data base structure containing a fact table in the
center, no redundancy, which is surrounded by a set
535
Figure 2. Example of a snowflake schema (adapted

schema or fact constellation) schema can model
from Kroenke, 2004)
multiple and interrelated subjects. Therefore, it
is usually used to model an enterprise-wide data
warehouse. A data mart, on the other hand, is
similar to a data warehouse but limits its focus to
a department subject of the data warehouse. Its
scope is department-wide. The star schema and
snowflake schema are geared towards modeling
single subjects. Consequently, the star schema or
snowflake schema is commonly used for a data
mart modeling, although the star schema is more
popular and efficient (Han & Kamber, 2006).
cIF vs. multidimensional

of the denormalized star schema and the normal-
ized snowflake schema (see Figure 3). The Two major design methodologies have been paid
starflake schema is used in situations where it is more attention in the design and development of
difficult to restructure all entities into a set of data warehouses. Kimball (1996, 1997) proposed
distinct dimensions. It allows a degree of crossover MD architecture. Inmon et al. (2000) proposed
between dimensions to answer distinct queries the CIF architecture. Imhoff, Galemmco, and
(Ahmad et al., 2004). Figure 3 illustrates the Geiger (2004) made a comparison between the
starflake schema. two by using important criteria, such as scope,
What needs to be differentiated is that the three perspective, data flow, and so forth. One of the
schemas are normally adopted according to the most significant differences between the CIF and
differences of design requirements. A data ware- MD architectures is the definition of the data mart.
house collects information about subjects that For MD architecture, the design of the atomic-
span the entire organization, such as customers, level data marts is significantly different from
items, sales, and so forth. Its scope is enterprise- the design of the CIF data warehouse, while its
wide (Han & Kamber, 2006). Starflake (galaxy aggregated data mart schema is approximately the
Figure 3. Example of a starflake schema (galaxy schema or fact constellation) (adapted from Han &
Kamber, 2006)
536
same as the data mart in the CIF architecture. MD architectural approaches, namely, the ERD or
architecture uses star schemas, whereas CIF archi- normalization techniques for the data warehouse
tecture uses denormalized ERD schema. This data and the star schema data model for multidimen-
modeling difference constitutes the main design sional data marts. A CIF architecture with only a
difference in the two architectures (Imhoff et al., data warehouse and no multidimensional marts is
2004). A data warehouse may need both types of almost useless and a multidimensional data-mart-
data marts in the data warehouse bus architecture only environment risks the lack of an enterprise
depending on the business requirements. Unlike integration and support for other forms of business
the CIF architecture, there is no physical reposi- intelligence analyses.
tory equivalent to the data warehouse in the MD
architecture. data-driven vs. metric-driven
The design of the two data marts is predomi-
nately multidimensional for both architecture, Currently, the most popular data model for data
but the CIF architecture is not limited to just this warehouse design is the dimensional model
design and can support a much broader set of (Bellatreche & Mohania, 2006; Han & Kamber,
data mart design techniques. In terms of scope, 2006). In this model, data from OLTP systems
both architectures deal with enterprise scope and are collected to populated dimensional tables.
business unit scope, with CIF architecture putting Researchers term a data warehouse design based
a higher priority on enterprise scope and MD on this model as a data-driven design model since
architecture placing a higher priority on business the information acquisition processes in the data
unit scope. With CIF architecture, the information warehouse are driven by the data made available
technology (IT) side tackles the problem of sup- in the underlying operational information systems.
plying business intelligence source data from an Another view of data warehouse design is called
enterprise point of view. With MD architecture, its the metric-driven view (Artz, 2006), which begins
proponents emphasize the perspective of consum- by identifying key business processes that need
ing business unit data. For data flow, in general, to be measured and tracked over time in order
the CIF approach is top-down, whereas the MD for the organization to function more efficiently.
approach is bottom-up. The difference between Advantages of data-driven model include that it
the two in terms of implementation speed and cost is more concrete, evolutionary, and uses derived
involves long-term and short-term trade-offs. A summary data. Yet the information generated
CIF project, as it is at enterprise level, will most from the data warehouse may be meaningless to
likely require more time and cost up front than the user owing to the fact that the nature of the
the initial MD project, but the subsequent CIF derived summary data from OLTP systems may
projects tend to require less time and cost than not be clear. The metric-driven design approach,
subsequent MD projects. MD architecture claims on the other hand, begins first by defining key
that all its components must be multidimensional business processes that need to be measured and
in design. Conversely, CIF architecture makes no tracked over time. After these key business pro-
such claim and is compatible with many different cesses are identified, then they are modeled in a
forms of business intelligence analyses and can dimensional data model. Further analysis follows
support technologies that are not multidimensional to determine how the dimensional model will be
in nature. For MD architecture, retrofitting is populated (Artz, 2006).
significantly harder to accomplish. Imhoff et al. According to Artz (2006), data-driven model
(2004) encourage the application of a combina- to a data warehouse design has little future since
tion of the data modeling techniques in the two information derived from a data-driven model
537
is information about the data set. Metric-driven & Kamber, 2006), with which “an organization
model, conversely, possibly has some key impacts can exploit the planned and strategic nature of
and implications because information derived the top-down approach while retaining the rapid
from a metric-driven model is information about implementation and opportunistic application of
the organization. Data-driven approach is domi- the bottom-up approach” (p. 129), when such an
nating data warehouse design in organizations at approach is necessitated in the undergoing orga-
present. Metric-driven, on the other hand, is at nizational and business scenarios.
its research stage, needing practical application
testimony of its speculated potentially dramatic materialized view
implications.
One of the advantages of data warehousing ap-
top-down vs. bottom-up proach over traditional operational database ap-
proach to the integration of multiple sources is
There are two approaches in general to building that the queries can be answered locally without
a data warehouse prior to the data warehouse accessing the original information sources (The-
construction commencement, including data odoratos & Sellis, 1999). Queries to data ware-
marts: the top-down approach and bottom-up houses often involve a huge number of complex
approach (Han & Kamber, 2006; Imhoff et al., aggregations over vast amount of data. As a result,
2004; Marakas, 2003). Top-down approach starts data warehouses achieve high performances of
with a big picture of the overall, enterprise-wide query by building a large number of materialized
design. The data warehouse to be built is large views (Bellatreche & Mohania, 2006; Jixue, Mil-
and integrated. However, this approach is risky list, Vincent, & Mohania, 2003; Lee, Chang, &
(Ponniah, 2001). A top-down approach is to design Lee, 2004; Theodoratos & Sellis, 1999). One of
the warehouse with an enterprise scope. The focus the common problems related to the materialized
is on integrating the enterprise data for usage in views is view selection. It seems infeasible to store
any data mart from the very first project (Imhoff all the materialized views as we are constrained by
et al., 2004). It implies a strategic rather than an some resources such as data warehouse disk space
operational perspective of the data. It serves as the and maintenance cost (Bellatreche & Mohania,
proper alignment of an organization’s information 2006). Therefore, an appropriate set of views
systems with its business goals and objectives should be selected among the candidate views.
(Marakas, 2003). In contrast, a bottom-up ap- Studies have shown that materialized view selec-
proach is to design the warehouse with business- tion problem is proven to be an NP-hard problem
unit needs for operational systems. It starts with (Bellatreche & Mohania, 2006; Lee et al., 2004).
experiments and prototypes (Han & Kamber, Baralis, Paraboschi, and Teniente (1997); Mistry,
2006). With bottom-up, departmental data marts Roy, Sudarshan, and Ramamritham (2001); and
are built first one by one. It offers faster and easier Theodoratos and Sellis (1997) researched on
implementation, favorable return on investment, solving the materialized view selection problem,
and less risk of failure, but with a drawback of but they considered only the intermediate query
data fragmentation and redundancy. The focus of results that appeared in the given workloads’
bottom-up approach is to meet unit-specific needs execution plan as candidate materialized views
with minimum regards to the overall enterprise- (Lee et al., 2004). As a result, the views were
wide data requirements (Imhoff et al., 2004). excluded from the candidate view space if they
An alternative to the previously discussed two joined the relations not referred to in the queries
approaches is to use a combined approach (Han even though the views could have been used in
538
the optimal query execution plan (Bellatreche & Parallel processing is based on a parallel
Mohania, 2006; Chang & Lee, 1999). Lee et al. database, in which multiprocessors are in place.
(2004) developed a solution for identifying the Parallel databases link multiple smaller machines
candidate view space of materialization, which to achieve the same throughput as a single, larger
demonstrates that the candidate view space can be machine, often with greater scalability and reli-
optimized by join-lossless property, partial-view ability than single processor databases (Singh,
and union-view. These methods can present better 1998). In a context of relational online analytical
results when the database schema, especially the processing (ROLAP), by partitioning data of RO-
join relations, gets more complex. LAP schema (star schema or snowflake schema)
among a set of processors, OLAP queries can be
data Partitioning and executed in a parallel, potentially achieving a linear
Parallel Processing speedup and thus significantly improving query
response time (Datta, Moon, & Thomas, 1998;
Data partitioning is the process of decompos- Tan, 2006). Given the size of contemporary data
ing large tables (fact tables, materialized views, warehousing repositories, multiprocessor solu-
indexes) into multiple small tables by applying tions are crucial for the massive computational
the selection operators (Bellatreche & Mohania, demands for current and future OLAP system
2006). A good partitioning scheme is an essential (Dehne, Eavis, & Rau-Chaplin, 2006). The as-
part of designing a database that will benefit from sumption of most of the fast computation algo-
parallelism (Singh, 1998). With a well performed rithms is that their algorithms can be applied into
partitioning, significant improvements in avail- the parallel processing system (Dehne et al., 2006;
ability, administration, and table scan performance Tan, 2006). As a result, it is sometimes necessary
can be achieved. Singh (1998) described five to use parallel processing for data mining because
methods of partitioning: (1) hashing algorithm to large amounts of data and massive search efforts
distribute data uniformly across disks, (2) round- are involved in data mining (Turban, Aronson,
robin partitioning (assigning a row to partitions in & Liang, 2005). Therefore, data partitioning and
sequence), (3) allocating rows to nodes based on parallel processing are two complementary tech-
ranges of values, (4) schema partitioning to tie a niques to achieve the reduction of query processing
table to a particular partition, and (5) user-defined cost in data warehousing design and development
rules to allocate data in a particular partition. (Bellatreche & Mohania, 2006).
Bellatreche and Mohania (2006) and Bellatreche,
Schneider, Mohania, and Bhargava (2002) on data mining
the other hand offer two types of partitioning:
horizontal and vertical. In horizontal fragmenta- As a process, data mining endeavors require certain
tion, each partition consists of a set of rows of the steps necessary to achieve a successful outcome.
original table. In the vertical fragmentation, each A step common to data mining is infrastructure
partition consists of a set of columns of the original preparation. Without the infrastructure data min-
table. Furthermore, horizontal fragmentation can ing activities simply will not occur. Minimum
be divided into two versions: primary horizontal requirements for the infrastructure are: a hardware
partitioning and derived horizontal partition. The platform, database management system (DBMS)
former one is performed using predicates that are platform, and one or more tools for data mining
defined on that table, the later one results from (Marakas, 2003). The hardware platform, in most
predicates defined on another relation. cases, is a separate platform than that which
originally housed the data. The data mining en-
539
vironment is usually a client/server architecture contrast, data mining is the process of forming
or a Web-based architecture (Turban et al., 2005). general concept definitions by observing specific
To perform data mining, data must be removed examples of concepts to be learned. It is a process
from its host environment and prepared before of business intelligence that can be used together
it can be properly mined. This process is called with what is provided by IT to support company
extraction, transformation, and loading (ETL) decisions (Giudici, 2003). A data warehouse has
process, in which the data is scrubbed/cleaned uses other than data mining. However, the fullest
and transformed according to the requirements use of a data warehouse must include data mining
and then loaded to the hardware platform, usually (Marakas, 2003), which in turn, at an upper level,
a data warehouse or data mart (Han & Kamber, discovers knowledge for the user.
2006; Marakas, 2003; Turban et al., 2005). A
DBMS is the fundamental system for the data-
base. Sophisticated tools and techniques for data Future trends
mining are selected based on mining strategies
and needs. A well-developed infrastructure is to Future information systems research will continue
be a pre-mining assurance for a successful data with the study of problems in information systems
mining endeavor. management, including systems analysis and
Data mining is the analysis of observational design (Banker & Kauffman, 2004). According
data sets to find unsuspected relationships and to Cunningham et al. (2006), there are no agreed
to summarize the data in novel ways that are upon standardized rules for how to design a data
both understandable and useful to the data owner warehouse to support CRM and a taxonomy of
(Hand, Mannila, & Smyth, 2001). Another term CRM analyses needs to be developed to determine
that is frequently used interchangeably with data factors that affect design decisions for CRM data
mining is KDD. KDD was coined to describe warehouse. Enterprises are moving towards build-
all those methods that seek to find relations and ing the operational data store, which derives data
regularity among the observed data and was gradu- from enterprise resource planning (ERP) systems,
ally expanded to describe the whole process of solutions for real-time business analysis. There
extrapolating information from a database, from is a need for active integration of CRM with
the identification of the initial business objectives operational data store for real-time consulting
to the application of the decision rules (Giudici, and marketing (Bellatreche & Mohania, 2006).
2003). Although many people treat data mining as In the data modeling area, to develop a more
a synonym for KDD, there are differences between general solution for modeling data warehouse
the two. Technically, KDD is the application of the current ER model and dimensional model
the scientific method to data mining (Roiger & need to be extended to the next level to combine
Geatz, 2003). Apart from performing data min- the simplicity of the dimensional model and the
ing, a typical KDD process model includes a efficiency of the ER model with the support of
methodology for extracting and preparing data object oriented concepts (Raisinghani, 2000).
as well as making decisions about actions to be
taken once data mining has taken place. The KDD
process involves selecting the target data, prepro- conclusIon
cessing the data, transforming them if necessary,
performing data mining to extract patterns and Several data warehousing and data mining de-
relationships, and then interpreting and assessing velopment and design methodologies have been
the discovered structures (Hand et al., 2001). In reviewed and discussed, followed by some trends
540
in data warehousing design. There are more is- Bellatreche, L., Schneider, M., Mohania, M., &
sues in relation to the topic but are limited due to Bhargava, B. (2002). PartJoin: An efficient stor-
the paper size. Some of the methodologies have age and query execution for data warehouses. In
been practiced in the real world and accepted by Proceedings of the 4th International Conference
today’s businesses. Yet new challenging method- on Data Warehousing and Knowledge Discovery
ologies, particularly in data modeling and models (DAWAK’02) (pp. 109-132).
for physical data warehousing design, need to be
Chang, J., & Lee, S. (1999). Extended condi-
further researched and developed.
tions for answering an aggregate query using
materialized views. Information Processing
Letters, 72(5-6), 205–212. doi:10.1016/S0020-
reFerences
0190(99)00147-7
Ackoff, R. I. (1967). Management misinformation Cunningham, C., Song, I., & Chen, P. P. (2006,
systems. Management Science, 14(4), 147–156. April-June). Data warehouse design to support
doi:10.1287/mnsc.14.4.B147 customer relationship management analyses.
Ahmad, I., Azhar, S., & Lukauskis, P. (2004). Journal of Database Management, 17(2), 62–84.
Development of a decision support system using Datta, A., Moon, B., & Thomas, H. (1998). A case
data warehousing to assist builders/developers for parallelism in data warehousing and OLAP.
in site selection. Automation in Construction, Proceedings of the 9th International Workshop
13, 525–542. doi:10.1016/j.autcon.2004.03.001 on Database and Expert Systems Applications
Artz, J. M. (2006). Data driven vs. metric driven (DEXA’98) (pp. 226-231).
data warehouse design. In J. Wang (Ed.), Ency- Dehne, F., Eavis, T., & Rau-Chaplin, A. (2006).
clopedia of data warehousing and mining (pp. The cgmCUBE project: Optimizing parallel data
223-227). Hershey, PA: Idea Group. cube generation for ROLAP. Distributed and
Banker, R. D., & Kauffman, R. J. (2004, March). Parallel Databases, 19, 29–62. doi:10.1007/
The evolution of research on information systems: s10619-006-6575-6
A fiftieth-year survey of the literature in Man- Giudici, P. (2003). Applied data mining: Statistical
agement Science. Management Science, 50(3), methods for business and industry. West Sussex,
281–298. doi:10.1287/mnsc.1040.0206 England.
Baralis, E., Paraboschi, S., & Teniente, E. (1997). Gorry, G. A., & Scott Morton, M. S. (1971). A
Materialized view selection in multidimensional framework for management information systems.
database. In Proceedings of the 23rd International Sloan Management Review, 13(1), 1–22.
Conference on Very Large Data Bases (VLDB’97)
(pp. 156-165). Han, J., & Kamber, M. (2006). Data mining:
Concepts and techniques (2nd ed.). San Francisco:
Bellatreche, L., & Mohania, M. (2006). Physical Morgan Kaufmann.
data warehousing design. In J. Wang (Ed.), En-
cyclopedia of data warehousing and mining (pp. Hand, D., Mannila, H., & Smyth, P. (2001). Prin-
906-911). Hershey, PA: Idea Group. ciples of data mining. Cambridge, MA: MIT Press.
Hoffer, J. A., Prescott, M. B., & McFadden, F. R.
(2007). Modern database management (8th ed.).
Upper Saddle River, NJ: Pearson Prentice Hall.
541
Imhoff, C., Galemmco, M., & Geiger, J. G. (2004). Ponniah, P. (2001). Data warehousing fundamen-
Comparing two data warehouse methodologies. tals: A comprehensive guide for IT professionals.
(Database and network intelligence). Database New York: John Wiley & Sons, Inc.
and Network Journal, 34(3), 3–9.
Raisinghani, M. S. (2000). Adapting data model-
Inmon, W. H., Terdeman, R. H., & Imhoff, C. ing techniques for data warehouse design. Journal
(2000). Exploration warehousing: Turning busi- of Computer Information Systems, 4(3), 73–77.
ness information into business opportunity. New
Roiger, R. J., & Geatz, M. W. (2003). Data
York: John Wiley & Sons, Inc.
mining: A tutorial-based primer. New York:
Jixue, L., Millist, W., Vincent, M., & Mohania, Addison-Wesley.
K. (2003). Maintaining views in object-relational
Singh, H. S. (1998). Data warehousing: Concepts,
databases. Knowledge and Information Systems,
technologies, implementations, and management.
5(1), 50–82. doi:10.1007/s10115-002-0067-z
Upper Saddle River, NJ: Prentice Hall PTR.
Kimball, R. (1996). The data warehouse toolkit:
Tan, R. B. (2006). Online analytical processing
Practical techniques for building dimensional
systems. In J. Wang (Ed.), Encyclopedia of data
data warehouses. New York: John Wiley & Sons.
warehousing and mining (pp. 876-884). Hershey,
Kimball, R. (1997, August). A dimensional model- PA: Idea Group.
ing manifesto. DBMS, 10(9), 58–70.
Theodoratos, D., & Sellis, T. (1997). Data ware-
Kroenke, D. M. (2004). Database processing: house configuration. In Proceedings of the 23rd
Fundamentals, design and implementation (9th International Conference on Very Large Data
ed.). Saddle River, NJ: Prentice Hall. Bases (VLDB’97) (pp. 126-135).
Lee, T., Chang, J., & Lee, S. (2004). Using rela- Theodoratos, D., & Sellis, T. (1999). Designing
tional database constraints to design materialized data warehouses. Data & Knowledge Engi-
views in data warehouses. APWeb, 2004, 395–404. neering, 31(3), 279–301. doi:10.1016/S0169-
023X(99)00029-4
Levene, M., & Loizou, G. (2003). Why is the
snowflake schema a good data warehouse de- Turban, E., Aronson, J. E., & Liang, T. (2005).
sign? Information Systems, 28(3), 225–240. Decision support systems and intelligent systems
doi:10.1016/S0306-4379(02)00021-2 (7th ed.). Upper Saddle River, NJ: Pearson Edu-
cation Inc.
Marakas, G. M. (2003). Modern data warehous-
ing, mining, and visualization: Core concepts. Wixon, B. H., & Watson, H. (2001). An empirical
Upper Saddle River, NJ: Pearson Education Inc. investigation of the factors affecting data ware-
housing success. MIS Quarterly, 25(1), 17–41.
Mistry, H., Roy, P., Sudarshan, S., & Ramam-
doi:10.2307/3250957
ritham, K. (2001). Materialized view selection and
maintenance using multi-query optimization. In
Proceedings of the ACM SIGMOD International
Conference on Management of Data 2001 (pp. key terms And deFInItIons
310-318).
Dimensional Model: A dimensional model
contains a central fact table and a set of surround-
542
ing dimension tables, each corresponding to one Materialized View: Mmaterialized view are
of the components or dimensions of the fact table. copies or replicas of data based on SQL queries
Dimensions: Dimensions are the perspectives created in the same manner as dynamic views
or entities with respect to which an organization (Hoffer et al., 2007, p. 298).
wants to keep records (Han & Kamber, 2006, p. Metric-Drive Design: Metric-drive design is
110). a data warehousing design approach which begins
Entity-Relationship Data Model: An entity- by defining key business processes that need to
relationship data model is a model that represents be measured and tracked over time. Then they are
database schema as a set of entities and the rela- modeled in a dimensional model.
tionships among them. Parallel Processing: Parallel processing is
Fact Table: A fact table is the central table in the allocation of the operating system’s process-
a star schema, containing the names of the facts, ing load across several processors (Singh, 1998,
or measures, as well as keys to each of the related p. 209).
dimension tables. Star Schema: Star schema is a modeling dia-
Knowledge Discovery in Databases (KDD): gram that contains a large central table (fact table)
KDD is the process of extrapolating information and a set of smaller attendant tables (dimension
from a database, from the identification of the tables) each represented by only one table with
initial business aims to the application of the a set of attributes.
decision rules (Giudici, 2003, p. 2).
This work was previously published in Encyclopedia of Decision Making and Decision Support Technologies, edited by Fred-
eric Adam and Patrick Humphreys, pp. 236-244, copyright 2009 by Information Science Reference (an imprint of IGI Global).
543
544
Chapter 2.14
Facilitating Design of Efficient
Components by Bridging
Gaps Between Data Model and
Business Process via Analysis
of Service Traits of Data
Ning Chen
Xi’an Polytechnic University, China
AbstrAct nents. In addition, the authors present the design of

a toolkit that automatically applies these patterns
In many large-scale enterprise information system to bridge the gaps to ensure reduced development
solutions, process design, data modeling and soft- time, and higher solution quality.
ware component design are performed relatively
independently by different people using various
tools and methodologies. This usually leads to gaps IntroductIon
among business process modeling, component de-
sign and data modeling. Currently, these functional With the development of information technology,
or non-functional disconnections are fixed manu- enterprise information becomes more complex and
ally, which increases the complexity and decrease tends to change more frequently; consequently
the efficiency and quality of development. In this enterprise should adjust its business according to
chapter, a pattern-based approach is proposed to market, which requires enterprise IT system to
bridge the gaps with automatically generated data be flexible and agile enough to response to the
access components. Data access rules and patterns changes. Now, business process modeling consists
are applied to optimize these data access compo- of service modeling, data modeling and component
modeling, which are the three main threads in
DOI: 10.4018/978-1-60566-330-2.ch007 enterprise IT system solution design (Ivica, 2002;
Facilitating Design of Efficient Components by Bridging Gaps Between Data Model and Business Process
Mei, 2003). They are usually performed relatively However, the existing enterprise modeling
independently, for different roles employ differ- methods remain largely unharnessed due to the
ent methodologies. The result is in a gap among following shortages: (1) They lack the automation
process model, data model and components, which of analysis mechanism which makes the enterprise
requires significant amount of efforts to fill in the unresponsive to the enterprise changes and increases
gap. Enterprise information system is an appli- the maintaining overhead of the evolution of these
cation with dense data (Martin, 2002) and mass models; (2) Some enterprise models are just con-
data access. Both functional and non-functional ceptual models and should be analyzed by hand.
aspects, such as system response time and data Others employ the complex mathematical models
throughput etc., are satisfied in system integration for analysis, which are hard for the business users
in order to provide efficient data access within to comprehend and manipulate. (3) The knowledge
process execution. Meeting these requirements reuse is difficult for the business users due to the
is a challenge presented to the solution designed, heterogeneity of the enterprise models.
which will greatly affect the efficiency of system In order to tackle the above problems, through
development. Therefore, how to build the relation- deep analysis of business process modeling and
ship model between business process and data data modeling, we extract process data mapping
model, and how to use the orchestration model to and data access flow to build data access com-
automatically generate data access components are ponents for bridging business process and data
two questions that have great impact to software model. Furthermore, a pattern is automatically
development. applied to data access component for facilitating
an efficient service.
relAtIon Works
Process/dAtA
The existing enterprise modeling approaches relAtIonsHIP model
are focused on two domains including peer-to-
peer enterprise system and multilayer Enterprise In present environment for software develop-
Modeling. ment, different tools are used by separate roles
David (2004) presents a loosely coupled in business process modeling, data modeling,
service-composition paradigm. This paradigm software component designing and coding. These
employs a distributed data flow that differs mark- tasks are so independent that the whole software
edly from centralized information flow adopted development becomes rather complex. Take the
by current service integration frameworks, such IBM develop studio as an example, we need to
as CORBA, J2EE and SOAP. Distributed data use modeling and programming tools such as
flows support direct data transmission to avoid WBI-Modeler, Rational Software Architect and
many performance bottlenecks of centralized WSAD-IE (Osamu, 2003). The development
processing. In addition, active mediation is used procedure contains the following steps:
in applications employing multiple web services
that are not fully compatible in terms of data 1. The analyst will analyze requirement to
formats and contents. design the use case using UML.
Martin Fowler and Clifton Nock summarize 2. By analyzing the relationship between
customary patterns of enterprise application ar- enterprise entities, the data model designer
chitecture to accelerate development of enterprise will design the data model, and create the
modeling (Martin, 2002; Clifton, 2003). database on the basis of UML.
545
3. The process analyst will design the abstract component, but also these information can be
business process using WBI Modeler. used developer to analyze the process and data
4. The software engineer will design the relationship model, and consequently for devel-
functions and APIs for components using oper to optimize data access activity, produce
RSA. approximate index for data model, create views
5. The developer will implement the software and apply data access patterns (Clifton, 2003),
components and access the database using which can enhance the data access performance
RSA. (Figure 1).
6. The process developer will develop the A business process usually contains some sub
executable process based on the abstract processes, and a sub process usually contains some
process model, and assemble the software activities, where activity is operation on data.
components as web services using WSAD- Let {P1, P2, P3, …Pr, …} be a process set in
IE. process model.
7. The system deployer will run the developed Definition 1: <Frequency of process > Let
business model (EAR file) on WBI-SF. a set PS contain some sub-processes, which are
processes directly or indirectly invoked by a busi-
Obviously, a good command of process or- ness process P , as denoted by PS = {Pr }r ∈I ,
chestration, OOD and UML is a prerequisite for where I is index set. Pr is invoked in process P
a designer to complete the solution design. The N r times, the frequency of process Pr is defined
figure1 presents the relationships among design
flows of business process, software components, as N r / ∑ N r .
r ∈I
and database. Definition 2 <Activity frequency of sub-pro-
cess > If the kth data-access-activity in sub process
Pr is invoked nr ,k times, the activity frequency
dAtA Access comPonent
of sub-process Pr is defined by nr ,k / ∑ nr ,k ,
k ∈I
Identification of Frequent Activity where I is index set.
Definition 3 <Frequency of activity > The
Business process provided much global informa- frequency of activity ar ,k is defined as the ratio
tion on the whole. Not only can these information of times of data access to the total access, i.e.
be used for developer to generate the data access nr ,k / ∑ ∑ nr ,k .
Pr ∈PS k ∈I
Figure 1. Map relationships between process service and data service
546
Box 1. Algorithm: application strategy of cache

Definition 4 <Frequent querying activity> pattern
Defined as activity with frequency of activity
Input: data access flow
greater than frequent-querying-activity threshold Output: cache pattern of data access
MAXSearchAF . Step1: Analyze the data access flow, then find all frequent
querying activity, which form the activity set T;
Definition 5 <Frequent updating activity> Step2: For all activity a∈T, given the corresponding data
model D by user interface, if the activation of D is 0, static
Defined as activity with frequency of activity cache pattern is applied; otherwise, if the activation of D is
greater than frequent-updating-activity threshold not equal to 0(0<active≤1), timing update cache pattern is ap-
MAXUpdateAF . Kactive
plied, the period of updating is , where Kactive is a
Frequency of activities can be computed by active
traversing data access flow, and frequent query- constant representing ratio of activation;
Step3: The developer can adjust the collection strategy of
ing activity and frequent updating activity can be cache by user interface;
identified based on rules. Step4: Generate cache pattern code according to cache updat-
ing strategy and collect strategy.
Automatic Application Box 2. Cache pattern rules

of cache Pattern RULE1:IF ActivityEstimate = static AND VolumeEstimate <=
Pre-Fetching Threshold
We can represent the optimization with rules. THEN latest least replace strategy
RULE2:IF ActivityEstimate = active AND VolumeEstimate <=
According to the customized threshold, frequent Pre-Fetching Threshold
data access activities can be selected, and then THEN non-active expiration strategy
RULE3:IF ActivityEstimate = veryactive AND VolumeEsti-
rule-analyzing system can use rules to recommend mate <= Pre-Fetching Threshold
approximate data access patterns. The perfor- THEN fix-time expiration strategy
RULE4:IF ActivityEstimate = static AND VolumeEstimate >
mance index and user interface can be added to Pre-Fetching Threshold
identify cache pattern. Box 1 shows a strategy of THEN latest least replace strategy
RULE5:IF ActivityEstimate = active AND VolumeEstimate >
configuration of data access. Pre-Fetching Threshold
Knowledge base of cache pattern strategy THEN non-active expiration strategy with limited
queue
stores the criterion how to apply cache patterns, RULE6:IF ActivityEstimate = veryactive AND VolumeEsti-
and recommends corresponding configuration mate > Pre-Fetching Threshold
THEN fix-time expiration strategy with limited queue
of cache pattern and cache parameters according
to activation information and different querying
Box 3. Rules for determining cache pattern pa-
condition capacity provided by user. The rules for
rameters
selection of proper cache pattern and configuration
of cache parameter are shown in Box 2 and 3. Cache collection period
Tontime
Tcollector = • hcollector + Tontime , where
Tresponse
toolkIt And result AnAlysIs Tontime denotes the change period of data items, Trespnse denotes
response time of querying operation, ηcollector is a constant
ratio. Cache queue capacity Qqueue=max{Qmax, Volume} , where
Based on Eclipse3.0 IDE and JSDK1.5, we de- Qmax denote the maximum queue, Volume denotes different
veloped a pattern-based tool-box to facilitating query condition volume.
efficient service, including process/data relation

analyzer, data access component builder and data
access optimizer, which can run alone or cooper- Through an analysis of process model file and
ate with other modeling tool as Eclipse plugin, as data model file, process/data relation analyzer can
shown in Figure 2. construct map relation files between processes in
547
Figure 2. Toolkit
N client inquiring transaction time of JavaBeans
component and cache pattern component, as shown
in Figure 5. The experimental result shows that
consumption time of data access components with
cache is less than consumption time of JavaBeans
component for big transaction. However, infor-
mation queried by data access components varies
frequently, cost of cache data validation will rise,
due to renewing or substituting cache items of
cache queue in small interval, thus decreasing
performance of data access component. In dif-
ferent scenario, effect of parameter-selection of
cache pattern on performance will be discussed
in future study.
conclusIon
process model and tables in data model; then on the
basis of map relation files, data access component In this chapter, a pattern-based approach to facili-
builder can generate data access components, for tate an efficient service is developed to automate
example JavaBeans or EJB code; finally cache pat- the analysis of large-scale enterprise information
tern builder integrates cache pattern with generated system. The approach will effectively reduce the
data access components (Figure 3 and Figure 4). number of system development problems by auto-
In order to show effect of cache pattern com- matic bridging of the gaps between development
ponent on optimizing data access, we simulate team and system analyst efficiently and effectively.
In addition, automatically generated approach,
Figure 3. Automatic generation of code: Genera-

tion of class representing data
Figure 4. Automatic generation of code: Genera-
tion of class representing EJB
548
Figure 5. Cache pattern performance

reFerences
Chen, N. (2008). A Quick Development Frame-

work Based on Process/Data Relationship Model.
In J. Zhou (Ed.), Proceedings of the 2008 Inter-
national Conference on Computer Science and
Information Technology. (pp. 597-600). Washing-
ton, DC, USA: IEEE Computer Society.
Clifton, N. (2003). Data Access Patterns: Data-
base Interactions in Object-Oriented Applications.
USA: Addison-Wesley Professional Press.
David, L., Jun, P., Kincho, H. L., & Gio, W.
(2004). Efficient Integration of Web Services with
Distributed Data Flow and Active Mediation. In
flexible data access components can provide data
M. Janssen (Ed.), ACM International Conference
access services, hiding the access complexity and
Proceeding Series: Vol. 60. Proceedings of the 6th
satisfying system’s functional requirements. The
international conference on Electronic commerce.
result is an approach for producing a pattern-based
(pp. 11-20). New York, NY, USA: ACM.
application, which completely improve on data
access and minimizes the amount of code devel- Ivica, C., & Magnus, L. (2002). Challenges of
opment required (Chen, 2008). component-based development. Journal of Sys-
Further effort in this area is required in en- tems and Software, 61(3), 201–212. doi:10.1016/
hancing the flexibility to deal with the increas- S0164-1212(01)00148-0
ing complexities of business process design, and
Martin, F. (2002). Patterns of Enterprise Ap-
improving the performance with cluster analysis
plication Architecture. USA: Addison-Wesley
adjusting the granularity of components.
Professional Press.
Mei, H., Cheng, F., Feng, Y., & Yang, J. (2003).
AcknoWledGment ABC: An Architecture Based, Component Orient-
ed Approach to Software Development. Journal
Our thanks to the IBM China Research Labora- of Software, 14(4), 721–732.
tory (CRL) in Beijing for helping us investigate
the challenge. We specially owe thanks to inves- Osamu, T. (Ed.). (2003). Exploring WebSphere
tigation participants for their ideas and efforts, Studio Application Developer Integration Edition
including Guanqun Zhang (IBM). V5, IBM RedBook SG24-6200-00. USA: IBM.
This work is supported by IBM University
Joint Research Project (Process/Data Orchestrated
Solution Design), 2005.
of IGI Global).
549
550
Chapter 2.15
The Utilization of Semantic
Web for Integrating
Enterprise Systems
Dimitrios Tektonidis
ALTEC S.A., Greece
Albert Bokma
University of Sunderland, UK
AbstrAct gies for the formation of ONAR framework and

its application for service oriented application
Integrating enterprise system has become an issue integration (SOAI).
of sharing information rather than transforming
information due to the increasing complexity and
the heterogeneity of the applications. The transi- IntroductIon
tion from application centric to integration centric
enterprise application integration (EAI) requires The integration of enterprise application such as
methods and technologies that will enable and enterprise resource planning (ERP) systems due
facilitate the definition of shared information. to their internal complexity, has lead many EAI
The use of ontologies semantic Web and tech- solution vendors to create solutions based on the
nologies can improve the existing EAI methods structure and the semantics of the application.
by providing a framework capable to define Nowadays the problem of integration is confronted
shared information. Ontologies-based enterprise with technologies (like Enterprise Java Beans) that
application integration (ONAR) framework uti- provide sophisticated and advanced techniques
lizes semantic Web technologies to define shared for technical interfaces.
information among heterogeneous systems. The The most recent orientations in enterprise
present chapter presents the utilization of ontolo- application integration (henceforth EAI) present
new techniques that provide methods to define and
exploit semantics of complex application. Still this
DOI: 10.4018/978-1-59904-648-8.ch006
The Utilization of Semantic Web for Integrating Enterprise Systems
definition is application centric and it cannot be two layers for the definition of concepts and the
shared among other heterogeneous applications. association to the application resources.
Tektonidis, Vontas, Hess, and Meschonat (2002) The semantic definition of the concepts is based
have stressed the problem of integration as a on the integration requirements. This implies the
problem of information sharing not as a problem concepts and their relation define the information
of adaptation that is very common case for ERP need to be exchanged. This “integration centric”
systems. approach follows the principals of modern enter-
In the technological level, service oriented prise application integration (EAI) specification
application integration (SOAI), as it is presented (Linthicum, 2004) contrary to traditional integra-
in Linthicum (2004), exploits the capabilities for tion approaches that aim to improve the definition
the functional description of Web services that are based on the semantics of the applications. Figure
used for the actual integration. This section focuses 1 presents the difference between traditional EAI
on the creation of an integration framework based (data transformation) and integration-centric ap-
on SOAI that utilizes semantic Web technologies proach.
(W3C, 2006) in order to enrich the semantics of The formation of an integration centric infor-
the exchanged information. mation definition EAI methodology required the
The approach used follows the ontologies utilization and the adaptation of semantic Web
based enterprise application integration (ONAR) technologies. ONAR methodology utilizes onto-
approach presented by Tektonidis et al. (2005) logical definitions for the definition of semantic
that utilizes Web ontologies to create semantic and uses association techniques in order to as-
conceptualizations of the business concepts that sociate concepts with the information repositories
exist inside an application. This conceptualization of the applications.
is used for the creation and the registration of the
Web services in a UDDI based registry. Figure 1. The difference between data transforma-
tion and semantic integration (Source: Capgemini)
utIlIZInG semAntIc
Web For enterPrIse
APPlIcAtIon InteGrAtIon
The need for the utilization of semantic Web de-

rives from the requirement for sharing information
instead of exchanging information. The adaptation
of ontologies for software engineering extends
further than the semantic definition of a domain.
Liu and Halper (1999) and Chiang (2001) proved
that the utilization of the ontologies requires certain
modifications in the principals of the frame-based
ontologies languages. The utilization of ontolo-
gies in order to define concepts for application
integration involves also characteristics of the
system that are integrated. However in order to
avoid definition of concepts based on the structure
and the logic of the applications we have defined
551
Integration centric semantics mitted to the financial system. The definition of

the concept “order” varies from case to case. For
The main objective of our research is to define example the integration between the production
the integration semantics based on the integra- planning system and the production system (plant
tion centric approach. The integration centric automation system) would include further infor-
approach presented in Tektonidis, Bokma, Oatley, mation regarding the machines and the personnel
and Salampasis (2005) aims to orient the design that should be involved excluding all financial
and development of the integration semantics information.
and components to the specific requirements of Therefore, according to the integration case
the integration case. the definition of a concept changes. The problem
The statement that integration centric approach is confronted today with two main approaches
is based on the specific requirements is a claim that are (a) application centric approach and (b)
made in comparison with existing techniques and shared vocabulary approach. Application centric
methodologies. As integration requirements, we approach provides techniques, mechanisms and
imply the information to be shared between the tools that facilitate the adaptation from one sys-
heterogeneous applications. tem to another. In the previous example the three
Figure 2 presents in a simplified example of applications would be adopted to the logic of the
how information about an Order is shared between financial system in order to retrieve information
four different applications. For the financial sys- regarding the submitted order. Technically, this is
tem, the “order” concept consists of the “customer accomplished with an application programming
name”, the “product name” and the “delivery date” interface (API) that the financial system offers to
property. However the other three systems do not external applications. This interface follows the
have internally the same definition about “order” application logic of the financial system and its
but they can understand all the concepts (order) utilization is actually an adaptation of the other
and properties (customer name, product name, applications to the financial system logic. The
delivery date) even if they can not process them definition of semantics following this case would
because there not a shared conceptualization of be defined based on the semantics of one system.
the order that every application can “understand”. The reduction to one “central” system semantics,
The definition of the order is not based on a however, exclude the semantics that exists to other
specific system but on the need to integrate these systems and not to the central application.
four different systems when a new order is sub- Shared vocabularies provide a common defini-
tion of all concepts that appear in all applications
that constitute the integrated system. Therefore in
Figure 2. Sharing information between heteroge-
the previous example the definition of the concept
neous applications
“order” would be more descriptive (having all the
attributes of all applications) and one for all the
possible integration cases. This would complicate
the conceptualization of the concept “order” in-
cluding semantics that in most of the integration
case would not be used.
Thus instead of defining the semantics based
on one application semantics or to the aggrega-
tion of all application semantics we propose
that the definition of integration case semantics
552
should be defined based on the requirements of may be divided in different level of abstrac-
the integration. The application centric approach tion. In the previous chapter, domain ontologies
defines differently the concepts according to enabled different levels of ontologies based to
the integration case. Following the example of their abstraction. An integration case in order to
Figure 3 the order concept should have different be complete may be very detailed in the concepts
conceptualization according to the integration that is based on and abstract to the rest of the
case. Thus in a case where the CRM system concepts that are only to be defined for reasons
requires information regarding the consuming of completion. In the example of Figure 3, for
habits of a customer from the financial system and integration case between the financial system
then the order conceptualization would contain and the CRM system information regarding the
only information about the customer name and customer purchasing habits may be very detailed
the product name while the integration between while other concepts like its order delivery date
the production planning system and the financial to be abstract. The abstract part of the ontology
system would include only the delivery date and is part that the application is not able to process
the product name. while the detailed part is data that the application
For the previous simplified example, it is uses in the integration case. Thus ontologies are
stressed that although inside an application the not meant to define the data to be exchanged, but
definition of a concept is univocal inside an inte- their semantics.
grated system the concept may have different The following section presents how ontologies
hypostasis according to the integration case. have been used as modeling technique to define
The requirement is how ontologies would im- semantic conceptualizations.
prove the multiple definition of a concept enable
different conceptualization for each integration
case. Ontologies according to Fensel (2004) are not sHArInG InFormAtIon
created to define data but to represent knowledge. usInG oWl ontoloGIes:
The knowledge of a domain does not follow the tHe onAr metHodoloGy
strict and precise definition of data structure that
is found in E-R diagrams or UML class diagrams. ONAR methodology follows the frame-based
Although ontologies provide an explicit and logic that modern ontology languages use in order
formal definition of a domain this definition to define conceptualizations. During our research,
Figure 3. The graph of a conceptualization ontology produced by ONAR concepts and services designer
553
Tektonidis et al. (2003) has found that languages Figure 3 presents the conceptualization of the
like the Web ontology (W3C, 2006) are capable in summary of Deed of Incorporation document
providing us with formal definitions for business used in public administration applications. This
concepts as OWL classes. ontology contains all the concepts (OWL classes)
In comparison with UML class diagrams, and their properties (OWL datatype properties)
OWL is more descriptive enabling the defini- as well as their relation (OWL object properties).
tion of semantics for the business concepts and The conceptualization is not based on the informa-
their attributes. The entities inside the ONAR tion system of the participating local authorities
conceptualization methodology inherit all the (Prefecture, Chamber of Commerce), but on an
features of the OWL classes. However OWL as ontology that is common for all authorities.
an ontology based language often permits the It can be argued that the graph does not rep-
abstract definition of domains. For EAI that can resent a formal OWL structure since for example
lead to the definition of conceptualizations that there are properties (OWL data-types) like the
cannot be reflected to real data since information property “ID” that is presented more that once
systems need to formalize abstract concepts into in the graph. Our framework uses OWL-Lite
data structures. as ontology language however for reasons of
The venture of having an ontology language presentation based on the end users feedback we
for defining information that is exchange between support that this kind of ontology representation
information systems has been confronted with the is more convenient and less complex to users that
definition of a restricted usage of OWL. In this are not familiar to ontology engineering.
attempt, graphical environment was developed in Summarizing the conceptualization capabili-
order the definition of the OWL ontologies to be ties of our framework we are using the OWL-Lite
constrained by certain rules that would prevent and more precisely: (See Figure 4)
from the false usage of OWL. ONAR concepts
and services designer (ONAR, 2006) enables the 1. Inheritance is permitted allowing some class
creation of conceptualization ontologies (OWL) features to be inherited from one class to
providing the business consultant a reduced OWL- another. The relation between the parent
Lite ontology (W3C, 2005). class and child classes follows the same rules
that both object-oriented engineering and
Figure 4. The generic structure of the association ontology
554
frame-based logic supports, however mul- defined from only one concept and vice versa.
tiple inheritance is not permitted. This This ensures that in the association of the logical
constraint is due to the fact that in complex model to the system resources will not create
information system polymorphism of some indeterminism.
concepts will unduly increase the complexity Both entities and properties are enriched with
of conceptualization. (a) notional properties like lexical descriptions,
2. Entities are allowed to have properties that (b) the ability to define the inverse of a concept
correspond to OWL datatype properties. A and (c) the ability that a property can potentially
property can belong to more than one entity be derived from another property. These enrich-
and is contrary to the usual principles of ments are used to increase the inference for the
object oriented modeling. This principle semantics of the concepts and their properties
derives from the frame-based logic of OWL inside a model following the OWL principles.
ontologies and extends the object oriented The most important reason for the utilization
engineering where a property belongs to of ontologies is the capability to define semantic
only one class. relations between the concepts of a domain. Me-
3. Relations between concepts can be defined cella and Batini (2000) introduced the concept
using of application wrapper as metalayer enables the
a. Functional property: Defines the redefinition of concepts of the application in order
“one to many” relation between the to be used from other heterogeneous applications.
instances (objects) of two concepts The lack of the inclusion of semantics limits the
b. Inverse functional property: capabilities of EAI approaches like application
Complementary to functional property wrappers and object oriented middleware (OOM)
in order to facilitate lexical descriptions techniques.
c. Symmetric property: Symmetric Defining information using data definition
properties are used to declare the one methods like UML class diagrams or entity rela-
to one elation between the instances tion (E-R) schemas the produced conceptualiza-
(objects) of two concepts. tions unavoidably follow the data structure of an
d. Transitive properties: Contrary to the application. This is the actual limitation that our
definition of OWL in ONAR framework needs to overcome. Software engineering
work Transitive properties are used for methods have limited capabilities for semantics
the division of complex concepts. For definition. For example, the semantics in the
example the concept “material” can relation between two concepts that both in UML
have properties regarding its financial or in E-R schemas do not have formal methods
features (prize, cost), logistics features to define.
like (size, weight, quantity) and so Our work has focused on creating semantics
forth. Using transitive properties the in the relations between concepts. The existing
concept can be divided to more than semantics of relations that are expressed in object
one concept according to its semantic. oriented class diagrams in UML or entities rela-
tions (E-R) diagrams pose limits to the semantic
However we have to limit the syntax of OWL, two concepts may have at the logical level.
which permits the existence in a relation of more Therefore adapting the four types of relations
than one concept as ranges and domains. In our (object relations) that are defined in OWL we
approach (in a relation) if the domains contain are achieved to increase the inference capabilities
more than one concept then the range should be of the conceptualizations. The relation between
555
concepts in this level represents semantic relation that enables the mapping of the conceptualization
between the concepts. These concepts may have ontology to the system repository.
a syntactic relation, however this is not requisite. The association ontology is related to the sys-
This detaches the definition of the concepts from tem structure and associates the concepts of one
their data realizations. Thus the definition of the conceptualization ontology to the system reposi-
concepts is based to the integration requirements. tory. The ontology enables the definition of the:
Concepts relations are defined based on their
domain and their range. The domain is consisted 1. Category of repository: Category of the
of set of concepts that subject of the relation. The repository defines level of the repository
range for a relation is the object or the objects of (e.g., whether is a flat file system or an
the relation. According to Wache, Vögele, Visser, RDBMS or an application server)
Stuckenschmidt, Schuster, and Neumann (2001) 2. Type of repository: Defines the technology
the definition of information based on ontologies of the repository (e.g. DCOM application
can be unlimited regarding the number of concepts server or JavaBeen application server)
that consist of a domain and the range. However 3. OWL ontology location: The conceptualiza-
in our research based on e-business (Tektonidis tion that the association ontology is related
et al., 2002), e-government (Tektonidis et al., 4. Mapping information: Information that
2005) and e-health (Tektonidis & Bokma, 2005) relates every logical concept to the resources
use cases the simultaneous existence of multiple of the repository like the access method or
domain and range ontology created problems technical information like the data type and
regarding the comprehension of the ontology as the size
well as in the association of the concepts to the
repository of the information system. With the disassociation of the conceptualiza-
tion from the repository of the system and the usage
mapping conceptualizations of the association ontology we have achieved:
to system repositories
1. To increase the reusability of the con-
Developing an approach that is independent from ceptualizations ontologies: The same
the application logic of each application neces- conceptualization ontologies can be used
sitate the existence of mechanisms and technique to different information system.
to associate logical conceptualizations (OWL 2. To reduce cost of maintainability when
ontologies) to the real application repository. In changes happened in to the application
addition, our approach is independent also from repository
the type of the repository (whether is a RDBMS 3. To enable the creation of common vocabular-
or an application server). ies that can be shared between heterogeneous
In our attempt to enrich OWL with elements applications
that would enable as to associate the conceptualiza-
tion to the system we have discovered that these In our work so far we have achieved the ap-
enrichments would create a complex ontology file plication of the ONAR framework to relational
that will carry both semantic and syntactic ele- databases, ODBC compliant repositories and ap-
ments. Another important issue was the portability plication server based on Web services (WSDL).
of the ontologies from different kind of systems. In the near future we aim to include other type
Therefore we have created an association ontology of repositories.
556
APPlIcAtIon InteGrAtIon bAsed 3. Design phase: The definition of the software

on semAntIc Web servIces instance takes place. The integrator defines
the input and the output list based on the
onAr Integration Process conceptualization ontologies.
4. Deployment phase: Where the software
The ONAR integration methodology has been instance is deployed (creation of the ser-
developed during this research as the proposed vice model) and becomes an integration
procedure for integrating application following the component.
ONAR approach based on the prototype system. 5. Publication phase: The profile of the soft-
The methodology guides integrators that are not ware component is published in the registry.
familiar with the ONAR approach, ontologies
or the technological and technical issues of the The definition, once in every integration case,
prototype system. the integration ontology and the fact that the con-
Figure 5 presents the phases of the integration ceptualization ontologies derived from the initial
process that are the following: integration ontology facilitate the insertion of new
applications to the integrated system (Tektonidis
1. Conceptualization phase: In this phase the et al., 2005). Thus the prototype system should
requirements of the integration are defined. automate the procedure in order the development
The integrator defines the concepts and their of the software instance to be leveraged in terms
relations that comprise the integration ontol- of complexity and maintainability.
ogy. During this phase also the integrator
defines the conceptualization ontologies Designing Web Services Based
based on the applicable semantics for each on the Conceptual Schema
application. The ontologies development is
based on the principals of the development The entire Web services development cycle includ-
phase of the methontology (Gòmez et al., ing the design, the deployment and registration of
1997) in order to define the ontologies. the Web services to the universal description, dis-
2. Association phase: The conceptual schema covery and integration (UDDI) registry is related
(conceptualization ontology) is associated to the conceptualization ontology. The design of
to the legacy system resources (data). the software instance is the process where ele-
ments of a conceptualization are used to define
Figure 5. ONAR integration methodology
557
Figure 6. ONAR service model
the input and the output of the Web service. The onAr Framework Architecture
design of a Web service consists of its syntactic
definition and its semantic description. The requirement and the objective that have been
The syntactic definition defines the input and set in the previous sections required the develop-
the output as well as the behavior of the Web ment of several components that comprise ONAR
service. This definition is based one conceptualiza- framework architecture, presented in Figure 7.
tion of the conceptual schema. Starting from one ONAR framework implementation was based
basic entity the Web service designer can use all on the methodology and in the existing state-of-
the concepts that are necessary for his definition. the-art technologies. The selection of the tech-
The syntactic definition (ONAR service model) nologies that materialize the conceptual compo-
of the service contains all the semantic and syn- nents of the architecture and it is not unique since
tactic relations between the basic entity (primary other technologies may substitute them.
concept) and the secondary concepts. However the conceptual components role and
The ONAR service model (Figure 6) is an XML the architecture materialized into a framework the
document contains the functional description of the ONAR methodology. A summary of the role and
service based on the conceptualization ontology. functionality of the components are the following:
The reason for having a new definition docu-
ment for the syntactic definition of the service,
apart from maintenance purposes, is to increase Figure 7. ONAR prototype system architecture
the portability of the Web services. We have to
implement the necessary functionality that enables
us to automatically generate the source code (C#)
of a Web service based only to the conceptualiza-
tion ontology the association ontology and the
ONAR service model. Therefore, if two applica-
tions share the same conceptualization ontologies
but have different structures (different association
ontologies) they can exchange Web services.
558
1. Concepts and services designer: The ONAR Ontology Design Tools

graphical environment that enables the inte-
grator to design and develop of the software The definition of the integration and conceptu-
instances and enables the administrator to alization ontologies required a design tool that
remotely administer and maintain the ONAR would enable the design of the ontologies. During
server the research we have reviewed several ontology
2. ONAR deployment server: The component design environments and tools like Protégé and
that undertakes the administrative tasks of WebODE.
the ONAR server As aforementioned, integration and conceptu-
1. ONAR publication server: The com- alization ontologies follow a specialized version
ponent that enables the administration of OWL Lite. In addition, the facilitation of the
of the ONAR repository (enriched creation of association ontology required an envi-
UDDI) ronment that would simplify the procedures in the
2. ONAR repository: The semanti- association phase. Therefore, we have developed
cally enriched UDDI based service a graphical environment that integrates the five
repository phases of the ONAR integration methodology.
3. Semantic repository: The pool of ONAR concepts and services designer (Figure
derived ontologies (integration and 8) enables:
conceptual) used inside an integrated
system a. The graphical design and syntactical valida-
tion of the integration conceptualization
Additional information for the ONAR pro- ontologies
totype components regarding technologies and b. The development of the association ontolo-
specifications is included in the ONAR Web gies and the necessary repository validity
site. The following sections analyse the tools and checking
components of the ONAR framework. c. The design and deployment of the software
instances
d. The design and publication of the software
instances profiles
Figure 8. ONAR concepts and services designer
559
We present in the following sections some of It can be argued that the proposed ontology
the features of the integrated environment however representation is closer to the UML class diagram;
further information can be found in the project however it provides a simplified view of the
Web site: http://research.altec.gr/ONAR. ontology without limiting its capabilities.
Ontology Representation software Instance creation
One of the most important issues in order ONAR Software instances are the software components
framework to facilitate the creation of ontologies that realize integration rules based on ontological
was the representation of ontology. OWL is a definition of the integration case. For the ONAR
frame-based language however according to the prototype system software instances are material-
W3C definition data type properties do not belong ized using Web service technology. The selection
to one class but to a collection of classes. of Web services over other technologies (like
A graphical representation of this principal CORBA, DCOM or EJB) has been based to the
would complicate the design and representation fact that Web services follow the message-oriented
of the ontology. During our research (Tektonidis logic. There are many pros and cons to this selec-
et al., 2005) the representation that is closer to tion, however it remains a technological issue and
the UML class diagram than the ontological the evolution of technologies may in the future
representation used from other tools (Protégé provide better solutions.
OWL plug-in) simplified the representation of The software instance is divided in two parts:
the ontology.
Every class in Figure 9 appears to have a 1. The software instance profile (informative
set of properties that belong only to the class. part) that contains the semantic description
For example, the “ID” property that appears in for the use of the software instance based
“production order”, “production” and “material” on the integration ontology. The structure
is one datatype property that has as domain the of the software instance profile follows the
three classes (Figure 10). principal of OWL-S profile ontology. The
Figure 9. Ontology representation using ONAR concepts and services designer
Figure 10. The definition of the “ID” property
560
Figure 11. ONAR software instance service model
OWL-S profile has been extended in order ONAR Repository

to be integrated with the repository that fol-
lows the UDDI standard that is described ONAR prototype system has a service oriented
later on. architecture (henceforth SOA) therefore in order
2. The service model (performative part) that to leverage the capability of this technologies
includes information regarding the function- required a repository that would enabled the pub-
ality of the software instance. The service lication of the software instance descriptions. Al-
model (Figure 11) includes the parts of the though ONAR repository is not directly associated
conceptualization ontology that are related to research issues, it is a core component because
to the software instance as well as their as- it serves internal purposes of the framework by
sociation with the repository of the applica- reducing the effort needed for the discovery of the
tion. The service model has XML structure appropriate software instance. ONAR framework
and it can be transferable (Tektonidis & internal architecture is service oriented and the
Bokma, 2005) to applications that contain existence of a UDDI based repository is considered
the conceptions in their conceptualization mandatory for the description and discovery of
ontology. This way the heterogeneous ap- the software instance (Web services).
plications can also share software instances
with the same semantics. The transfer of the
service model (and thus a software instance)
Figure 12. The portability of the software instances
is accomplished with the substitution of the
association ontology element of the source
application with the elements of the target ap-
plication. Figure 12 presents the process than
enables the transport of a software instance
between two heterogeneous applications.
561
Figure 13. Integration of the OWL-S profile ontol-

enriched UDDI enables semantic descriptions of
ogy with UDDI (Paulocci et al., 2001)
the services.
ONAR framework used as base the work of
Paulocci et al. (2001) in order to create a seman-
tic repository to register the services. ONAR re-
pository is an UDDI based repository that uti-
lizes the semantics of OWL-S profile ontology.
However in order to exploit the advantage of the
semantic description of the software instance
ONAR repository includes more detailed descrip-
tions of the inputs and outputs of the services.
Figure 14 presents the extensions of the OWL-
S profile in order to support the semantics of the
ONAR framework. More specifically OWL-S
input and OWL-S output classes have been en-
riched with four characteristics namely (a) domain
(b) range (c) sub property of and (d) cardinality
in order to be able to include the semantics of the
software instances.
Thus ONAR repository based on the discovery
mechanisms enables semantic searches of the
To this respect Paulocci, Kawamura, Payne, necessary software instances that can be used in
and Sycara (2001) integrated the ontological ser- complex integration cases.
vice representation definition of OWL-S profile
with UDDI. The UDDI has been enriched with
a set of TModel in order to include the concepts
of OWL-S profile ontology (Figure 13). The
Figure 14. Extensions in OWL-S profile (Source: Tektonidis et al., 2006)
562
conclusIon And FurtHer Work Corcho, O., Fernández-López, M., & Gómez-
Pérez, A. (2003). Methodologies, tools and
The utilization of ontologies for the conceptual- languages for building ontologies. Where is
ization of existing system based on an integra- their meeting point? Data & Knowledge En-
tion centric approach requires validation and gineering, 46(1), 41–64. doi:10.1016/S0169-
evaluation of our framework. Thus we aim first 023X(02)00195-7
to validate whether ONAR framework can be
Dogac, A., Kabak, Y., & Laleci, G. (2004). Enrich-
applied to a wide variety of Information system.
ing ebXML registries with OWL ontologies for
In this direction over this last year, we have been
efficient service discovery. In Proceedings of the
collecting feedback from the registered user of the
14th International Workshop on Research Issues
framework. Currently we have examined e-busi-
on Data Engineering, Boston.
ness e-health (Artemis, 2005) and e-government
(EU-Publi.com, 2005) integration cases. Eu-Publi.com IST-Project (2005). Facilitating
In addition, we aim to create an evaluation cooperation amongst European public adminis-
framework that will enable us to evaluate ONAR tration employees. Retrieved February 13, 2007,
to existing approaches based on its (a) maintain- from http://www.eu-publi.com
ability, (b) complexity, (c) portability and (d)
Linthicum, D. (2004). Next generation applica-
adaptability in order to assess quantitatively its
tion integration: From simple information to web
performance to other integration frameworks.
services. Addison-Wesley.
Finally, we are currently working on technical
improvements and enrichments of the framework Liu, L., & Halper, M. (1997). Incorporating se-
mainly for its adaptability feature in order for mantic relationships into an object oriented data-
ONAR to be used from different information base system. In Proceedings of the 32nd Hawaiian
systems. International Conference on Systems Science.
Mecella, M., & Batini, C. (2000) Cooperation
of heterogeneous legacy information systems: A
reFerences
methodological framework. In Proceedings of the
W3C (2006). Semantic Web description. Re- 4th International Enterprise Distributed Object
trieved February 13, 2007, from http://www. Com-puting Conference, Makuhari, Japan.
w3.org/2001/sw/ ONAR. (2006). Ontologies based enterprise appli-
Artemis, I. ST-Project (2005). A Semantic web cation integration framework. Retrieved Februaty
service-based P2P infrastructure for the interoper- 13, 2007, from http://research.altec.gr/ONAR
ability of medical information systems. Retrieved OWL. (2006). W3C OWL 1.0 syntax specification.
February 13, 2007, from http://www.srdc.metu. Retrieved February 13, 2007, from http://www.
edu.tr/webpage/projects/artemis/ w3.org/TR/owl-ref/
Chiang, C. (2001). Wrapping legacy systems for Paolucci, M., Kawamura, T., Payne, T. R., &
use in heterogeneous computing environments. Sycara, M. (2002). Semantic matching of web
Information and Software Technology, 43(8), services capabilities. In Proceedings of the In-
497–507. .doi:10.1016/S0950-5849(01)00160-4 ternational Semantic Web Conference (ISWC)
(pp. 333-347).
563
Ratnakar, V., & Gil, Y. (2002). A comparison of Tektonidis, D., Vontas, A., Hess, U., & Meschonat,
(Semantic) mark-up languages. In Proceedings of J. (2002). Handling the shop-floor of an industry
the 15th International FLAIRS Conference, Special through ERP systems: A functional integration
Track on Semantic Web, Pensacola, Finland. model. In Proceedings of the International Con-
ference on E-business & E-Work 2002, Prague.
Tektonidis, D., & Bokma, A. (2005). The utiliza-
tion of ontologies for e-health systems integra- Themistocleous, M., Irani, Z., O’Keefe, R., & Paul,
tion. In Proceedings of the 10th International R. (2001). ERP problems and application integra-
Symposium on Health Information Management tion issues: An empirical survey. In Proceedings
Research, Thessaloniki, Greece. of the 34th Hawaiian International Conference on
Systems Science, Big Island.
Tektonidis, D., Bokma, A., Oatley, G., & Salam-
pasis, M. (2005). ONAR: An ontologies-based ser- Wache, H., Vögele, T., Visser, U., Stuckenschmidt,
vice oriented application integration framework. H., Schuster, G., Neumann, H., & Hübner, S.
In Proceedings of the 1st International Conference (2001). Ontology-based integration of informa-
on Interoperability of Enterprise Software and tion: A survey of existing approaches. In Proceed-
Applications, Geneva, Switzerland. ings of the IJCAI-01 Workshop: Ontologies and
Information Sharing, Seattle, Washington.
This work was previously published in Agent and Web Service Technologies in Virtual Enterprises, edited by Nicolaos Proto-
geros, pp. 97-114, copyright 2008 by Information Science Reference (an imprint of IGI Global).
564
Section III
Tools and Technologies
This section presents extensive coverage of the technology that informs and impacts enterprise information systems.
These chapters provide an in-depth analysis of the use and development of innumerable devices and tools, while also
providing insight into new and upcoming technologies, theories, and instruments that will soon be commonplace.
Within these rigorously researched chapters, readers are presented with examples of the tools that facilitate and
support the emergence and advancement of enterprise information systems. In addition, the successful implementa-
tion and resulting impact of these various tools and technologies are discussed within this collection of chapters.
566
Chapter 3.1
Data Warehouse Maintenance,
Evolution and Versioning
Johann Eder
University of Klagenfurt, Austria
Karl Wiggisser
University of Klagenfurt, Austria
AbstrAct of all, being aware of their existence. Second,

once you know that a change took place, it is
Data Warehouses typically are building blocks of important to know which change (i.e., knowing
decision support systems in companies and public about differences between versions and relations
administration. The data contained in a data ware- between the elements of different versions). For
house is analyzed by means of OnLine Analytical data warehouses this means that changes are
Processing tools, which provide sophisticated fea- identified and represented, validity of data and
tures for aggregating and comparing data. Decision structures are recorded and this knowledge is used
support applications depend on the reliability and for computing correct results for OLAP queries.
accuracy of the contained data. Typically, a data This chapter is intended to motivate the need
warehouse does not only comprise the current for powerful maintenance mechanisms for data
snapshot data but also historical data to enable, warehouse cubes. It presents some basic terms
for instance, analysis over several years. And, as and definitions for the common understanding and
we live in a changing world, one criterion for the introduces the different aspects of data warehouse
reliability and accuracy of the results of such long maintenance. Furthermore, several approaches
period queries is their comparability. Whereas data addressing the problem are presented and classi-
warehouse systems are well prepared for changes fied by their capabilities.
in the transactional data, they are, surprisingly,
not able to deal with changes in the master data.
Nonetheless, such changes do frequently occur. IntroductIon
The crucial point for supporting changes is, first
The standard architecture for data warehouse
DOI: 10.4018/978-1-60566-756-0.ch010 systems are multidimensional databases, where
Data Warehouse Maintenance, Evolution and Versioning
transactional data (cell values) are described in at that time. Comparing the numbers of 1990 and
terms of master data (dimensions and dimension 1991, where the organization itself did not change,
members). Whereas today’s commercial systems may indicate a massive increase of inhabitants.
are well prepared to deal with changes in the In reality, the 1991 number also contains the 16.4
transactional data, they are, surprisingly, not able million people of former East-Germany. Eurostat,
to deal with changing master data in a satisfactory for instance, takes this into account, and presents
way. Nonetheless, such changes frequently occur numbers of the united Germany also for the years
(restructuring in organizations, new laws, mergers before 1991. Another example for an unclear
and acquisitions, product portfolio restructuring, inclusion are the Baltic countries or Slovenia.
etc.). All these changes have to be represented in They did not even exist before 1991, but were
the information systems, and thus, must some- parts of other countries, which, of course, never
how be modeled also in the data warehouse. For were parts of the European Union. An alternative
data warehouses the adequate representation and to presenting such “adjusted data” is to display
treatment of such changes is even more crucial the “historical truth”, i.e. include the numbers
than in standard database applications, since data of different countries only after they joined the
warehouses are intended to represent also histori- European Union. This may make sense in some
cal data which – changes occurring – might be situations, in others, such results may be useless.
quite incompatible. An example demonstrating the effect of
A simple example illustrating the problem of changing semantics could be to retrieve the
missing data is querying the number of inhabit- Gross National Product of the countries in the
ants in the European Union for the last 25 years. European Union from 1983 to 2008. Besides
This query seems rather straightforward and the the problems induced by the structural changes
numbers should not leave much space for interpre- described above, i.e. whether and how to include
tation. But, one has to be aware of some changes: numbers for a specific country, this query illus-
First of all, the geopolitical entity “European trates the changing semantics problem: As of
Union” only exists since 1993, succeeding the 1999 and 2002, a common European currency,
“European Community”, which itself was origi- the Euro, was introduced as deposit currency
nally named “European Economic Community”. and cash money respectively, in many – but
Furthermore, in the considered period (1983 to not all – of the member countries. Thus, before
2008), the European Union grew from 12 to 27 1999 the Gross National Product of different
members. Finally, with the reunification of East- countries was expressed in the local currency,
and West-Germany in 1990 one of the member but as of 1999 it is given in Euro. Before 1999
countries had a massive internal reorganization. So for comparing the GNP of different countries,
if querying the number of inhabitants from 1983 to it is obvious that the numbers must be brought
2008, how can the resulting numbers be compared? to a common base, i.e. the same currency, to
When querying this data from the Eurostat website, be comparable. But what about statistics for a
one has to choose the “geopolitical entity” (EU-27, single country? For Austria, 1 Euro exchanges
EU-25, one or more counties, …) for which the 13.7603 Austrian Schillings. So, someone com-
data should be retrieved. If, for instance, EU-25 paring the Austrian Gross National Product from
is chosen, the population for these 25 countries is 1990–2006 without considering the Euro would
returned also for the years before their member- notice a giant retracement in the year 1999. But,
ship. But of course, the overall sum of returned of course, someone who knows about the Euro
inhabitants for the year 1987 does not match the can divide each value given in ATS by 13.7603
real number of people living in the European Union and then compare the values.
567
Besides such simple “unit changes” there time consuming and also error prone, espe-
may be also more complex semantic changes for cially if the affected dimensions are large
dimension members. Consider a query analyzing and the changes are rather small.
the unemployment rate in the European Union. 3. Dealing with the influence of structure
Not only that it is calculated in different ways for changes on the cell data: Some structural
various countries, the calculation mode has also changes may have heavy influence on the
been changed several times in the last few years, cell data. One of the main reasons for data
for instance whether people who are attending warehouse maintenance is comparability of
coursed offered by federal employment offices are cell data. Now comparing cell data stemming
counted as unemployed or not. Such calculation from before and after a structural change
methods may be contained in the data warehouse may be very complex or even impossible,
definition as formulae for a certain member. because of the influence of the structure
Generally, the three basic aspects of change changes on the cell data. In this context two
management in data warehouses can be identified major problems can be identified
as follows: a. Missing Data: If elements are inserted
or removed from the structure, cell data
1. Being aware about changes happening: may not be available for the whole pe-
First of all, to be able to manage changes, it riod of analysis (e.g. missing data for
is necessary to be aware of their existence. new countries in the European Union).
This awareness can easily be seen from two b. Incorrect Data: If structure elements
typical reasons for data warehouse structure are changed, their semantics may
changes. The first reason are changes in the change. This could have an influence
real world. that is represented by the data on the calculation of cell values. Thus,
warehouse, for instance creation of a new if comparing cell values from before
department, a merge of different depart- and after the change, equal values may
ments, or new countries joining the European have a different meaning and vice versa
Union.. The second reason for modifications (e.g. different methods for calculating
in the data warehouse are changing require- the unemployment rate).
ments, for instance analyzing not only the
turnover in a company, but also the gain, or These examples illustrate the problems induced
keeping track of unemployment rates, which by changing structures on a very simple level.
were not recorded before. Froeschl, Yamada and Kudrna (2002) call this the
2. Identifying the changes in the system: problem of footnotes in statistics, i.e. many values
As today’s typical data warehouse systems have to be tagged with their correct semantics.
typically do not support changing structure When being aware of such semantic and structural
data, they also may not be able to provide changes, interpreting “strange” results may be
information about them, for instance some cumbersome but possible. But if someone does
sort of a change log, even if the happened, not even know that there were changes, analyzing
e.g. by some automatic ETL process recre- query results may be impossible or, even worse,
ating the data warehouse from its sources. lead to wrong decisions. One possible solution to
The administrator is then, being aware that this problem is data transformation, i.e. viewing
changes happened, in charge of finding the old data under new structures or vice versa. One
modifications and correctly dealing with may define the semantics of a query, and the system
them. Executing this task manually is very has – provided that the necessary information is
568
available – the ability to include only the desired as the real world tends to change, such changes
data, exclude undesired data and adjust the data must also be represented in the cube structure.
to the desired structure and semantics. Changes can happen on the schema level, i.e.
dimensions or categories are changed, or on the
instance level, i.e. the members are changed.
bAckGround: tHe Figure 1 shows examples for schema and in-
dAtA WAreHouse stance changes. It contains three different version
mAIntenAnce Problem of a cube structure in a car dealer’s data warehouse.
For sake of simplicity, only the Cars dimension is
The standard architecture and modeling approach depicted. The dealer sells different car models of
for data warehouses is the multidimensional data different brands. Each model has assigned a user
cube. A multidimensional data cube consists of a defined attribute Engine Power. Traditionally, for
set of Dimensions, each of them comprising a set German cars this is given in horsepower, whereas
of Dimension Members (also called simply Mem- for English models it is given in kilowatt. In the
bers). Dimensions are hierarchically organized second version there are various instance changes:
into a set of Categories or Dimension Levels, the new model BMW1 is introduced and Silver
each of them having assigned a set of members. Spirit is renamed to Silver Spirit II. In version 3
The members themselves are also hierarchically there is also a schema change. Due to the merge
structured, accordingly to the categories they are of BMW and Rolls Royce on the one hand and
assigned to. Dimensions and categories define Mercedes and Chrysler on the other hand, a new
the schema of the cube, whereas the members dimension level is introduced. Of course for this
are called the instances. Schema and instances new level also new members are created and the
together define the cube structure. Selecting one existing members are relocated accordingly. The
member from each dimension defines a Data Cell, brand Puch is discontinued, the model attached
containing either a Cell Value or a NIL value. to it is now sold under the brand of Mercedes.
Although there is no standardized terminology The new brand Chrysler with one car model is
in data warehousing till today, these basic terms introduced, whereas the Phantom V is removed
are widely accepted. The terms Fact and Measure from the product portfolio. The attribute for the
are sometimes used with different semantics. In engine power is unified to kilowatt. All these
this chapter they are used as follows: A fact is a structure modifications are due to changes in the
dimension member in the Fact Dimension, repre- data warehouse’s application domain. A modifica-
senting a certain subject of analysis (for instance tion resulting from a changed requirement can for
Turnover). The term measure is used synonym instance be the introduction of a member profit
to cell value. in the facts dimension, depicting the car dealers
wish to keep track of his profit.
Problems of cube From the view of structural changes, these
structure changes reorganizations do not cause any problems. But
what about the data? Table 1 shows the cell data
It is obvious that the cube structure is modeling a for the cube in the years 2005, 2006 and 2007, i.e.
certain part of the real world. For instance, depart- for the Versions 1, 2 and 3 respectively. A possible
ments and divisions of a company will somehow be chart resulting from this data without considering
represented in a cube used in that company. Thus, changes may look like shown in Figure 2. Without
there may be a Company dimension comprising knowing about the changes, interpretation of this
members for the departments and divisions. And, chart is not possible.
569
Figure 1. Schema and instance changes in the example company
Table 1. Cell data for the example company
Cars sold Cars sold Cars sold

2005 2006 2007
All Cars 111 All Cars 145 All Cars 181
BMW 40 BMW 67 BMW&Rolls-Royce 71
BMW 3 30 BMW 1 20 BMW 67
BMW 5 10 BMW 3 35 BMW 1 22
Rolls-Royce 7 BMW 5 12 BMW 3 34
Silver Spirit 3 Rolls-Royce 6 BMW 5 11
Phantom V 4 Silver Spirit II 3 Rolls-Royce 4
Mercedes 44 Phantom V 3 Silver Spirit II 4
C 320 34 Mercedes 49 Daimer-Chrysler 110
CLK 320 10 C 320 38 Mercedes 75
Puch 20 CLK 320 11 C 320 40
Modell G 20 Puch 23 CLK 320 11
Modell G 23 Modell G 24
Chrysler 35
Voyager 35
570
Figure 2. Example charts for changed structure
levels of the maintenance Problem from simple reaggregation to complex data trans-
formations, because, for instance, the unit of a fact
As defined above, the cube’s structure consists of has changed. For the example given above, a data
schema and instances. A data warehouse mainte- adjustment may be the recalculation for the brand
nance system must be able to keep track of changes Mercedes, representing that Model G is now sold
in both of them. Thus, on the schema level it must under this brand.
provide operations for Insertion, Deletion and
Change of dimension and categories. Category Impacts of data Warehouse
changes are, for instance, addition or deletion of changes on cell data
user defined attributes. Also the hierarchical rela-
tions between different categories may be subject The different types of changes in data warehouse
to change. On the instance level, the system must structures affect the assigned cell data in different
provide operations for the Insertion, Deletion ways. The following impacts on the cell data can
and Change of dimension members, as well as be identified:
operations for changing the hierarchical relations
between the dimension members. • Additional Cell Data: Additional data is
Recording and managing structure changes most times expressed by new structure ele-
is only one aspect of the maintenance problem. ments not depending on previously existing
The cell data contained in the data warehouse elements, either on the schema (dimension,
depends on the structure. Thus, modifications categories) or on the instance (dimension
of the structure may lead to inconsistencies in members) level. The problem with ad-
the cell data. So it may have to be adjusted to be ditional cell data is that it is oftentimes
consistent again. Such adjustments could range not available for the past thus, leading to
571
missing data. An example could be new sions available or not. This also applies to data
countries joining the European Union. warehouse maintenance. With Data Warehouse
• Removed Cell Data: When dimension Evolution every modification (or every set of
members or categories are completely re- modifications applied in one transaction) of the
moved from the data warehouse structure, structure leads to a new version of this structure
also the assigned cell data should no longer and the previous version is lost. The contained cell
be part of the cube. For instance, data from data has to be transformed to be consistent with the
countries leaving the European Union new structure. Queries can only be executed based
should not be contained in statistics any on the newest structure. With Data Warehouse
longer. Versioning also every modification (or every set
• Restructuring of Cell Data: Restructuring of modifications applied in one transaction) of the
cell data can happen in various forms: ag- structure leads to a new version of this structure,
gregation hierarchies may change, new ag- but the previous version is kept available and may
gregation levels may be created, dimension be restored. Existing cell data does not need to
members may be split or merged, the unit be adapted, but can be stored following is origin
of a fact may change. All these restructur- structure version. This enables queries spanning
ings have in common that the cell data for multiple structure versions.
the new version can – to a certain extent Comparing these two possibilities reveals
– be calculated from the old structure. Not advantages and disadvantages for both of them:
dealing with such changes may lead to When using the evolution approach, one does
incorrect data. Examples for this type of not need to keep track of old versions, which
changes may be the split of a country in reduces data management effort and as all data
the European Union into two independent is adapted to the current version, no adaptations
countries, or the changing the currency. have to be done during runtime, which means a
• Change of Cell Data Calculation: Cell better query performance, compared to the ver-
data may also be calculated from various sioning approach. On the other hand, evolution
external data sources by some given for- lacks the possibility of multiversion queries and
mulae. Such calculation rule may change. as the data has to be transformed, change opera-
Not dealing with such changes may again tions may take quite a long time. The versioning
lead to incorrect data. An example for this approach allows multiversion queries, and data
type may be changing the calculation rule does not have to be transformed when applying
for the unemployment rate. modifications. On the other hand, when running a
• Changing Cell Data: Typically, cell data multiversion query, data may have to be adapted,
contained in a data warehouse is non-vol- which leads to worse performance, compared to
atile, thus not changing any more (Inmon the evolutionary approach. Furthermore, keeping
2005, p. 29). Therefore changing cell data track of past versions is a non-negligible data
is typically not considered during of data management effort.
warehouse maintenance.
Aspects of time in the
data Warehouse evolution data Warehouse
and versioning
Time is a key issue in data warehousing, because
When it comes to changing systems, one question when supporting the decision process, comput-
always arising is whether to keep previous ver- ing historical trends may be necessary (Rizzi and
572
Golfarelli, 2006). In data warehouses one can nance where all past versions of the data warehouse
distinguish three different kinds of time: are kept available. Dealing with changes on the
data level, mostly insertion of new data, is not
1. Valid Time of Cell Data: In a multidimen- part of data warehouse maintenance, but part of
sional cube usually the Time dimension is a data warehouse’s normal operation. (Eder and
used to represent the history of changing Wiggisser, 2009, p. 1)
transaction data. Cell data, which is related
to a specific dimension member from the
time dimension is valid at the time point or tHe comet temPorAl
interval represented by that member. dAtA WAreHouse
2. Transaction Time of Cell Data: Based on
the assumption of nonvolatility, for most Eder and Koncilia (Eder & Koncilia, 2001; Eder,
application the transaction time of cell Koncilia, & Morzy, 2002) suggest the COMET
values, i.e. the time when data is current in Metamodel for temporal data warehouses. This
the data warehouse, is not considered to be model is based on the multidimensional data
relevant. When this assumption is violated model. All elements (dimensions, categories,
and transaction data which is already stored dimension members, …) and relations between
in the data warehouse is changed afterwards, them get assigned a timestamp [Ts; Te] where Ts
transaction time bay become an important identifies the first point in time where the given
aspect, as it allows traceability for query element is valid, and Te defines the last point in
results (Rizzi and Golfarelli, 2006). time of the validity of the element. An ending
3. Valid Time of Structure Data: Data ware- timestamp of NOW defines elements for which
house structures need to be maintained. the end of validity is yet unknown. Operations
Structure elements may be inserted, deleted for inserting, deleting, and updating structure
or changed. To be able to determine the data elements are provided, as well as operations for
warehouse’s structure at a given point in changing hierarchical relations between these
time, structures have to be assigned a valid elements, for instance creating a new aggregation
time. Today’s data warehouse systems typi- hierarchy. On the member level, also the complex
cally provide very little support for changing operations Split – i.e. a single dimension member
structures. is separated into several members – and Merge
– i.e. several dimension members are combined
Data warehouse maintenance mostly deals with into a single one – are supported.
the third aspect, i.e. valid time of structure data. An advantage of this approach is its indepen-
Subsuming the above the following definition dence from the data warehouse system in use
can be given for data warehouse maintenance, (Eder, Koncilia, & Wiggisser, 2006). The ver-
evolution and versioning:The process and meth- sioned structure data and the metadata is stored
odology of performing changes on the schema independently from the data warehouse system
and instance level to represent changes in the in a relational database, thus virtually any data
data warehouse’s application domain or require- warehouse system can be supported with tempo-
ments is called Data Warehouse Maintenance. rality. As COMET works independently from any
Data Warehouse Evolution is a form of data particular data warehouse system, the changes log
warehouse maintenance where only the newest between two versions may not always be readily
data warehouse state is available. Data Warehouse available to be integrated in the COMET database.
Versioning is a form of data warehouse mainte- For that purpose, the framework provides a change
573
detection mechanism based on directed acyclic far more complex than just such “unit transforma-
graphs. COMET compares the graph representing tions”. For instance in the case of a split, there
the last known structure version in the database may exist different weighting factors for each
to the graph representing the current version in successor, defining the split ratios.
the data warehouse systems and suggests and edit The set of all these transformation functions
script, which represents the differences between may be modeled as a matrix or as a directed acy-
the two graphs. Such a change detection can only clic graph (Eder & Wiggisser, 2008). The matrix
be semiautomatic, because to definitely confirm approach allows simple and intuitive modeling
the calculated differences, external knowledge of the transformation process. The cell data is
which can only be provided by a human user is represented as n-dimensional matrix, transforma-
necessary. tion data is modeled as two-dimensional matrix
The COMET approach supports versioning for each dimension. The transformation is done
and therefore all previous structure versions are by standard matrix multiplication. As many data
recorded. If an element changed at time Ti, a new cubes are only sparsely populated and also the
version of this element with valid time [Ti; NOW] amount of changes is assumed to be rather small,
is created in the relational database. The previous this matrix multiplication is not the most efficient
version is updated by assigning the ending time- way for transformation. The graph approach
stamp Ti-1. For creating a new structure element is intended to give a more efficient representa-
at time Ti, the respective data is inserted into the tion. Cell data and transformation functions are
relational database and given the timestamp [Ti; represented by nodes and edges, respectively.
NOW]. Deleting a structure element at timepoint Thus, only populated cells have to be included.
Ti is done by changing the ending timestamp from As each edge stands for a multiplication and each
NOW to Ti-1. It is important to note that no data is node stands for a cell, it is desirable to have the
physically deleted from the relational database. transformation graph minimized, i.e. containing
All elements with the same end of validity the least possible number of nodes and edges. For
build a so called structure version, which repre- independent transformations – that are transforma-
sents a version of the cube valid at a certain point tions which do not affect different versions of the
of time. Structure versions are contiguous, thus same member – it is possible to change the order
there are no holes in the version history of a cube. in which these transformations are executed, thus
A particular version of a structure element may to minimize the graph.
be part of several structure versions, but within The architecture of a prototypic implementa-
a single structure version, only one version of an tion of the COMET Metamodel, called DWT is
element can be valid. shown in Figure 3[REMOVED REF FIELD].
Between different versions of a dimension The whole system is based on the backend da-
member so called transformation functions are tabase, which contains the versioned structure
defined, which allow to transform the assigned information, the transformation functions and the
cell values between different structure versions cell data. DWT is independent from any specific
with a certain weighting factor w. Such a trans- data warehouse system. For each system to be
formation may for instance be the transformation supported, an import and export interface has to
from Austrian Schillings into Euro for all money be implemented. These interfaces are responsible
related members in the fact dimensions, for which for reading/writing structure and cell data from/
new versions have been created in the year 2002. to the particular system. Structure data read from
In this case the factor w would have a value of an external data warehouse system is compared
0.07267. Of course such transformations may be to the last version stored in the backend database
574
(if available) by the difference detection compo- otHer APProAcHes AddressInG

nent. If there is some external knowledge about tHe mAIntenAnce Problem
the changes available in computer readable form
it can be imported as well. With the differences There are several other approaches dealing with
generated the version management component the data warehouse maintenance problem. Most
creates a new structure version and stores it into of them are scientific ones, but two of them are
the backend database. When reestablishing a included in commercial products. In this section
particular structure version from the backend some of these approaches are presented and com-
database, the structure selection component reads pared by their capabilities.
all necessary data from the backend database.
Cell data and transformation data are handled Approach of kimball: slowly
by data transformation component. The export changing dimensions
interface is responsible for writing the structure
data the transformed cell data into the external Kimball (1996) was probably the first discovering
data warehouse system. the need for evolving structures in a data warehouse
and introduces three methods for dealing with
so called slowly changing dimensions. The first
method simply suggests overwriting old instances
with their new values, thus tracking a change his-
Figure 3. Architecture of DWT (Adapted from Eder and Wiggisser (2006))
575
tory is not possible. The second method consists Approach of Hurtado,

of creating a new instance for each change. This mendelzon and vaisman
will create a version history, but means additional
effort in data management. One has to introduce The approach of Hurtado, Mendelzon and Vaisman
a surrogate key, because the natural primary keys (Hurtado, Mendelzon & Vaisman, 1999; Vais-
may not be unique any longer. For relating vari- man & Mendelzon, 2001) proposes a temporal
ous versions of an instance to each other, creating multidimensional model and a temporal query
a time stamp for the validity of each version is language. It allows modifications of schema and
proposed. The third method proposes creating a instances. The temporal dimension schema is
new attribute for the instance, so the original and defined as a directed acyclic graph. Each node
the current attribute value can be saved. This can, represents a dimension level. The edges connecting
of course, only handle two versions of an instance. the levels are labeled with a time interval denot-
All three methods are quite straightforward and ing the time when the edge is valid. The same
only allow very basic modifications on the instance approach is used for dimension members, i.e.
level. Impacts of structure changes on the cell the nodes represent dimension members and the
data are not considered. Furthermore, they are edges between them are labeled with their valid
only applicable for slowly changing dimensions time. This model assumes that only the relations
and not for what Kimball calls a rapidly changing between the nodes, and not the nodes themselves,
monster dimension (Kimball and Ross, 2002). For change. The temporal query language TOLAP can
this type of changes, they suggest swapping out be used to execute queries over a set of temporal
frequently changing attributes (e.g. age or income dimensions and fact tables.
of a customer) into so called minidimensions. With
their help the original member remains unchanged Approach of blaschka, sapia
and if an attribute changes, just another member and Höfling: Fiesta
in the minidimension is used.
The Framework for Schema Evolution in Multidi-
Approach of yang and Widom mensional Databases (FIESTA) (Blaschka, Sapia
and Höfling, 1999) supports a schema design
Yang and Widom (1998) present an approach that technique and some evolution operations. The
allows building a temporal data warehouse for tem- authors derive the need for an evolution methodol-
poral and non-temporal data sources. This work ogy from the fact that the data warehouse design
is based on the assumption of a data warehouse process is an iterative process. The proposed
being only a view over the sources. It contains evolution algebra supports modifications on the
temporal query language and a framework to cre- schema level. Instance evolution is not supported
ate and incrementally update temporal views over directly, but the instances are adapted automati-
the history of instances. As such a history may cally according to the changes on schema level,
not always be available in the sources, auxiliary either physically in the database or logically with
data is stored in the data warehouse to enable views and filters.
self-maintaining views.
Approach of Quix
Quix (1999) provides a framework for quality-

oriented evolution of data warehouse views. The
author proposes a data warehouse process model
576
to capture dynamic aspects by representing the members and hierarchical relations. All of the
data warehouse evolution as a special process. The elements on the instance level are timestamped,
main focus is on the maintenance of evolving data schema elements are not evolvable. Furthermore,
warehouse views. Metadata is provided to keep it contains a so called confidence factor, describing
track of the history of changes in the conceptual the reliability of data, for instance if it is source
and logical schema. Consistency rules are defined or mapped data, and a mapping relationship that
guarantee consistency when quality factors have describes the mapping between two versions of a
to be re-evaluated. dimension member. Operations are only defined
on the member level. There are operations for
Approach of sarda creating new members, removing and reclassifying
(i. e. moving) members, and create associations
Sarda (1999) presents a formal model for a between versions of a member.
temporal data warehouse that supports multiple
hierarchies, a symmetric treatment of dimensions Approach of kaas, Pedersen
and measures and many-to-many relationships and rasmussen
among dimension levels. The authors provide a
mapping from their formal temporal ware house Kaas, Pedersen and Rasmussen (2004) present
model to a relational model. History is recorded their approach for supporting schema evolution
for dimension members as well as for relations for star and snowflake schema. In contrast to
between them. other approaches, this methodology takes into
account the special needs of an evolving star or
Approach of ravat and teste snowflake schema using a relational database as
data storage. They provide a rich set of changing
Ravat and Teste (2000) introduce an object- operations, including operations for inserting and
oriented approach to data warehouse modeling. deleting dimensions, categories and dimension
They define a Warehouse Class Extension (the members. For each of the operations, the impact
instances extracted from the sources) as current on existing queries and the complexity of applying
state, a set of historical states, and a set of archived the operation is evaluated. These evaluations are
states for the instances. Whereas historical states compared for the case of an underlying star and
are available on detailed level, archived states can snowflake schema. From these comparisons the
only be queried on an aggregated level. The main authors conclude that a star schema is more robust
focus is laid on modeling of data warehouses. in case of structure evolution. Data transforma-
tion is not captured in this paper, but considered
Approach of body, miquel, as future work.
bedard and tchounikine
Approach of Golfarelli,
Body, Miquel, Bedard and Tchounikine (2003) lechtenbörger, rizzi and vossen
present their approach for handling evolution in
multidimensional structures. As basis for their Golfarelli, Lechtenbörger, Rizzi and Vossen (2004;
approach they define a temporal multidimensional 2006) present their approach for schema version-
model.Based upon this model a set of evolution ing in data warehouses. Based on a graph model
operators is defined. The multidimensional model (called schema graphs) of the data warehouse
comprises the typical elements of a data warehouse schema they present their algebra for schema
structure, i. e. dimensions, levels, dimension modifications. This approach supports version-
577
ing, therefore, past versions are not lost. Based constellation, a report showing the historical truth
on those schema versions the authors describe a and a report showing comparable results. This ap-
mechanism to execute cross-version queries, with proach is limited to basic operations on dimension
the help of so called augmented schemas. For cre- members and does not allow to transform cell data
ating such an augmented schema, an old schema between structure versions.
version is enriched with structure elements from
a subsequent version, so that the data belonging Approach of kAlIdo
to the old schema version can be queried as if it
followed the new version. The KALIDO Dynamic Information Warehouse
(2004) supports some aspects of data warehouse
Approach of malinowski and Zimanyi maintenance. Dealing with changes is realized by
the so called Generic Data Modeling. The data
Malinowski and Zimanyi (2006) present an warehouse model consists of three categories of
approach for representing time-varying data in data: the transaction data, describing the activi-
dimensions. Based on their MultiDimER (Ma- ties of the business and the facts associated with
linowsky & Zimanyi, 2004) they present a set them, the business context data, which is the
of temporal extensions that allow conceptual analog to the instances, and the metadata, which
representation of time-varying levels, attributes among others comprises parts of the schema.
and hierarchies. They describe the versioning With evolving the business context data, instance
of levels and hierarchies. When describing time evolution is supported.
varying levels, they actually refer to modeling
changes in the members contained in this level. comparison of Approaches
With respect to hierarchies, the authors distinguish
between temporal levels and non-temporal rela- After having presented the various approaches,
tions between them, temporal levels and temporal Table 2 shows a classification of them with respect
relations between them, and non-temporal levels to the following features:
and temporal relations between them. Based on
these scenarios the authors extend their Multi- 1. Level of Maintenance: Does the approach
DimER metamodel to include the capability of support schema maintenance, instance
expressing these changes. Additionally, means for maintenance or both of them?
transforming MultiDimER into a classical Entity 2. Type of Historization: Does the approach
Relationship model are proposed. As the approach support versioning or evolution?
is more concerned about conceptual modeling of
changes, considerations about dealing with cell Of course, the desired feature combination
data are not included. would be an approach supporting versioning on
schema and instance level, because this offers
Approach of sAP AG the biggest flexibility. On the other hand, it can
be seen that the current commercial approaches
Besides these scientific approaches, there are both support maintenance only on instance level.
also two commercial products including sup- But nevertheless, selection considering these
port for changing structures in data warehouses. two features only may not be sufficient. Table
SAP AG (2000) presents an approach capable of 3 compares some the approaches (two of them
producing four different types of reports: a report quite established and two rather new ones) with
using today’s constellation, a report using an old respect to a set of additional features. The selec-
578
Table 2. Classification of approaches
Versioning Evolution
Schema and Instance Eder et al., Malinowski et al. Hurtado et al., Kaas et al.
Schema only Golfarelli et al. Blaschka et al., Quix
Instance only SAP AG, Ravat et al., Sarda, Body et al. Kimball, Yang et al, Kalido
tion of approaches for the comparison is based on 5. Formal Temporal Model: Is a formal model
their similarity, i.e. the selected approaches follow for the temporalization provided?
similar principles, which allows to define a set of 6. Changes in Time/Fact dimension: Does the
features applicable to all of them. A plus in a cell approach allow changes in the Time and/or
expresses that the respective approach supports Fact dimension?
the feature. A minus means that the respective 7. Modeling Technique: Is the approach based
feature is not supported by that approach. on a particulare technique for multidimen-
sional modeling, possibly presented by the
1. Valid time for hierarchical relations: Does authors?
the approach support valid time for the rela-
tions between dimension members? related Problems
2. Valid time for dimension members: Does
the approach support valid time for dimen- Besides the classical maintenance requirements
sion members? of keeping managing structural changes in a data
3. Data transformation or multiversion warehouse, maintenance methodologies can also
queries: Does the approach support que- be used to facilitate so called what-if-analysis.
ries spanning several structure versions or Bebel, Wrembel and others (2004, 2007) present
transformation of cell data between structure an approach for the management of multiversion
versions? data warehouses. They differentiate between real
4. Complex Operations: Does the approach versions and alternative versions. The former are
support operations other then insert, delete, used to historicize data warehouse modifications
update (for instance split or merge)? resulting from real world changes. Alternative ver-
Table 3. Features of different maintenance approaches
Eder
Hurtado et al. Golfarelli et al. Malinowski et al.
et al.
Valid time for hierarchical relations + + + +
Valid time for dimension members + - - +
Data transformation / multiversion queries + - + -
Complex
+ + - -
operations
Formal Temporal Model + + + +
Changes in Time/Fact dimension + - + -
Modelling
- - + +
Technique
579
sions facilitate to create several versions, each of Future trends

them representing a possible future situation and
then apply what-if-analysis on them. Additionally, Current commercial systems assume that the data
alternative versions may be used to simulate data warehouse structure is defined at design time and
warehouse modification for optimization purposes. does not change afterwards. Therefore the support
Shahzad, Nasir and Pasha (2005) present of structure modifications is rather limited. On the
a similar approach for enabling evolution and other hand, real-world systems are used in evolv-
versioning in data warehouses that supports the ing application domains. Thus, the demand for
simulation of business scenarios with alterna- modifications is present, because the system has
tive versions. Based on the formal model of a so to be consistent with its application domain and
called versioned multidimensional schema, they requirements. Despite the fact that more effort is
introduce a set of operations on the schema and put into integrating maintenance capabilities into
instance level. All the operations are defined by commercial data warehouse systems, current prod-
a changing-algebra defined upon the multidi- ucts are still not well prepared for this challenge.
mensional model. The versioning function sup- Schema and instance maintenance are quite
ports both versioning and evolution. To support elaborated in current research approaches, but
simulation of business scenarios so called real efficient cell data transformation between differ-
versions and alternative versions may be created. ent structure versions is still subject to research.
The problem of data transformation following the The two main problems with data transforma-
change operations is not dealt with. tion are first of all to define semantically correct
Another instance of data warehouse main- transformation functions, which express the
tenance is the so called View Maintenance. The user’s requirements and expectations. The second
classical data warehouse maintenance deals with problem is the huge amount of data, which has
structural changes, and often assumes the data to be dealt with. Related to data transformation
warehouse structure to be rather independent from are also multiversion queries. The problem with
the underlying sources. View maintenance, on such queries is again their semantic definition, i.e.
the other hand, interprets the data warehouse as whether and how to include cell data related to
a materialized view of the sources. Thus, changes elements which are not valid of the whole period
in the sources directly affect the data warehouse. of consideration.
For instance, Zhuge, Garcia-Molina, Hammer and
Widom (1995) present an approach to synchronize
changes in the source data to the materialized conclusIon
view. The main problem here is to decide whether
to update or to recalculate the view from scratch. Maintenance is an important aspect in the data
But as this, and also similar, approaches only deal warehouse domain. Typically, data warehouse
with changes of transaction data they are out of systems are used in a changing environment thus
scope for this chapter. Bellahsene (2002) presents the need for evolving systems is inevitable. This
an approach for structural view maintenance, i.e. chapter provides some simple, but yet illustrating
updating the view definition with changes from the examples that motivate the need for data ware-
underlying data sources. The presented operations house maintenance.
only cover addition and deletion of attributes of Founded on basic data warehouse concepts, the
the source schema. Also here the main question levels of data warehouse maintenance are intro-
is, whether an update of the view is possible and duced. Data warehouse structure may change on
cheaper than recalculation. the schema (i.e. dimensions and categories) and
580
instance (i.e. dimension members) level. Not only Dynamic Information Warehouse, K. A. L. I. D.
the structure elements themselves, but also the re- O. (2004). A technical overview. Retrieved May
lations between them may change. When keeping 8, 2007 from http://www.kalido.com
track of structure changes, one has also to decide
Eder, J., & Koncilia, C. (2001). Changes of di-
whether to use an evolutionary or versioning ap-
mension data in temporal data warehouses. In
proach. Keeping available prior versions may for
Y. Kambayashi, M. Mohania, & W. Wöß (Eds.),
instance be necessary for legal reasons. Besides
Proceedings of the 3rd International Conference
managing these structural changes, another im-
on Data Warehousing and Knowledge Discovery
portant aspect of data warehouse maintenance is
(pp. 284-293). Heidelberg: Springer.
dealing with the impact of structure changes on
the cell data. Modifications of structure data may Eder, J., Koncilia, C., & Morzy, T. (2002). The
corrupt the structural and/or semantic consistency COMET metamodel for temporal data ware-
of the associated cell data. houses. In A. Pidduck, et al. (Eds.), Proceedings
There are several approaches dealing with of the 14th International Conference on Advanced
the data warehouse maintenance problem. One Information Systems Engineering (pp. 83-99).
of them, namely the COMET metamodel for Heidelberg: Springer
Temporal data warehouses, is presented in some
Eder, J., Koncilia, C., & Wiggisser, K. (2006).
details. Several other approaches addressing the
Maintaining temporal warehouse models. In L.
data warehouse maintenance problem are shortly
Xu & A. M. Tjoa (Eds.), Proceedings of the IFIP
introduced. After their introduction, some of the
International Conference on Research and Practi-
presented approaches are compared with respect
cal Issues of Enterprise Information Systems (pp.
to several features. Approaches dealing with
21-30). Heidelberg: Springer
problems related to data warehouse maintenance
are presented to mark the boundaries of this re- Eder, J., & Wiggisser, K. (2008). Modeling trans-
search area. formations between versions of a temporal data
warehouse. In Proceedings of ER 2008 Workshops.
Heidelberg: Springer
reFerences
Eder, J., & Wiggisser, K. (2009). Data warehouse
Bellahsene, Z. (2002). Schema evolution in data maintenance, evolution and versioning. In L.
warehouses. Knowledge and Information Systems, Liu & T. Özsu (Ed.), Encyclopedia of Database
4(3), 283–304. doi:10.1007/s101150200008 Systems. Heidelberg: Springer.
Blaschka, M., Sapia, C., & Höfling, G. (1999). On Froeschl, K., Yamada, T., & Kudrna, R. (2002).
schema evolution in multidimensional databases. Industrial statistics revisited: From footnotes to
In M. Mohania & A. M. Tjoa (Eds.), Proceedings meta-information management. Österreichische
of the 1st International Conference on Data Ware- Zeitschrift für Statistik, 31(1), 9-34.
housing and Knowledge Discovery (pp. 153-164). Golfarelli, M., Lechtenbörger, J., Rizzi, S., &
Heidelberg: Springer. Vossen, G. (2004). Schema versioning in data
Body, M., Miquel, M., Bedard, Y., & Tchounikine, warehouses. In S. Wang et al. (Eds.), Conceptual
A. (2003) Handling evolutions in multidimen- Modeling for Advanced Application Domains,
sional structures. In U. Dayal, K. Ramamritham, & ER 2004 Workshops (pp. 415-428). Heidelberg:
T.M. Vijayaraman (Eds.), Proceedings of the 19th Springer.
International Conference on Data Engineering
(pp. 581-591). New York: IEEE Computer Society.
581
Golfarelli, M., Lechtenbörger, J., Rizzi, S., & Ravat, F., & Teste, O. (2000). A temporal object-
Vossen, G. (2006). Schema versioning in data oriented data warehouse model. In M. Ibrahim,
warehouses: Enabling cross-version querying J. Küng, & N. Revell (Eds.), Proceedings of the
via schema augmentation. Data & Knowledge International conference on Database and Expert
Engineering, 59(2), 435–459. doi:10.1016/j. Systems Applications (pp. 583-592). Heidelberg:
datak.2005.09.004 Springer
Hurtado, C., Mendelzon, A., & Vaisman, A. Rizzi, S., & Golfarelli, M. (2006). What time is it in
(1999). Updating OLAP Dimensions. In Proceed- the data warehouse? In A. Tjoa & J. Trujillo (Eds.),
ings of the ACM Second International Workshop Proceedings of the 8th International Conference
on Data Warehousing and OLAP (pp. 60-66), on Data Warehousing and Knwoledge Discovery,
New York: ACM Press. DaWaK 2006 (pp. 134-144). Heidelberg: Springer
Inmon, W. (2005) Building the data warehouse SAP. (2000). Multi-dimensional Modeling with
(4th ed.) New York: John Wiley & Sons. BW: ASAP for BW Accelerator. Retrieved May
8, 2007 from http://sap.com
Kaas, C., Pedersen, T. B., & Rasmussen, B. (2004).
Schema evolution for stars and snowflakes. In Sarda, N. (1999). Temporal issues in data ware-
Proceedings of the 6th International Conference house systems. In Proceedings of the Interna-
on Enterprise Information Systems (pp. 425-433). tional Symposium on Database Applications in
Non-traditional Environments (pp. 27-34). New
Kimball, R. (1996). Slowly changing dimensions.
York: IEEE Computer Society
DBMS Magazine, 9(4), 14.
Shahzad, M. K., Nasir, J. A., & Pasha, M. A. (2005).
Kimball, R., & Ross, M. (2002). The data ware-
CEV-DW: Creation and evolution of versions in
house toolkit. (2nd ed.) New York: John Wiley
data warehouse. Asian Journal of Information
& Sons.
Technology, 4(10), 910–917.
Malinowski, E., & Zimányi, E. (2004). OLAP
Vaisman, A., & Mendelzon, A. (2001). A tempo-
hierarchies: A conceptual perspective. In J. Persson
ral query language for OLAP: Implementation
& J. Stirna (Eds.), Proceedings of the 16th Inter-
and case study. In G. Ghelli & G. Grahne (Eds.),
national Conference on Advanced Information
Proceedings of the 8th International Workshop
Systems Engineering (pp. 477-491). Heidelberg:
on Database Programing Languages (pp. 78-96).
Springer.
Heidelberg: Springer
Malinowski, E., & Zimányi, E. (2006). A con-
Wrembel, R., & Bebel, B. (2007). Metadata
ceptual solution for representing time in data
management in a multiversion data warehouse.
warehouse dimensions. In M. Stumptner, S.
Journal on Data Semantics, 8, 118–157.
Hartmann, & Y. Kiyoki (Eds.), Proceedings of
the 3rd Asia-Pacific Conference on Conceptual Yang, J., & Widom, J. (1998). Maintaining tempo-
Modelling (pp 45-54). Newcastle: CRPIT ral views over non-temporal information sources
for data warehousing. In H-J. Schenk, F. Saltor,
Quix, C. (1999). Repository support for data
I. Ramos, & G. Alonso (Eds.), Proceedings of
warehouse evolution. In S. Gatziu et al. (Eds.),
the 1998 International Conference on Extending
Proceedings of the 1st International Workshop on
Database Technology (pp. 389-403). Heidelberg:
Design and Management of Data Warehouses (p.
Springer
4). CEUS-WS.org
582
Zhuge, Y., Garcia-Molina, H., Hamer, J., & Wi-

dom, J. (1995). View maintenance in a warehous-
ing environment. In M. Carey & D. Schreider
(Eds.), Proceedings of SIGMOD (pp. 316-327).
New York: ACM Press
plex Construction, edited by Ladjel Bellatreche, pp. 171-188, copyright 2010 by Information Science Reference (an imprint
of IGI Global).
583
584
Chapter 3.2
Information Supply Chains:
Restructuring Relationships,
Chains, and Networks
Hina Arora
Arizona State University, USA
T. S. Raghu
Ajay Vinze
AbstrAct events such as Hurricane Katrina have highlighted

the need for and value of ISCs by exposing the
Information supply chains (ISCs) take an in- vulnerability of supply chains to demand surges
formation-centric view of supply chains, where and supply disruptions. Mitigating supply chain
information is not only used to support supply vulnerabilities requires a mechanism that allows
chain operations, but also to create value for for continuously sensing the environment, detect-
customers and enable business partners to col- ing existing or anticipated vulnerabilities, and
lectively sense and respond to opportunities in a responding to these vulnerabilities in real time
networked ecosystem. Creating value in the ISC through information sharing and collaboration.
involves gathering, organizing, selecting, synthe- This chapter describes how the autonomic com-
sizing, and distributing information. In so doing, puting paradigm can be used to build resilient
ISCs should provide secure, confidential, reliable, information supply chains by restructuring the
and real time access to heterogeneous informa- associated relationships, chains, and networks.
tion, while ensuring that the right information is
delivered to the intended recipients at the right
time. In other words, security, information qual- IntroductIon
ity, and information lead-time delays are critical
performance determinants in ISCs. Recent disaster Supply chain vulnerability can be defined as
(Rice and Caniato, 2003) “an exposure to serious
DOI: 10.4018/978-1-60566-723-2.ch012
disturbance, arising from risks within the supply
Information Supply Chains
chain as well as risks external to the supply chain”. demand surge. In contrast, the Federal Govern-
Efficient supply chains should be responsive to ment relied on redundant capacity maintained in
demand surges and supply disruptions resulting the Strategic Petroleum Reserve to mitigate the
from internal and external vulnerabilities. Demand supply disruptions caused by Hurricane Katrina
surges and supply disruptions are characterized by in the Gulf of Mexico (Gross, 2005).
the probability of their occurrence, the magnitude Responding to demand surges and supply dis-
of the impact, and the ability to cope with them. ruptions therefore requires efficient redistribution
Firms can respond to these vulnerabilities by ei- and reallocation of resources based on real-time
ther, reallocating and redirecting existing capacity decision-making through information sharing and
(this involves prioritizing among customers), or, collaboration. In other words, there is a need for
maintaining redundant capacity (this involves an information-centric view of supply chains that
an inherent trade off between “just-in-time” lean integrates information gathering, collaboration
supply chains and maintaining inventory and and decision making in order to support supply
capacity slack “just-in-case” (Martha and Sub- chain operations and create value for custom-
bakrishna, 2002)). ers. This gives rise to the notion of Information
Mitigating supply chain vulnerabilities re- Supply Chains (ISCs). There are three drivers
quires complex coordination mechanisms among for ISCs, each with unique perspectives to help
a network of entities, systems and organizations. with resiliency in the supply chain. The first is
Various points of contact and information hand- an organizational perspective that examines the
offs leave open the possibility of errors. Effective differences and similarities between traditional
supply chain managers react to such disruptive supply chains and information supply chains in
events by making decisions in real-time (Raghu order to gain a better understanding of informa-
and Vinze, 2004). This requires high levels of tion supply chain requirements. The second is
information sharing, real-time responsiveness, a process orientation that considers the unique
and collaboration. informational challenges that arise in dynamic
Hurricane Katrina provided evidence of the decision-making environments such as those
importance of efficient decision-making, collabo- caused by supply chain disruptions. The third is
ration and information sharing in responding to a technological perspective that investigates the
demand surges and supply disruptions. Wal-Mart use of autonomic computing principles in building
was able to move supplies to areas hit by hurri- a resilient information-based supply chain. The
cane Katrina because it had an emergency opera- next three sections take a closer look at each of
tions center that was staffed around the clock by these orientations in turn. Section 5 presents an
decision-makers who had access to all of the com- illustrative example of how autonomic comput-
pany’s systems (Worthen, 2005). District managers ing principles can be used to build an ISC in the
could call in to request supplies, and the decision context of a healthcare supply chain that has been
makers, with the help of the logistics department disrupted by an Influenza pandemic.
would decide on how to relocate supplies to the
affected area. They also relied on historical point-
of-purchase-data from other hurricanes to forecast InFormAtIon suPPly cHAIns
consumer demand before and after the hurricane,
and used this data to stockpile emergency supplies A supply chain is a network of material, infor-
in distribution centers around the affected areas. mation and cash flows between suppliers (who
Wal-Mart therefore mostly relied on reallocat- provide raw material), manufacturers (who convert
ing and redirecting existing capacity to meet the raw material to final products), distributors (who
585
transport and deliver products to customers), and We therefore define information supply
customers. While material flows from suppliers chains (ISC) as a collection of information and
to the customers, information and cash flows in communication technologies to provide a secure
the opposite direction. integrated decisional environment that enables
Traditional supply chains take an inventory- business partners to collectively sense and respond
based approach, where information is treated as to opportunities and challenges in a networked
a supporting element in the value-adding process, eco-system.
not a source of value itself. The primary focus ISCs differ from traditional supply chains in
is on reducing material-flow lead-time delays. many respects (based on the concept of virtual
However, volatile, unpredictable demand, and value chains, Rayport and Sviolka, 1995). First,
complex coordination mechanisms in today’s ISC shifts away from the traditional supply chain
marketplace is driving the need for a high level model of a linear sequence of activities with de-
of information sharing, real-time responsive- fined points of inputs and outputs, to a matrix of
ness and collaboration. It is therefore becoming potential inputs and outputs that can be accessed
increasingly important to take an information- and distributed through a wide variety of chan-
based approach that concentrates on reducing nels. Second, ISCs use digital assets that are not
information-flow lead-time delays in addition to used up in their consumption. Hence, companies
material-flow lead-time delays. that create value with digital assets may be able
ISCs take an information-centric view of sup- to re-harvest them through a potentially infinite
ply chains, where information is not only used number of transactions. Third, ISCs enable
to support supply chain operations, but also to companies to achieve low unit costs of products
create value for customers. Creating value in the and services thus redefining economies of scale.
ISC involves gathering, organizing, selecting, Fourth, ISCs can help provide value across many
synthesizing and distributing information. In so different and disparate markets. Fifth, transaction
doing, ISCs should provide secure, confidential, costs are much lower in ISCs. Finally, ISCs enable
reliable and real-time access to heterogeneous companies to be market sensitive (demand-driven
information, while ensuring that the right informa- rather than forecast-driven). The differences are
tion is delivered to the intended recipients at the summarized in Table 1.
right time. In other words, security, information These differences not withstanding, ISCs
quality and information lead-time delays are criti- have a lot in common with traditional supply
cal performance determinants in ISCs. chains. They deal with demand and supply (of
Table 1. Differences between traditional supply chains and information supply chains
Traditional Supply Chain Information Supply Chain

Linear sequence of activities with defined points of inputs and Network of information sources and sinks.
outputs.
Physical assets used up in consumption. Digital assets not used up in consumption.
Economies of scale reachable only by increasing the number of Small companies can achieve low unit costs of production and
goods produced. services.
Companies usually constrained to one market. Companies can provide value in disparate markets.
Characterized by high transaction costs. Low transaction costs.
Forecast driven. Demand driven.
586
information), raw goods (data), finished goods by monitoring for attacks), but also produce value
(synthesized information), capacity limitations for the customers (say, by alerting the user of an
(hardware and human cognitive limitations), update or system shutdown for maintenance). As
fill-rates, lead-time delays, costs (of searching should be evident from this simple example, in
and synthesizing information), quality of goods, order to create value, the ISC should be capable
and supply chain disruptions. of gathering information from multiple sources,
Given the above characteristics of ISCs, it can compiling complex data, and making appropriate
be seen that they would be most relevant in sec- recommendations. In the case of IT networks,
tors such as Services, Finance, IT and Healthcare. this was traditionally done by the system admin-
Table 2 compares ISC characteristics across these istrator. However, due to bounded rationality
three domains from the organization, process and and cognitive limitations, human administrators
technology perspectives. Consider an IT network can deal with only simple decision making tasks
for instance. The network could be made up of and small amounts of information. In fact, this
several workstations, servers, routers and print- is true of handling disruptions in other dynamic
ers. Workstation users in this scenario are not decision-making environments such as finance
only users of information (such as checking the and healthcare as well. Concepts across these three
print queue, or looking for an idle machine to run varied domains share commonalities in decision-
a complex job), but also creators of information making processes as each of them demonstrate the
(such as a submitted print job). Such an ISC will need for dynamic decision-making.
not only use information to remain functional (say,
Table 2. ISC characteristics in three domains
Characteristic Domain
IT Finance Healthcare
Stakeholders System administrators, users Stock markets, shareholders Healthcare providers, patients
Information Character- Defined standards, high volume Relatively well defined stan- Heterogeneous standards or lack of
istics of information, real-time or batch dards, high volume of infor- them, high volume of information,
processing mation, real-time information real-time information processing
processing
Business Objectives System availability, reliability, applica- Timely transaction execution, System availability, reliability,
tion performance, security real-time information feed and security, confidentiality, real-time
delivery, security, reduce trading information feed and delivery
cost, maximize returns
Constraints Cognitive limitations of system admin- Information processing, discern- Cognitive limitations of providers,
istrators, limited resources ing macro economic trends and limited resources
analyst information
Resources CPU availability, memory availability, Stock prices, options, capital, Beds, medical staff, medicines, vacci-
intrusion monitoring data credits nations, epidemic surveillance data
Information node char- Peer-to-peer or client-server con- Peer-to-peer Peer-to-peer or client-server con-
acteristic figuration figuration
Domain specific issues Gathering and synthesizing requisite Trading costs Gathering and synthesizing requisite
information, false positives and false information, false positives and false
negatives negatives
Business Value of Auto- Freeing administrators of mundane Synthesizing and responding Enabling collaboration among pro-
nomic Approach tasks, facilitating intrusion detection, to market trends on real-time viders, facilitating epidemic con-
improved services for users. basis trol, improved quality of care for
patients.
587
extendInG Isc tHrouGH understanding of dynamic decision-making tasks

dynAmIc decIsIon mAkInG and decision support for this class of problems
remains limited (Gonzalez, 2005).
As mentioned earlier, responding to demand Dynamic decision making is characterized by
surges and supply disruptions requires efficient the following four properties (Edwards, 1962;
redistribution and reallocation of resources based Brehmer, 1992): (a) a series of decisions is re-
on real-time decision-making through information quired to reach the goal, (b) the decisions are
sharing and collaboration. This requires solving interdependent (later decisions are constrained
two types of decision-making problems: the Situ- by earlier decisions), (c) the state of the decision
ational Analysis Problem (how to monitor, repre- problem changes, both autonomously, and as a
sent and analyze the various positions, proposals consequence of the decision maker’s actions, and
for actions, hypotheses and evidences) and the (d) the decisions have to be made in a real-time
Conflict Resolution Problem (how to enable rapid environment. Dynamic decisions typically arise
collaborative decision-making through effective in military contexts, and extreme events such as
conflict resolution). IT can be a great enabler in natural or technological disasters (Mendonca,
enhancing our ability to cope with supply chain 2007). For instance, natural disasters such as Ka-
disruptions. IT can increase situational awareness trina create a demand surge and associated scarcity
among decision-makers and enhance coordination for food, water, medication and shelter. A series
through efficient communication and data shar- of decisions regarding resource requirements,
ing (voice, data and chat capabilities). However allocation, transportation and relative urgency
decision-making support for conflict resolution need to be made. The decisions are interdependent
in such environments is still very limited and not in that, the allocation of resources and attention
well understood. Decision support can greatly to one population affects the time and resources
enhance resource mobilization, allocation and spent on the remaining population. This in turn
coordination efforts, and provide a platform to test determines the extent of human and economic
various “what-if” scenarios. This can be invaluable loss associated with the disaster. The state of the
in dealing with supply chain disruptions. decision problem can also change autonomously
Decision-making can be described as a four- due to the arrival of a second hurricane or other
step process (Sprague, 1980; Simon, 1960): (a) natural factors. In other words, the state of the
intelligence comprised of searching the environ- decision space at any given moment is dependent
ment for problems, (b) design involving analysis both on the characteristics of the natural disaster,
and development of possible courses of action, (c) and the decisions made by the decision makers.
choice consisting of selecting a particular course And finally, decisions have to be made in real
of action, and (d) implementation of the action. time in order to be most effective.
Decision support systems (DSS) provide support These issues resonate with the finance, health-
for complex decision-making and problem solving care and IT domains too. Consider for instance
tasks (Sprague, 1980; Shim et al, 2002). DSS also the finance domain, which is characterized by
include support for communication, coordination high volumes of financial data. A series of deci-
and collaboration. A good DSS should improve sions regarding stocks, bonds and options need
both the efficiency (ratio of result over effort) and to be made in order to maximize profits. The
the effectiveness (goal attainment) of the decision- decisions are interdependent in that, the alloca-
making process (Shim, 2002; Bots and Lootsma, tion of funds from a limited budget to one type
2000). While DSS technology has evolved signifi- of stock or option affects the amount that can be
cantly in the last three decades (Shim et al, 2002), invested in the others. The state of the decision
588
problem can also change autonomously due to Three different decision support mechanisms
external macro-economic trends. In other words, are therefore prevalent in the literature for dynamic
the value of the portfolio is dependent both on the decision making (Gonzalez, 2005; Huguenard and
decisions of the investor, and external economic Ballou, 2006): (a) outcome feedback, in which
trends not in the investor’s control. And finally, decision makers are provided with feedback on
decisions have to be made in real time in order the performance results of their decisions, (b)
to be most effective. cognitive feedback, in which decision makers are
The limitations of bounded rationality are even given instructions on how to perform the decision
more pronounced in dynamic decision-making task, and (c) feedforward, where decision makers
tasks. Decision makers are often unable to compre- are provided with the models and tools to perform
hend the side effects of their actions thereby causing “what-if” analysis of potential decisions. It has
performance to degrade. As Radner (2000) says, been shown that outcome feedback alone is an
“…in any even semi-realistic decision problem, ineffective form of decision support (Lerch and
the DM does not know all of the relevant logical Harter, 2001; Gonzalez, 2005). However, cogni-
implications of what he knows. This phenomenon tive feedback and feedforward in combination with
is sometimes called the failure of logical omni- outcome feedback resulted in better performance
science”. Feedback delays and poor feedback qual- (Sengupta and Abdel-Hamid, 1993).
ity further exacerbate performance. The real-time Outcome feedback requires data gathering and
nature of decisions also introduces an element of analysis capabilities to provide useful information
stress into dynamic decision-making, and in order to the decision makers. Feedforward requires
to cope with this stress, the decision maker reverts simulation and modeling capabilities to provide
to simpler, more task oriented modes of operation, optimal solution alternatives. Cognitive feedback
thereby negatively influencing decision perfor- requires the ability to learn from experience and
mance (Brehmer, 1992). Situational awareness in come up with rules of thumb or action policies.
a dynamic decision making environment requires The effective and efficient functioning of an ISC
two overlapping cognitive activities (Lerch and therefore requires the capability of monitoring
Harter, 2001) that compete for the decision maker’s large amounts of data (knowledge synthesis),
attentional resources: (a) monitoring or tracking analyzing it (situational analysis), planning a
of key system variables for information regarding course of action and executing or recommend-
present and expected conditions, and (b) control or ing it (conflict resolution). The ISC should be
the generation, evaluation and selection of alterna- self-configuring, that is, capable of altering its
tive actions that can change the system. Control problem-determination and conflict resolution
can be achieved through feedback (selecting an solutions in the face of changing business policies.
action based on current system information), or And, it should be self-optimizing, that is, capable
feedforward (selecting actions based on a predicted of providing the optimal solution in the face of
future state of the system). While feedforward changing objectives and constraints.
can improve decision quality, decision makers
are more likely to use available feedback control
than feedforward, since the former requires less AutonomIc comPutInG
cognitive effort. Monitoring and feedback can be enAblement For Iscs
improved through improved collection, processing
and delivery of information. Feedforward can be It should be clear from the previous section, that
improved through modeling and simulation aids building ISCs requires the technology enablement
that project future states of the system. of supply chains. Essentially, there is a need for
589
an information-based view of supply chains with and/or synthesize the quantities stated in the condi-
the capability of monitoring large amounts of data, tion, and execute the stated actions whenever the
analyzing it, planning a course of action based condition is satisfied. Goal policies describe the
on stored knowledge and high level policies, and conditions to be attained without specifying how
executing or recommending a course of action. As to attain them. Goal policies are more powerful
we have seen, these needs play out across domains than action policies because they can be speci-
such as emergency response, IT and finance. The fied without requiring detailed knowledge of that
Autonomic Computing paradigm addresses all of element’s inner workings. Autonomic elements
these requirements, and can be used to implement employing goal policies must possess sufficient
ISCs in all of these domains. modeling and planning capabilities to translate
Autonomic computing has been suggested as goals into actions. Utility policies specify the
a new paradigm to deal with the ever-increasing relative desirability of alternative states either by
complexity in today’s systems (Kephart and Chess, assigning a numerical value or a partial or total
2003). Autonomic systems are composed of self- ordering of the possible states. Utility functions
managed elements. Self-management requires that are even more powerful than goal policies because
the system be self-configuring (capable of goal- they automatically determine the most valuable
driven self-assembly with the help of a central goal in any given situation. Autonomic elements
registry), self-optimizing (especially with respect employing utility policies must possess sufficient
to resource allocation), self-healing (the system modeling and planning capabilities to translate
as a whole should be capable of dealing with the utilities into actions.
failure of any constituent part) and self-protecting The AM also assumes the existence of a com-
(against undesirable system behavior due to bugs mon knowledge base that it continuously uses and
or unanticipated conditions, and against system modifies according to its experiences and policies
penetration by attackers). Each autonomic element (Kephart and Walsh, 2004). For instance, when
consists of an autonomic manager and a set of considering utility policies, the decision problem
managed components. The autonomic manager is essentially one of choosing values of decision
continuously monitors the managed components, variables so as to maximize the utility function.
analyzes the data they generate, plans actions if Since the relation between the decision variables
required, and executes them (the MAPE model) and process outcomes is oftentimes dynamic and
in order to achieve the self-management aspects non-stationary, the agent continuously learns about
of the system. the process outcomes and updates its knowledge
The AM relies on high-level system policies to repository.
guide its goals (White et al, 2004). A policy is a rep- Autonomic elements can therefore be used
resentation, in a standard external form, of desired to gather data, synthesize information, and dis-
behaviors or constraints on behavior. High-level tribute it. The elements rely on their knowledge
system policies refer to high-level objectives set base and high-level policies to decide on what to
by the administrators for the autonomic systems, gather, how to synthesize, and whom to distribute
while leaving the task of how they are achieved to. Every element can therefore be a consumer
to the autonomic managers. At least three forms and a provider of information in the ISC. While
of policy have been identified: action policies, Autonomic Computing has been used in the past
goal policies and utility policies (Kephart and to manage resources in the IT context, it can be
Walsh, 2004). Action policies are typically of the extended to other domains to manage physical
form IF (Condition) THEN (Action). Autonomic resources as illustrated in the next section in the
elements employing these policies must measure healthcare context.
590
An Isc cAse study In ing number of human casualties. This can result in
PublIc HeAltH a demand surge and associated scarcity of health-
care resources such as beds, staffing, supplies and
This section presents an illustrative example of equipment due to demand surges. Surge capacity
how autonomic computing principles can be used is a health care system’s ability to expand quickly
to build an ISC in the context of a healthcare sup- beyond normal services to meet an increased
ply chain that has been disrupted by an Influenza demand. Cooperating through mutual aid at the
pandemic. regional level can help improve surge capacity.
The Healthcare System can be characterized This requires real-time collaboration among vari-
as a collaborative endeavor between physicians, ous entities (such as hospitals and pharmacies) in
patients, healthcare organizations, insurance the Healthcare ISC. However, the complexity of
companies, laboratories and public health that are the resource redistribution and allocation problem
intricately connected in a healthcare information can quickly overwhelm human decision makers.
supply chain (Figure 1). While the IT ISC deals This is because, resource allocation in this context
with IT resources such as memory and CPU usage, results in a dynamic decision making environ-
the healthcare ISC deals with healthcare resources ment. A series of decisions regarding vaccination
such as beds, medicines, vaccinations, and hospital policy, anti-viral allocation and patient priorities
staff. Data in the healthcare ISC typically pertains need to be made. The decisions are interdependent
to the patient in terms of health records, medical in that, the allocation of resources and attention
images etc. The healthcare ISC can create value to one population affects the time and resources
for the patients by ensuring patient data is avail- spent on the remaining populations. This in turn
able to the right entity at the right time, monitoring determines the extent of the pandemic. The state
for epidemics, and optimizing resource allocation of the decision problem can also change autono-
during an epidemic. mously due to virus mutations and other natural
A large-scale public health emergency such as factors. In other words, the state of the pandemic
an epidemic outbreak can result in an overwhelm- at any given moment is dependent both on the
Figure 1. A typical Healthcare ISC. Primary data users are in green, secondary data users are in blue,
and tertiary data users are in orange
Consulting Physician
ID-C-H-R-P-B-I
Clinical Laboratory
ID-R-B-I
State Health Departments

ID-C
Patient Primary Care Physician

ID-C-H-R-P-B-I ID-C-H-R-P-B-I
Medical & Pharmaceutical

Research
C-H-R-P
Retail Pharmacy Health Insurance Company

ID-P-B-I ID-C-P-B-I
591
characteristics of the pandemic, and the decisions the scarcity and improve surge capacity. Therefore,
made by the decision makers. And finally, deci- on encountering a resource shortage, the healthcare
sions have to be made in real time in order to be entities contact an Autonomic Manager (level 3)
most effective. Application of self-management to optimally re-distribute resources between enti-
and self-optimization concepts can greatly relieve ties at the regional level. The Autonomic Manager
the cognitive challenges faced by decision makers uses a “utility policy”, to maximize utilities across
in this context. the various healthcare entities, subject to resource
A prototype system based on these ideas was and cost constraints.
built (Arora, Raghu, Vinze, 2006). Autonomic The Autonomic Toolkit2 is used to implement
principles of self-optimization and self-configu- the Autonomic Manager functionality. Each re-
ration were used to address demand surges in the gional healthcare entity communicates its resource
context of healthcare information supply chains requirements (or excesses) to the Autonomic
that have been disrupted by an epidemic. The Manager in the form of a Common Base Event
system was built using a multi-agent systems (CBE). Effective analysis of event data requires
platform and the Autonomic Computing Toolkit, the data to be reported in a consistent manner.
to illustrate how an autonomic computing ap- Since the data is typically heterogeneous and
proach can facilitate resource allocation decisions distributed, the only way to reliably interpret the
in responding to public health emergencies. The data, correlate it with events from other diverse
multi-agent toolkit, Repast1, was used to simulate sources and respond effectively is to use a com-
a Smallpox outbreak. The Autonomic Toolkit mon event format with a canonical vocabulary.
was used to implement the Autonomic Manager This is accomplished by modeling events using the
functionality. Regional healthcare entities com- CBE, which standardizes the format and content
municate their resource requirements (or excesses) of events and provides the foundation for auto-
to the Autonomic Manager. This information is nomic computing. A CBE describes a situation
analyzed for resource shortages and excesses. A that has occurred. Table 3 shows an example of
utility-based constrained optimization problem, some parts of a Common Base Event that could
constrained by high-level budget and resource be used to report an epidemic outbreak and as-
policies, is used to find the optimal resource sociated resource availability.
redistribution solution. Custom resource models are defined in order
Autonomic computing principles are used to to provide the Autonomic Management Engine
build an ISC in this context as shown in Figure 2.
A smallpox outbreak occurs within a population Figure 2. Autonomic resource allocation proto-
(level 1). Those susceptible to the disease get type system
infected. As the epidemic spreads, casualties start
trickling into the care entities (level 2). Timely
intervention in the form of immunization (to re-
duce the number of susceptibles), quarantine (to
reduce the number of contacts with the infected)
and treatment (to help the infected recover) can
help contain the epidemic. However, the exponen-
tial nature of the spread of epidemics can cause
an overwhelming number of casualties leading
to a scarcity of containment resources. Pooling
resources at the regional level could help alleviate
592
Table 3. Example of a CBE in the healthcare context
Field Name Example Data Description

Version 1.0.1 The version of the CBE format that is being used.
creationTime 2005-09-01T01:02:03.456Z The time that the situation was detected.
Severity 2 The severity of the event from the point of view of the entity that
is reporting it.
globalInstanceId C1F2ACFE… A unique value that identifies the event instance such as a GUID
(Globally Unique Identifier).
sourceComponentId [source identifier] Identifies who was affected by the situation.
Location [source location] Identifies the location of the source.
locationType [source location type] The type of location.
reporterComponentId [reporter identifier] Indicates who reported the event (e.g. unique identifier for a hospital,
doctor, etc.).
Location [reporter location] Identifies the location of the reporter (e.g. location of hospital).
locationType [reporter location type] The type of location (e.g. hospital).
Situation ---- The situation that has occurred (contains information in the next
three rows).
categoryName ReportSituation Category of the type of situation that is being reported.
situationType STATUS Provides additional information associated with each situation
category.
ResoningScope EXTERNAL Defines whether the impact of the situation is internal or external
to the source.
ExtensionName ReportableDisease Extensions are used to provide information that is specific to a situ-
ation. There must be a canonical definition for this data.
ExtendedDataElement[0] name=’DiseaseCategory’ The first entry in this extension.
type=’string’
values=’A’
ExtendedDataElement[1] name=’DiseaseName’ The second entry in this extension.
type=’string’
values=’Smallpox’
ExtendedDataElement[..] … Additional extended data elements could be used.
(AME) the requisite information to manage perishable. Third, resources are often required in
healthcare resources. Each regional healthcare bundles. For instance, transferring medicines from
entity communicates its resource requirements one region to another requires trucks and drivers,
(or excesses) to the Autonomic Manager. The just as administering vaccinations requires nurses.
Autonomic Manager uses this information to These additional constraints result in analytically
come up with an optimal allocation mechanism. intractable problems, making it difficult to come
This can be a complex proposition due to a num- up with perfectly optimal solutions under all pos-
ber of reasons. First, resource requirements are sible objectives and constraints. Model execution
often time-bound. For instance, vaccinations are is therefore done in two phases – training and re-
typically effective only in the susceptible stage. finement. During the training phase, agents make
Second, resource availability may also be time- random decision choices and observe process
bound. For instance, resources such as medicines outcomes. This builds a repository of informa-
and vaccinations are consumable, and may also be tion (knowledge base) over time and forms the
593
basis for the agents’ informed decisions during reFerences

the refinement phase. Then, given the repository,
agents systematically relate the decision choices Arora, H., Raghu, T. S., & Vinze, A. (2007, Janu-
and the process outcomes (or their utility values). ary). Optimizing regional aid during public health
Models of the environment further enhance learn- emergencies: An autonomic resource allocation
ing. This two-phase mechanism allows the system approach. 40th Hawaii International Conference
to learn and explore in a protected environment on System Sciences (HICSS), HI.
and exploit the learning in real situations. Feed- Arora, H., Raghu, T. S., & Vinze, A. (Manuscript
back gained from real situations is used to update submitted for publication). Optimizing antiviral
learning patterns. allocation policy through regional aid during an
Initial feedback from this implementation sug- Influenza pandemic. Journal of Decision Sciences
gests that both process and procedural efficiencies Special Issue on Decision Making in the Health-
can be gained by using an autonomic approach Sector Supply Chain.
to building healthcare information supply chains.
This approach facilitates data monitoring and Arora, H., Raghu, T. S., Vinze, A., & Britten-
decision making capabilities, and ensures that ham, P. (2006). Collaborative self-configuration
the right information reaches the right entity at and learning in autonomic computing systems:
the right time. Preliminary results suggest policy Applications to supply chain. Poster, 3rd IEEE
implications for public health in terms of the International Conference on Autonomic Comput-
organizational, process and technology changes ing (ICAC), Dublin, Ireland.
that will have to be incurred in order to make the Bots, P. W. G., & Lootsma, F. A. (2000). Decision
healthcare supply chain more resilient (Arora, support in the public sector. Journal of Multi-Criteria
Raghu, Vinze, Submitted; Arora, Raghu, Vinze, Decision Analysis, 9(1-3), 1–6. doi:10.1002/1099-
2007). 1360(200001/05)9:1/3<1::AID-
MCDA262>3.0.CO;2-D
conclusIon Brehmer, B. (1992). Dynamic decision making:

Human control of complex systems. Acta Psy-
An ISC is a collection of information and com- chologica, 81(3), 211–241. doi:10.1016/0001-
munication technologies to provide a secure 6918(92)90019-A
integrated decisional environment that enables
Edwards, W. (1962). Dynamic decision theory
business partners to collectively sense and respond
and probabilistic information processing. Human
to opportunities and challenges in a networked
Factors, 4, 59–73.
eco-system. Such an information-centric view
can be instrumental in making supply chains Gonzalez, C. (2005). Decision support for real-
more responsive to demand surges and supply time, dynamic decision-making tasks. Organiza-
disruptions. It can enhance the ability of decision tional Behavior and Human Decision Processes,
makers to cope with supply chain disruptions by 96, 142–154. doi:10.1016/j.obhdp.2004.11.002
increasing situational awareness, enhancing co-
Gross, D. A. (2005, September 23). What FEMA
ordination, and presenting feasible alternatives in
could learn from Wal-Mart-less than you think.
an otherwise complex solution space. Autonomic
Slate. Retrieved from http://www.slate.com/
computing presents a promising environment to
id/2126832
implement ISCs, by providing a framework for
information gathering, analysis, learning, and
decision-making.
594
Huguenard, B. R., & Ballou, D. J. (2006). Dis- Rice, J. B., & Caniato, F. (2003, September/Oc-
patcher: A theory-based design for study of real- tober). Building a secure and resilient supply net-
time dynamic decision-making. Proceedings of work. Supply Chain Management Review, 22–30.
the 2006 Southern Association for Information
Sengupta, K., & Abdel-Hamid, T. K. (1993).
Systems Conference (pp. 117-122).
Alternative conceptions of feedback in dynamic
Kephart, J. O., & Chess, D. M. (2003). The vi- decision environments: An experimental inves-
sion of autonomic computing. IEEE Computer, tigation. Management Science, 39(4), 411–428.
36(1), 41–50. doi:10.1287/mnsc.39.4.411
Kephart, J. O., & Walsh, W. E. (2004). An artificial Shim, J. P., Warkentin, M., Courtney, J. F., Power,
intelligence perspective on autonomic computing D. J., Sharda, R., & Carlsson, C. (2002). Past,
policies. Fifth IEEE International Workshop on present, and future of decision support technol-
Policies for Distributed Systems and Networks ogy. Decision Support Systems, 33(2), 111–126.
(pp. 3-12). doi:10.1016/S0167-9236(01)00139-7
Lerch, F. J., & Harter, D. E. (2001). Cognitive Simon, H. (1960). The new science of management
support for real-time dynamic decision making. decision. New York: Harper and Row.
Sprague, R. H. (1980). A framework for the
doi:10.1287/isre.12.1.63.9717
development of decision support systems. MIS
Martha, J., & Subbakrishna, J. S. (2002, Sep- Quarterly, 4(4), 1–26. doi:10.2307/248957
tember/October). Targeting a just-in-case supply
White, S. R., Hanson, J. E., Whalley, I., Chess,
chain for the inevitable next disaster. Supply Chain
D. M., & Kephart, J. O. (2004). An architectural
Management Review (pp. 18-23).
approach to autonomic computing. IEEE Pro-
Mendonca, D. (2007). Decision support for impro- ceedings of the International Conference On
visation in response to extreme events: Learning Autonomic Computing.
from the response to the 2001 World Trade Center
Worthen, B. (2005). How Wal-Mart beat feds
attack. Decision Support Systems, 43, 952–967.
to New Orleans. CIO Magazine. Retrieved on
doi:10.1016/j.dss.2005.05.025
November 1, 2005, from http://www.cio.com/
Radner, R. (2000). Costly and bounded ratio- archive/110105/tl_katrina.html?CID=13532
nality in individual and team decision-making.
Industrial and Corporate Change, 9(4), 623–658.
doi:10.1093/icc/9.4.623
endnotes
Raghu, T. S., & Vinze, A. (2004, September).
Collaborative self-configuration and learning in
1
http://repast.sourceforge.net/
autonomic computing systems. IBM Proposal.
2
http://www-128.ibm.com/developerworks/
autonomic/overview.html
Rayport, J. F., & Sviokla, J. J. (1995). Exploiting
the virtual value chain. Harvard Business Review.
This work was previously published in Always-On Enterprise Information Systems for Business Continuance: Technologies for
Reliable and Scalable Operations, edited by Nijaz Bajgoric, pp. 205-217, copyright 2010 by Information Science Reference
595
596
Chapter 3.3
Benefits of Information
Technology Implementations
for Supply Chain Management:
An Explorative Study of
Progressive Indian Companies
Prashant R. Nair
Amrita University, Coimbatore, India
AbstrAct discussed. Fundamental changes have occurred

in today’s global economy. These changes alter
The usage of Information Technology (IT) in the relationship that we have with our custom-
organizations across the supply chain has become ers, our suppliers, our business partners, and
a determinant of competitive advantage for many our colleagues. Reflection on the evolving and
corporations. This chapter focuses on the usage of emerging IT trends like software agents, RFID,
IT tools for Supply Chain Management (SCM). web services, virtual supply chains, electronic
It also highlights the contribution of IT in help- commerce, and decision support systems, further
ing restructure the entire distribution set-up to highlights the importance of IT in the context
achieve higher service levels, lower inventory, of increasingly global competition. The rapid
and lower supply chain costs. An overview and adoption of the Internet for communication with
tangible benefits of the existing IT tools, which all stakeholders, seems to reflect the potential of
are widely deployed, is provided with focus on the new-age communication media. It has also
existing configuration considerations, available been observed that several progressive Indian
applications, and deployments in India. The role companies are extensively using emerging tools
of existing communication technologies in making like virtual supply chains, web services, RFID,
IT an enabler of SCM, is highlighted by addressing and electronic commerce to shore up their supply
a range of different point and enterprise solutions chain operations. However, adoption of tools like
in a variety of supply chain settings. Critical IT software agents and decision support systems for
demonstrations and implementations in SCM are supply chain integration by Indian companies, is
limited.
DOI: 10.4018/978-1-61520-625-4.ch021
Benefits of Information Technology Implementations for Supply Chain Management
IntroductIon to suPPly transportation control (e.g., owner-operat-

cHAIn mAnAGement ed, private carrier, common carrier, con-
tract carrier, or third party logistics (3PL)).
Supply Chain Management (SCM) is the manage-
ment of a network of interconnected businesses Supply chain execution is the management
involved in the ultimate provision of product and coordination of the movement of material,
and service packages required by end customers information, and funds across the supply chain.
(Harland, 1996). This term was coined by Keith This movement is bi-directional. The basic ele-
Oliver, a Booz Allen Hamilton executive, in 1982. ments of the supply chain are:
SCM spans all movement and storage of
raw materials, work-in-process inventory, and • Information: Integration of processes
finished goods from point-of-origin to point-of- through the supply chain to share valuable
consumption (supply chain). It also encompasses information, including demand signals,
the planning and management of all activities forecasts, inventory, transportation, and
involved in sourcing, procurement, conversion, potential collaboration
and logistics management activities. Importantly, • Inventory Management: Quantity and lo-
it also includes coordination and collaboration cation of inventory including raw materi-
with channel partners, which can be suppliers, als, work-in-progress (WIP), and finished
intermediaries, third-party service providers, and goods
customers. In essence, supply chain management • Cash Flow: Arranging the payment terms
integrates supply and demand management within and the methodologies for fund flow across
and across companies. More recently, the loosely entities within the supply chain.
coupled, self-organizing network of businesses
that cooperates to provide product and service Recent development in information technolo-
offerings, has been called the extended enterprise gies enables organizations to avail information
(Ross, 2006). easily in their premises. These technologies are
Supply chain management must address the helpful to effectively coordinate various sup-
following problems: ply chain activities. The cost of information is
• Distribution Network Configuration:

Figure 1. Integrated supply chain model
Number, location, and network missions
of suppliers, production facilities, distribu-
tion centers, warehouses, cross-docks, and
customers.
• Distribution Strategy: Including ques-
tions of operating control (central-
ized, decentralized or shared); delivery
scheme (e.g., direct shipment, pool point
shipping, Cross docking, Direct Store
Delivery (DSD), closed loop shipping);
mode of transportation (e.g., motor car-
rier, including truckload, parcel; railroad;
ocean freight; airfreight); replenishment
strategy (e.g., pull, push or hybrid); and
597
Figure. 2. Functional Roles of IT in SCM
decreased due to the increasing rate of technolo- cause its value to supply chain members was not
gies. In the integrated supply chain model (Fig.1), clearly understood. IT infrastructure capabilities
bi-directional arrow reflects the accommodation provide a competitive positioning of business ini-
of reverse material and information feedback tiatives like cycle-time reduction and implemen-
flows. The manager also needs to understand that tation of redesigned cross-functional processes.
information technology is more than just comput- Three factors have strongly impacted this
ers. It also includes computer data recognition change in the importance of information. First,
equipment, communication technologies, factory satisfying or rather pleasing customers has become
automation, hardware, and services. something of a corporate obsession. Serving the
customer in the best, most efficient, and effective
manner has become critical. Second, information
is a crucial factor in the managers’ abilities to re-
InFormAtIon tecHnoloGy
duce inventory and human resource requirement
As An enAbler oF suPPly
to a competitive level. Information flow plays a
cHAIn mAnAGement
crucial role in strategic planning.
Information sharing between partners in the
Prior to the 1980s, information flow between func-
supply chain is also critical and these integration
tional areas within an organization and between
attempts are accompanied by IT initiatives. Such
supply chain member partners, were paper-based.
IT initiatives include:
This paper-based transaction and communication
is slow. During this period, information was often
• use of bar-coding in logistics systems;
overlooked as a critical competitive resource, be-
598
• use of EDI to communicate between • providing information availability and

branches; visibility;
• use of material requirements planning; • enabling a single point of contact for data;
• enterprise Solutions like ERP; and • allowing decisions based on total supply
• Internet and web services for communica- chain information; and
tion between partners. • enabling collaboration with partners.
Early studies on the impact of Electronic IT in SCM enables great opportunities, ranging
Data Interchange (EDI) on Just-in-Time (JIT) from direct operational benefits to the creation of
shipments in the automobile industry showed strategic advantage. It changes industry structures
significant earnings with lowered shipment errors and even the rules of competition. IT is key in
(Srinivasan, 1994). supporting companies creating strategic advan-
In the seven principles of SCM, Anderson
tage by enabling centralized strategic-planning
(1996) pointed out that it is necessary to develop
with day-to-day centralized operations. In fact,
a supply chain-wide technology strategy that sup-
supply chain management tends to become more
ports multiple levels of decision-making and gives
market-oriented because of IT usage.
a clear view of the flow of products, services, and
Cisco reported savings of $500 million by re-
information. For this, an IT system should integrate
capabilities of three essential kinds. structuring its internal operations and integrating
For the short term, the system must be able processes with suppliers and customers with the
to handle day-to-day transactions and electronic help of web-based tools (Berger, 2009). Wal-Mart
commerce across the supply chain and thus help and Proctor & Gamble (P&G) have successfully
align supply and demand by sharing information demonstrated how information sharing can be
on orders and daily scheduling. From a mid-term utilized for mutual advantage. Through sound
perspective, the system must facilitate planning information technologies, Wal-Mart shares point
and decision-making; supporting the demand & of sale information from its many retail outlets
shipment planning and master production schedul- directly with P&G and other major suppliers
ing needed to allocate resources efficiently. To add (Anderson, 1996).
long-term value, the system must enable strategic Celestica, one of the world’s largest electronic
analysis by providing tools such as an integrated manufacturing services companies, has applied
network model, which synthesizes data for use a web-based IT tool to regulate its global supply
in high-level “what-if” scenario planning, to help base (Shore, 2001). IT has helped Celestica to
managers evaluate plants, distribution centers, improve its responsiveness to customers, thus
suppliers, and third-party service alternatives. helping its customer, Dell to maintain its delivery
The functional roles of IT in SCM have been promise to end-users.
outlined as follows (Auramo, 2005):
cHAllenGes In ImPlementInG
obJectIves And beneFIts oF InFormAtIon tecHnoloGy In
InFormAtIon tecHnoloGy In suPPly cHAIn mAnAGement
suPPly cHAIn mAnAGement
Any company that has undertaken the mission of
The objectives of IT in SCM are (Simchi-Levi, implementing an integrated supply chain man-
2003):
agement strategy with the use of IT tools, knows
that one of the greatest challenges that it faces is
599
the significant change in internal culture that is sion, such clarity in global distribution remains
required to make the supply chain redesign suc- largely restricted to major multinationals with deep
cessful. It is not an easy thing to re-condition people pockets and volumes large enough to justify the
to accept change, especially in organizations, hefty initial investment in IT, which can run into
where a certain mindset has prevailed for many millions of dollars.
years. However difficult it may be to accomplish, Towill (1997) states that “to survive, let alone
change can be implemented successfully when win, a company must be part of one or more sup-
directed by a strong and knowledgeable leader, ply chains producing world class performance.”
who understands the tools available for achieving Hence, companies need to work together and
positive change, as well as their role in initiating optimize the complete pipeline by establishing a
and sustaining these changes. seamless supply chain to maximize their market
Integrating new applications with existing and share. Significant and radical improvements in
legacy systems could also pose problems. Incom- individual business performance can be realized
patible systems at buyer and vendor facilities are only with the support of this holistic chain concept.
another management challenge to tackle. Sharing Process manufacturers and IT system vendors are
of data with diverse stakeholders like suppliers working to develop a filter to sift through the bar-
and customers, filtering, and mining data gener- rage of data from process control systems to move
ated and thereby finding “business” value of the important information to higher level IT systems.
data are other issues.
Disconnected enterprise systems create data
redundancy, errors and can lead to costly busi- exIstInG InFormAtIon
ness inefficiencies. Poor coordination between tecHnoloGy tools And
enterprise systems leads to flawed production APPlIcAtIons In suPPly
plans, increased supply chain pressure, and poor cHAIn mAnAGement
customer service. Lack of visibility of orders,
schedules, and shipments can lead to costly ad- electronic data Interchange (edI)
ministrative decision-making processes.
According to Macleod (1994), supply chain Introduced in the 1970s and popularized in the
managers increasingly want to automate all of the 1980s, Electronic Data Interchange (EDI) tech-
supply chain, from forecasting to distribution, and nology has been widely used by firms in supply
to link every element of the chain. More and more chains to facilitate transactions and information
companies want an integrated solution to enable exchanges. EDI is defined as computer-to-com-
them to see the entire supply chain at once. For puter exchange of structured data for automatic
instance, they want to know that if they drill down processing. EDI is used by supply chain partners
to forecast, they can see the demand history, which to exchange essential information necessary for
is a combination of data that has come from sales the effective running of their businesses. These
order processing, inventory management, and the structural links are usually set up between organi-
warehousing system. zations that have a long-term trading relationship.
According to Van Oldenborgh (1994), the abil- For example, some multiple retailers will supply
ity to reduce human intervention along the entire electronic point of sale (EPOS) data directly to
length of the supply chain can help dramatically suppliers, which in turn triggers replenishment of
in cutting logistics costs and boosting customer the item sold. Therefore, the consequence of this
satisfaction. Unfortunately, for many midsize type of strong link is that those suppliers will be
companies in these times of economic reces- able to build a historical sales pattern, which will
600
assist their own demand forecasting activities. EDI all its stores. These scanners updated inventory
also provides timely and accurate information numbers for individual items at point of sale and
about customers’ sales. Moreover, it is utilized enabled headquarters to easily aggregate sales
for sending invoices, bills of lading, confirmation and inventory data at its centralized IT depart-
of dispatch, shipping details, and any information ment. Later in 1987, a satellite communications
between supply chain partners (Rushton, 2000). network installation linked all the stores with
The benefits of EDI are quick process to informa- the headquarters with real-time inventory data
tion, better customer service, reduced paper work, (Johnson, 2002).
increased productivity, improved tracing and
expediting, cost efficiency, and improved billing. enterprise resource
With EDI, supply chain partners can overcome Planning (erP) systems
the distortions and exaggeration in supply and
demand information by improving technologies Enterprise Resource Planning (ERP) systems are
to facilitate real time-sharing of actual demand enterprise-wide information systems used for au-
and supply information. tomating all activities and functions of a business.
These transaction-based information systems are
bar coding and scanner integrated across the whole business. ERP systems
facilitate data capture for the whole business into a
Bar Codes are the representation of a number or single computer package, which provides a single
code in a form suitable for reading by machines source for all key business information activities,
(Rushton, 2000). Bar codes are widely used such as customer orders, inventory, and financials.
throughout the supply chain to identify and track Many companies now view ERP systems from
goods at all stages in the process. Bar codes are a vendors like Baan, SAP, and People Soft as the
series of different width lines that may be presented core of their IT infrastructure. ERP systems have
in a horizontal order, called ladder orientation, or a become enterprise-wide transaction processing
vertical order, called picket fence orientation. For tools, which capture the data and reduce the manual
example, goods received in a warehouse may be activities and tasks, which are normally associated
identified by the warehouse management system with processing financial, inventory, and customer
and added to stock held in the warehouse. When order information. ERP systems achieve a high
put away, the bar code is used to associate the level of integration by utilizing a single data model,
storage location with the bar-coded stock, and on developing a common understanding of what the
dispatch, the stock record is amended. The use of shared data represents and establishing a set of
bar codes can speed up operations significantly. rules for accessing data. In addition to the huge
On the other hand, the problems can occur if bar costs that are involved in procuring an ERP ap-
codes are defaced or the labels fall off in transit. plication, installation of such systems will entail
Bar code scanners are most visible in the checkout widespread change within the organization. It will
counters of supermarkets and hypermarkets. The have implications in terms of Business Process
bar code specifies the name of the product and its Re-engineering (BPR), changes in organizational
manufacturer. Bar codes are also used for track- structure, people, and change management. Many
ing of moving items such as components in PC companies have benefited from using this system
assembly operations and automobiles in assembly while some have experienced severe problems
plants. In 1983, with bar codes printed on most with their application. Generally, they also require
goods, Wal-Mart introduced checkout scanners in a lot of customization and training for each user.
601
Warehouse management systems example of such a system is developed by Target

Corporation and NTE. Initially, Target was mak-
Warehouse management systems are systems that ing transportation requests manually for inbound
control all the traditional activities of warehouse shipments (Thongchattu, 2007). There was limited
operations. Areas covered usually include receipt visibility for shipments and as a result of this, there
of goods, allocation or recording of storage loca- were more number of less-than-truckloads, which
tions, replenishment of picking locations, produc- was not cost-effective. Implementation of the new
tion of picking instructions or lists, order picking, system resulted in Target vendors submitting the
order assembly, and stock rotation. Some systems relevant freight information electronically with
are used in conjunction with radio frequency (RF) increased speed and efficiency. The new system
communication equipment. This equipment can resulted in improved cost controls, better labor
be mounted on forklift trucks. The warehouse planning and reduced administrative overheads.
management system communicates with the RF
system and directs the activities of the warehouse Inventory management systems
staff (Thongchattu, 2007). For example, when
picking, it will provide the tasks for the opera- During the mid to late 1990s, retailers began
tive to carry out. Once the task is complete, the implementing modern inventory management
operative updates the system and is directed to systems, made possible in large part by advances
the next task. This has the advantage of updating in computer technology. The systems work in a
the stock holding in real-time. There are highly circular process, from purchase tracking to inven-
sophisticated systems that control the opera- tory monitoring to re-ordering and back around
tions of fully automated warehouses. This may again. Retailers such as Target, Lowe’s and Best
include automated storage and retrieval systems, Buy stock tens of thousands of items from all over
automated guided vehicles, and other devices the world. Wal-Mart alone stocks items made in
that are relatively common in today’s modern more than 70 countries, according to its corporate
warehouse such as conveyors, carousels, and website. It is estimated that at any given time,
sortation systems. A number of computer models the Arkansas-based retailer manages an average
have now been developed to assist in the planning of $32 billion in inventory. With those kinds of
of warehouse design and configuration. These are numbers, having an effective, efficient inventory
generally very sophisticated 3D simulation models control system, or inventory management system,
that provide a graphic, moving illustration on the is imperative. Wal-Mart’s system helps it maintain
computer screen of the layout of the warehouse. its signature ‘everyday low prices’ by telling store
managers about the products that are selling and
transportation those, which are taking up shelf and warehouse
management systems space. Inventory management systems are the
rule for such enterprises, but smaller businesses
Transportation Management Systems provide and vendors use them, too. The systems ensure
more visibility into shipments and orders. Sched- customers always have enough of what they want
uling issues are also addressed on time. Multiple and balance that goal against a retailer’s financial
transportation options can be explored as a result need to maintain as little stock as possible. Factors
of earlier visibility into the supply chain. Timely such as quicker production cycles, a proliferation
communication and status reports can also be ob- of products, and multi-national production con-
tained. By having control on its supply chain, busi- tracts make them a necessity. Modern inventory
nesses can make efficient routing decisions. An management systems must have the ability to
602
track sales and available inventory, communicate agriculture, information technology, and fast-
with suppliers in near real-time and receive and moving consumer goods (FMCGs). ITC initiated
incorporate other data, such as seasonal demand its e-Choupal a few years back with a focus to
(Crosby, 2009). streamline its dealings with Indian farmers. This
is one of the mega projects in rural India. It has
covered more than 40,000 villages in ten states,
suPPly cHAIn mAnAGement including Uttar Pradesh, Madhya Pradesh, Andhra
In IndIA: successFul Pradesh, Maharashtra, Rajasthan, and Karnataka.
demonstrAtIons oF With the initial launch in June 2000, e-Choupal
InFormAtIon tecHnoloGy has become one of the largest initiatives in rural
usAGe In IndIAn comPAnIes India. Today, the services through this system
serve more than 50,000 villages through over
tanishq 7,000 kiosks in various states. The initiative was
launched with a view to providing instant access
One of the largest Indian business conglomerates to real-time information in regional languages for
in India, TATA through its group company, TITAN crop cultivation, as well as empowering farmers
industries, wanted to open a new line of jewel- with decision-making ability. A web-based ini-
lery, under the brand name, Tanishq in 1996. The tiative of ITC’s international business division,
supply chain of Tanishq consisted of raw material e-Choupal is one of the leading programs that
suppliers, manufacturing divisions, carry and offer the farmers with the necessary informa-
forward agents (CFA), and retail stores or bou- tion and services for the enhancement of farm
tiques. It was necessary to connect and integrate productivity and improvement of farm-gate price
69 boutiques and 32 CFAs of the new jewellery realization, as well as reduction in transactional
chain. A web-based IT solution, “Gold Mine” was costs. e-Chaupal offers farmers access to the latest
used to connect factory with the boutiques and local and global information on weather, scientific
CFA. The system offered visibility on the status farming practices and market prices at the village
of orders placed by boutiques and production itself through the Web portal - all in Hindi.
was synchronized with an internal ERP system. ITC plans to operate such kiosks so that they
Sales information flowing in from the boutique can create an electronic stock exchange for the
gave managers full track of the effectiveness of marketing of agricultural commodities through the
their marketing programmes, category-wise and use of ICT. The agri-business division of ITC has
price band-wise. Bulletin boards and online dis- implemented e-Choupal project in order to tackle
cussion forums further added to the application. the most pressing demands of the agriculture sector
Today Tanishq is the largest branded jeweler in in India. The solution provided under this project
India with a market share of around 25% (Kan- serves farmers through several essential modes,
nabiran, 2005). as well as addresses their basic needs. Through
the e-Choupal portals, farmers can access the
Itc’s e-choupal: landmark latest local and global information on weather,
in rural deployment scientific farming practices, and market prices at
the village level. It also facilitates the supply of
ITC Limited is one of the premier and diversi- high quality farm inputs as well as purchases of
fied conglomerates in India. From a traditional produce at the farmers’ doorstep. This IT-backed
tobacco and cigarette producer, it has grown into business solution includes a computer, stationed
a group of companies with interests in hotels, at a farmer’s house, and is linked to the Internet
603
via VSAT connection. This single point of com- advantage from the ERP implementation was
munication serves an average of more than 500 regarding the management of inventory. Before
farmers within a radius of 5 km. Each e-Chaupal ERP implementation, the company’s practice of
comes with an expenditure of nearly US $ 3000 monthly inventory reviews frequently led to time
with a maintenance cost of somewhere between lag in processing orders (ERP Implementation at
US $ 100 and US $ 200 annually. However, the BPCL, 2009).
host farmer- sanchalak incurs some operating
expense and gets substantial benefits from all mahindra and mahindra
e-Choupal transactions. From an IT perspective,
the solution is one of the landmarks in the rural Mahindra & Mahindra started its IT initiatives
segment having created an IT-based integrated in 1995 with the launch of a dedicated network
supply chain for the company also. The web- across 8 manufacturing locations, 33 area of-
driven project offers scalability and effectiveness fices and 4 branch offices. In June 1996, M&M
of communication mechanism, as well as real-time launched its website mahindraworld.com, which
information and facilitates collaboration between was one of the first corporate websites in India.
various parties for the full spectrum of farmer In 2000, M&M extended the Intranet to three
needs (Abbas, 2009). websites dedicated to suppliers and dealers of
the automotive and farm equipment sectors. In
bPcl the same year, it started Customer Relationship
Management (CRM) and SCM initiatives, which
In 2001, Bharat Petroleum Corporation Limited helped in analyzing customer buying behavior
(BPCL), a leading player in the Indian petroleum and product development with the help of data
industry, successfully implemented an Enterprise mining tools. In 2003, these initiatives played an
Resource Planning (ERP) solution. Implementa- important part in M&M’s farm equipment divi-
tion began in April 2000 after the company decided sion winning Japan’s prestigious Deming award
to integrate all its activities through the ERP pack- (Dutta, 2009).
age SAP R/3. BPCL divided its IT initiatives into a
three-pronged strategy, wherein it planned to create larsen and toubro
a communication network within the organiza-
tion, create a basic information network for the Larsen and Toubro (L&T) is one of the largest en-
entire corporation, and process transactions with gineering companies in India. Its largest division,
customers all over the country. The strategy was ECC handles turnkey industrial and infrastructure
devised after the company divided the organiza- projects in civil, mechanical, electrical, and in-
tion into six Strategic Business Units (SBU) and strumentation engineering. L&T also offers large
conducted a detailed evaluation of the company as turnkey construction services. Headquartered in
a whole. The organization was restructured to help Chennai and with more than 50 years of experi-
focus on specific customer segments and address ence and expertise, ECC commanded a leadership
their individual needs. SAP R/3 helped BPCL to position in the industry. ECC structured itself into
successfully launch its e-business initiatives, the seven regions in order to handle the complexity in
first of which was to allow its customers to track terms of stage of completion, size of project, and
the status of their orders online. This not only the geo-political situation. Each region catered to
allowed the company to retain existing custom- the demand of specialized construction services in
ers, but also helped in attracting new customers. that region and had its own materials department.
According to company sources, BPCL’s biggest The materials department of ECC also got itself
604
ISO-9002 certified; indicating the level of impor- comprising about 4,000 redistribution stockists,
tance, it gave to the management of its suppliers. covering 6.3 million retail outlets reaching the
ECC adopted a sound and efficient supply chain entire urban population, and about 250 million
management strategy in order to keep the supply rural consumers. Hindustan Unilever distribution
chain costs to the minimum. Its quality policies covers over 1 million retail outlets across India
emphasized minimum landed costs, transparent directly and its products are available in over 6.3
practices, and optimum inventory levels. One con- million outlets in India, i.e. nearly 80% of the retail
stant challenge that ECC faced was in integrating outlets in India. HUL, in its endeavor to move
and collaborating its supply chain participants. The from the existing push-based planning system to
root cause of this challenge was the construction a pull-based system, wanted to develop a SCM
industry itself. The project sites being spread out solution that would ensure informed decisions are
at various locations and the large supplier base made during procurement, manufacturing, replen-
made it difficult for ECC to minimize its supply ishment, and distribution. Specifically, the distri-
chain cost. In order to handle the complex chal- bution operation was suffering because of high
lenge of implementing an effective supply chain margin of errors. This problem was compounded
management system in the company, L&T ECC by increasing instances of out-of-stock inventory,
opted to implement a web-based SCM solution which led to demand-supply mismatches. HUL
built on Microsoft technologies - SQL Server wanted a solution that could provide visibility
2000 and Windows 2000 Server in early 2000. across its entire value chain.
The system worked on an online order and of- An IT-powered system has been implemented
fline delivery strategy. The company could now to supply stocks to redistribution stockists on a
streamline its entire vendor network. ECC also continuous replenishment basis. The objective
arranged a proper training program for its vendors is to catalyze HUL’s growth by ensuring that the
in order to familiarize them with the functioning right product is available at the right place in
of the system (Supply Chain Management in L&T right quantities, in the most cost-effective man-
ECC Division, 2009). ner. For this, stockists have been connected with
the company through an Internet-based network,
Hindustan unilever called RSNet, for online interaction on orders,
dispatches, information sharing, and monitoring.
Hindustan Unilever Limited (HUL) is India’s RS Net covers about 80% of the company’s turn-
largest Fast Moving Consumer Goods (FMCG) over. Today, the sales system gets to know every
company, touching the lives of two out of three day what HUL stockists have sold to almost a
Indians with over 20 distinct categories in home million outlets across the country. RS Net is part
& personal care products, and foods & beverages. of Project Leap, HUL’s end-to-end supply chain,
HUL’s brands like Lifebuoy, Lux, Surf Excel, Rin, which also includes a back-end system connect-
Wheel, Fair & Lovely, Ponds, Sunsilk, Pepsodent, ing suppliers, all company sites, and stretching
Close-up, Lakme, Brooke Bond, Kissan, Knorr- right up to stockists. The IT solution for SCM has
Annapurna, and Kwality Walls are household improved HUL’s proactive planning capability,
names across the country and span many categories manufacturing, and distribution efficiency, which
- soaps, detergents, personal products, tea, coffee, has in turn helped the company ensure a more
branded staples, ice cream, and culinary products. responsive and dynamic system of distribution.
They are manufactured over 40 factories across It has enabled the company gain visibility across
India. The operations involve over 2,000 suppli- its supply chain, reduce distribution lead-time and
ers and associates. HUL’s distribution network, minimize cost incurred on its supply chain. Avail-
605
Figure 3. RFID reader and tag communication flow diagram.
ability of stock has increased considerably from 65 supports handling the logistics of moving finished
per cent to 90 per cent. HUL has also been able to goods from the source to stores i.e. transportation
distribute its stocks more equitably with marked and warehouse management systems (Pantaloons
decline in mal-distribution from 19 per cent to 6 - Information Technology in the Supply Chain:
per cent of its total volume transported. Manual Case Study, 2009).
intervention has come down from 40 per cent to
sub-zero levels. Direct dispatches from factories edI Implementations in India
to wholesaler network have increased. Hindustan
Unilever, which once pioneered distribution in The EDI network involves the ministry of com-
India, is today using IT to reinvent distribution merce, customs, airlines, ports, directorate gen-
- creating new channels, and redefining the way eral of foreign trade, banks, regional licensing
current channels are serviced. In the process, it authorities, and other organizations. The major
is converging product availability, with brand EDI users in India presently are governmental
communication and brand experience. Although, departments of commerce and customs, port trusts,
HUL was using the ERP solution, “MFG/ PRO,” shipping agents, couriers, custom house agents,
its parent, Unilever is presently implementing educational institutes, importers/exporters, min-
SAP ERP package across all geographies (Supply istry of transport etc. There are two approaches
Chain Management Solution for Hindustan Lever: to implementing EDI. Many large organizations
Case Study, 2009). acquire or build their own proprietary systems,
often in association with their business partners.
Pantaloon The other approach is to work with a VAN (value
added network) provider like VSNL and NIC,
Pantaloon, one of India’s largest retail apparel which provides EDI transaction services, secu-
giants, uses the SAP retail solution in a bid to rity, document interchange assistance, standard
use IT for supply chain management. The solu- message formats, communication protocols, and
tion supports product development, sourcing, communication parameters for EDI (Report of the
procurement, collaboration, and communication CEFACT Rapporteur for Asia, 2009).
between supply chain partners. The solution also
606
key erP Implementations in India frequency signals. The need to minimize operat-
ing costs and employed assets has resulted in the
All the key players in the ERP global market are adoption of radio frequency technology to track
practically present in India. This includes SAP with inventories within a supply chain down to the item
their flagship product R/3, Baan Company with level, thus reducing channel volume and enhancing
their Baan IV product, Oracle with their Oracle forecasting and planning capabilities (D’Avanzo,
Applications and the world-class ERP product, 2004). In this system, data is transmitted by a
Marshall from the rising Indian star Ramco portable device, called a tag, which is read by an
Systems. The other major players in the global RFID reader and processed according to the needs
ERP Market, namely, PeopleSoft has entered the of a particular application. The data transmitted
Indian market only very recently. Yet another by the tag may provide identification or location
leading product MFG/ PRO from QAD has been information, or specifics about the product tagged,
present for a while with its major two customers such as price, color, date of purchase, etc. (RFID
being Hindustan Unilever and Godrej. SAP has Systems, 2009).
been exceptionally successful in India with nearly Reno GmbH, one of Europe’s largest shoe
two-thirds of the Indian market share. The major companies, operating more than 700 stores in 15
industrial houses Tata, Reliance, Essar, Mahindra, countries, plans to embed wireless RFID chips
and Kirloskar have embraced SAP. Baan has been in shoes sold at stores across the continent. Reno
very successful in major manufacturing companies has been using RFID technology to track product
such as TVS. Oracle has been a playing a dominant shipments from its factories to its stores for sev-
role in the telecom centre with a stronghold among eral years but has not yet used the technology to
all cellular phone companies. Ramco Marshall has track individual products inside each store. For
a good client base among the process industry in individual product tracking, wafer-thin RFID chips
the south and a few public sector undertakings are being designed for shoes from Reno’s Asian
(Sadagopan, 2009). production facilities. By integrating RFID tags
into its shoes, Reno aims to curb theft of boxed
products, products on display, and the shoes cus-
emerGInG And neW InFormAtIon tomers try on inside the stores (Schneider, 2003).
tecHnoloGy tools For Schiff Nutrition International, a midsize company
suPPly cHAIn mAnAGement based in Salt Lake City, is a maker of vitamins and
nutritional supplements. Schiff is in the process
radio Frequency of deployment of the RFID technology in order
Identification (rFId) to continue doing business with Wal-Mart. In
2003, Wal-Mart began setting deadlines for sup-
RFID technology is fast replacing bar code in pliers to start using RFID tags on their shipments
various supply chain operations like inventory (Krotov, 2008).
tracking and management. The advantage that
RFID has over the bar code is that it can uniquely software Agents
identify the specific object such as when items
are produced, which lot does the item belong to, A software agent is a software system that has
and when will the items expire. In bar coding, the attributes of intelligence, autonomy, perception
reading device scans a printed label with optical or acting on behalf of a user. Agents can behave
laser or imaging technology. However, in RFID, autonomously or proactively. The intelligence of
the reading device scans a tag by using radio an agent refers to its ability of performing tasks
607
Figure 4. Web services architecture
or actions using relevant information gathered as used agents to transform its supply chain network
part of different problem-solving techniques such into a network of software agents, whose behaviors
as influencing, reasoning and application specific are programmed through rules (Cobzaru, 2003).
knowledge. One category of agents, monitoring
and surveillance agents are being used to ob- decision support systems
serve and report on equipment, usually computer
systems. These agents can also keep track of Decision Support Systems (DSS) are a specific
company inventory levels, observe competitors’ class of computerized information systems that
prices and relay them back to the company, watch supports business and organizational decision-
stock manipulation by insider trading and rumors, making activities. A properly designed DSS is
etc. Along with data mining agents, agent-based an interactive software-based system intended to
solutions are being considered for applications help decision-makers compile useful information
in SCM. For example, NASA’s Jet Propulsion from raw data, documents, personal knowledge,
Laboratory has an agent that monitors inventory, and/or business models to identify and solve
planning, and scheduling equipment ordering to problems and make decisions. Typical informa-
keep costs down, as well as food storage facilities. tion that a decision support application might
These agents usually monitor complex computer gather and present would be an inventory of all
networks that can keep track of the configura- current information assets (including legacy and
tion of each computer connected to the network. relational data sources, data warehouses, and data
Air Liquide America LP, a producer of lique- marts) or comparative sales figures between one
fied industrial gases, reduced its production and week and the next week. In SCM, there is always
distribution costs using agents. Merck and Co, a a likelihood of having disagreements among
leading research-driven pharmaceutical company parties for a certain decision-making process.
used agents to help it find more efficient ways to This phenomenon gets worse, when the business
distribute anti-HIV drugs. Proctor and Gamble environment becomes more competitive and tur-
608
bulent. Accordingly, decision support systems have electronic commerce

been used in various areas like logistics, inventory
management, facility design, sales analysis etc for Electronic commerce refers to the wide range of
negotiation among supply chain partners. Baan, a tools and techniques utilized to conduct business
leading ERP vendor unveiled an application, Baan in a paperless environment. Electronic commerce
Enterprise Decision Manager for aiding corporate therefore includes electronic data interchange,
decision-making. Major retailers like Wal-Mart, e-mail, electronic fund transfers, electronic pub-
Sara Lee, and Roebuck have increasingly started lishing, image processing, electronic bulletin
using Collaborative Forecasting and Replenishment boards, and shared databases. Companies are able
(CFAR), which uses DSS for jointly developing to automate the process of moving documents
forecasts. GAF Materials Corp, the largest manu- electronically between suppliers and customers.
facturer of asphalt-based roofing materials in the This system provides access to customers all
US, uses a freight-management DSS (Lee, 1999). over the world and thus eliminates geographical
limitations. Some of the e-commerce applications
Web services in the B2C (Business to Consumer) and B2B
(Business-to-Business) space, include e-Tailing
Web services are application interfaces acces- (using the Internet for selling goods and services
sible via Internet standards that use XML and over the Internet, examples are companies such
that employ at least one of the following stan- as Amazon and e-bay); e-Procurement (using
dards: Simple Object Access Protocol (SOAP), the Internet for purchase of goods and services
Web Services Description Language (WSDL) or which are not directly used in the main business
Universal Description, Discovery and Integration of a company like a car manufacturer procuring
(UDDI). These standards, and the next-generation stationery for its employees), and e-Auctions (Sites
standards that are being built on them, are defining on the web, which run conventional auctions).
the way that forward-thinking enterprises manage There are two types of e-Auctions: those that are
lightweight integration tasks. To fulfill orders, the carried out in real-time, where participants log in
retailer has to manage stock levels in warehouses to an auction site using a browser at a specified
(Coronado, 2006) . For stock re-ordering and time and bid for an article until the highest price
replenishment, manual means have been usually is reached and no other bids are forthcoming. The
employed. All this could be automated using web other type of site and the most common, is where
services as shown in example (figure 4). an item is offered for sale and a date advertised
In the example shown in Figure 4, there is one after which no more bids are accepted.
retailer, one logging facility, three warehouses,
three manufacturers, and one configurator, and electronic supply chains
thus a total of nine web services, which are all
integrated using the Universal Description, Dis- Electronic Supply Chains (ESC) refers to those
covery And Integration (UDDI) protocol. UDDI supply chains that are electronically facilitated
creates a standard interoperable platform that between or among participating firms. Also called
enables companies and applications to dynami- virtual supply chains, these are realized in two
cally use web services over the Internet. UDDI is forms, EDI-based or Internet-based. EDI gener-
a cross-industry effort driven by major platform ally connects firms through proprietary Value
and software providers, as well as marketplace Added Networks (VAN), whereas the Internet
operators and e-business leaders (Universal De- generally connects firms through open networks,
scription, Discovery and Integration, 2009). which use standard protocols. The ESC links trad-
609
Figure 5. CPC architecture
ing partners to allow them to buy, sell and move lastminute.com have achieved innovative results
products, costs, the introduction of the Internet using ESCs (Gunasekharan, 2004).
has brought about opportunities that allow firms
to transact with other enterprises electronically. collaborative Product commerce
Amazon is one such example. The virtual supply
chain also envisages use of Internet-based ap- Collaborative product commerce (CPC) is a set
plications to transact and exchange information of tools that allows companies to manage product
like product and inventory information with their information and share that information with sup-
downstream or upstream trading partners. Sup- pliers and partners through the web. CPC tools
ply chain initiatives like Collaborative Planning, typically combine information management,
Forecasting and Replenishment (CPFR), Vendor version-control, configuration management,
Managed Inventory (VMI), Efficient Customer workflow-management, and collaborative tools.
Response (ECR) and quick response have been CPC helps a company’s development strategy to
increasingly facilitated in the new e-supply chain become three-dimensional. This is by facilitat-
paradigm. Information sharing among suppli- ing the collaborative development of products,
ers, manufacturers, distributors, and retailers processes, and supply chain strategies such as
are greatly improved. American-On-Line and Build-to-Order (BTO) (Turner, 2001). In the fu-
610
ture, CPC is expected to co-ordinate and control International, a home furnishings textiles exporter,
virtually all supply, design, manufacturing, and headquartered in the small south Indian town
customer-relationship processes. Collaborative of Karur, has deployed RFID tags to reduce its
Product Commerce (CPC) embraces the product operating costs. As a result of this deployment,
design, development, and introduction processes production efficiency at their plants has improved
as well as the associated management of product by over 5 percent (Giridhar, 2009).
data and requires a coalition of internal and external The total RFID market (including tags, readers,
constituencies. The benefits of developing robust software and services) has been growing exponen-
CPC capabilities can be compelling. They typically tially over the last decade. Today it is around US
come in the form of integration of design processes $ 5.5 billion and the forecast is that it may touch
and supply chains, increased revenue and margin, US $ 25 billion by 2015 of which a large chunk
lower operating costs, better customer response, will come from emerging economies like China
JIT processes, and higher quality improvement. and India. This acceleration will be facilitated
Figure 5 shows components of a typical CPC by rapid declines in the RFID tag prices and ex-
solution (True, 2002). tensive use in various applications. One estimate
suggests that the RFID tag market will by several
billion tags today to tens of trillions of tags by
ImPlementAtIons oF emerGInG 2015 (RFID market to reach $7.26Bn, 2009).
InFormAtIon tecHnoloGy Forecasts also project over 1.3 million jobs will
tools In IndIAn comPAnIes be directly created by RFID technology by 2015
(RFID Workforce: Mining the RFID Technology
Several progressive Indian companies are exten- Gold Rush, 2009).
sively adopting emerging IT tools like RFID, web The electronic business market has also grown
services, electronic commerce, and virtual supply by leaps and bounds with the present market space
chains to shore up their supply chain operations. being estimated at almost US $ 1.8 billion (Ashish,
Leading Indian IT service providers like Tata 2009) with several players like Indian railways,
Consultancy Services, Infosys, and Wipro are skumars.com, e-bay, and all leading banks and
developing RFID-based solutions for supply chain financial institutions like ICICI, HDFC, IDBI etc.
management in sectors like manufacturing, retail, Several large manufacturers in India are also
telecom, government, hospitals, and educational architecting virtual supply chains. Maruti-Suzuki,
facilities like libraries. These solution providers the largest car manufacturer in India, which has
are also developing web services based packages more than 200 vendors and suppliers, has con-
to facilitate virtual supply chains, EDI, e-Tailing, nected to all vendors using web services and
and e-Procurement. Retail giants like Pantaloon, high-bandwidth networking. Vendors are linked
ITC, Wills Lifestyle, Madura Garments, Big Ba- through this Internet-based supply chain, which
zaar, and Total Mall have already implemented maintains online information regarding order
RFID at their retail outlets for product tracking status and delivery instructions. These have sub-
and warehouse management. Automobile major, stantially helped in reducing both inventory levels
Mahindra and Mahindra is another early adopter, and lead times required for the supply of various
using RFID for scheduling and logistics manage- components (Harikanth, 2009).
ment (Nallayam, 2009). The manufacturing giant, However, deployment of tools like DSS and
Ashok Leyland, is using RFID for plant assembly software agents for supply chain management
operations. Even small players have started to use in Indian companies is limited. A few IT service
RFID for their supply chain integration. Sabare companies like PTC and Rolta are offering CPC
611
solutions for manufacturing companies, which chain, the merchandize lies for a lesser time in
are mainly used for computer-integrated design the warehouse, thereby eliminating the need for
and manufacturing. Usage of DSS and software additional warehouse space. Earlier, about 20%
agents for SCM are at the research or experimen- of the non-saleable returns from the stores were
tal stages. Indian companies must rapidly adopt attributed to the reason of mishandling at the
these technologies for SCM and derive potential warehouse. With the project in action, since the
benefits. physical handling at the warehouse is reduced, the
percentage of non-saleable garments has reduced
Itc Wills lifestyle – An considerably. With unique identity of each tag,
rFId success story and faster reconciliation, the magnitude of manual
errors was also greatly reduced.
Wills Lifestyle is an apparel company owned by Every garment was also getting about seven
the diversified ITC group. The chain of Wills days earlier to the store. With some garments
Lifestyle stores offers a complete fashion wardrobe getting as much as an extra 10 to 15 day win-
comprising ‘Wills Classic’ formal wear, ‘Wills dow, ITC lifestyle retailing has seen a 2 percent
Sport’ relaxed wear, and ‘Wills Clublife’ evening growth in sales. In addition, the percentage of
wear, along with accessories for both men and manual errors has been reduced greatly because
women. Wills Lifestyle is also rated as among each tag has a unique identity and reconciliation
the top five luxury brands in India. is more accurate. Assuming a 0.1 percent error
In December 2006, Wills Lifestyle added a rate across three lakh products every month, the
technology enabler, RFID to their showrooms and company estimates saving worth US $ 120,000
warehouses. Deployed across 7 retail showrooms a year (Goswami, 2009).
and 2 warehouses, the project reaped wide-ranging
benefits to internal business users as well as mahindra and mahindra–rFId usage
consumers. RFID proved its mettle in terms of for scm in manufacturing sector
speeding up of existing processes, reducing the
time-to-market, efficient material handling, and Mahindra & Mahindra (M&M) is India’s largest
accuracy of books versus physical stock. Dur- farm equipment company and the largest manu-
ing the bar-code era, each piece of garment was facturer of tractors in India with sustained market
individually scanned and lots of manpower was leadership of over 20 years. Mahindra is also
involved, also at the end of the day accuracy of the first tractor company in the world to win the
stocks received/dispatched at the warehouse was Deming Prize. All of M&M’s plants have been
a key concern. With RFID in place, the time taken certified for ISO 9001, QS-9000, and ISO 14001.
to receive 1 carton (about 30 to 35 garments) was Inventory accuracy, shrinkage control, control-
reduced from about 5 to 8 minutes to about 20 ling delays in vendor shipments, and minimiz-
to 30 seconds. With the usage of RFID tunnels, ing dud production schedules were of critical
the output per person increased from 300 to 400 importance to M&M in their drive to improve
pieces per day to about 2000 to 3000 pieces supply chain efficiencies. The need to maintain
per day. Reduced scanning time enabled faster accurate information and avoidance of human
receiving and dispatch of merchandize, thereby errors were the factors that drove M&M to look
reducing the time to market. Thus, an extra 10 to for real-time data capture solutions. SkandSoft
15 day window was available leading to increased Technologies Private Limited, an upcoming IT
contribution to the business by over 2 percent. Ow- service provider, was chosen to provide an user-
ing to increased inventory velocity in the supply friendly, single-window solution to this complex
612
challenge. SkandSoft, after a comprehensive efficient and effective supply chain management
study of M&M’s existing inbound supply chain, technologies, practices, and policies. Customers’
developed an RFID-based SCM solution. The expectations are also increasing and companies
RFID solution was in tune with the evolving are prone to more and more uncertain environ-
specifications of RFID hardware and systems. It ments in the face of increasing competition. To
was also integrated with the legacy SAP MM and survive in these times, companies will find that
APO modules of M&M. The business results after their conventional supply chain integration will
deploying this solution were complete reduction in have to be expanded beyond their boundaries so
manpower involved in physical count check and as to integrate all stakeholders. Adoption of IT
stock verification. Picking and docking personnel tools is vital for such efforts.
saved 32% of their time. The company achieved This chapter discusses the role of IT as an
near-zero shrinkage of the items covered and 14% enabler in supply chain management and also
reduction in costs by FIFO compliance in storage highlights the vast benefits to companies with
(Success stories: Mahindra and Mahindra, 2009). a comprehensive IT strategy. An overview and
deployment of the present alignments of widely
Fedex India virtual supply chain deployed IT tools like EDI, ERP, bar codes, in-
ventory management, transportation management,
Blue Dart Express, is India’s premier integrated air and warehouse management systems is provided.
express carrier and logistic-services provider. Blue Successful IT implementations of SCM of progres-
dart is also the Indian global service participant sive Indian companies are also described. The basic
of Federal Express (FedEx), the world’s leading elements of emerging and new age tools like RFID,
air express transportation company. Using web software agents, decision support systems, web
services, Blue dart has evolved an e-strategy to services, electronic commerce, and virtual supply
integrate all partners and customers with its e- chains are reviewed. Rapid advances in Internet
shipping tools. All 14 of the company’s warehouses technology have changed the way companies do
are connected to this system. All customers can business. Enterprises must harness the power
track their shipments through single or multiple of these emerging information technology tools
waybills online. This solution is also ported to an for supply chain integration and collaboration.
e-mail or mobile response system. Customers can While it has been observed that usage of tools like
register and gain access entire waybill tracking RFID, web services, electronic commerce, and IT
data, which are available for 45 days online. This solutions to facilitate virtual supply chains, has
web-enabled solution with additional features been on the rise, Indian companies are yet to tap
like e-logistics, e-fulfillment, inventory control the vast potential of the application of software
etc, has also been integrated to the ERP system, agents and decision support systems in supply
“COSMATH,” which has been indigenously de- chain management.
veloped by the company (Federal Express India
supply chain, 2009).
reFerences
conclusIon Abbas, M. (2009). ITC’s e-Choupal: Land-

mark in rural deployment. Retrieved April 15,
Industry trends like globalization, outsourc- 2009, from http://www.cxotoday.com/India/
ing, customization, time to market, and pricing Case_Study/ITCs_e-Choupal_Landmark_in_Ru-
pressure have compelled enterprises to adopt ral_Deployment/551-99474-1004.html.
613
Anderson, D. L. Britt., F. E., & Favre, D. J. Dutta, S., & Kumar, A. (2009). Mahindra &
(1997). The seven principles of supply chain Mahindra’s E-Business Initiatives. ECCH Case
management. Supply Chain Management Review, Collection. Retrieved May 12, 2009, from from
(Spring), 31-41. http://www.asiacase.com/ecatalog/NO_FIL-
TERS/page-ERP-602916.html
Ashish, D. (2009). E-commerce in India. Plugge-
din. Retrieved April 20, 2009, from http://www. ERP Implementation at BPCL. Case Study. (2009).
pluggd.in/india/ecommerce-in-india-irctc-con- Retrieved May 2, 2009, from http://www.icmrin-
tribution-3961/ dia.org/casestudies/catalogue/IT%20and%20
Systems/ITSY005.htm
Auramo, J., Kauremaa, J., & Tanskanen, K. (2005).
Benefits of IT in supply chain management: An Federal Express India supply chain. (2009). Re-
explorative study of progressive companies. trieved May 15, 2009, from http://www.scribd.
International Journal of Physical Distribu- com/doc/2165098/Supply-Chain-Management-
tion and Logistics Management, 35(2), 82–90. at-Fedex-India
doi:10.1108/09600030510590282
Giridhar, C. (2009). India tests RFID waters.
Berger, A. (2009). Five steps to an eSynchronized Electronic Busines. Retrieved May 18, 2009, from
Supply Chain. Retrieved April 27, 2009, from http://www.edn.com/article/CA6388858.html
www.accenture.com/NR/rdonlyres/18099CFB-
Goswami, K. (2009). RFID Speeds Up Wills
1D5F-4FA7-BBC4-862EC465123D/0/esynchro-
Lifestyle Business. Real CIO World. Retrieved
nized_supply_chain_pov_ref.pdf
May 18, 2009, from http://www.cio.in/case-study/
Cobzaru, M. (2003). Agent-based Supply Chain rfid-speeds-wills-lifestyle-business
Management System. Unpublished doctoral dis-
Gunasekharan, A., & Ngai, E. W. T. (2004). Virtual
sertation, University of Calgary, Calgary, Canada.
Supply Chain Management. Production Planning
Coronado, A. E., Lyons, A. C., Michaelides, and Control, 15(6), 584–595. doi:10.1080/09537
Z., & Kehoe, D. F. (2006). Automotive supply 280412331283955
chain models and technologies: A review of
Haag, S., Cummings, M., & Philips, A. (2006).
some latest developments. Journal of Enterprise
Management Information Systems for the Infor-
Information Management, 19(5), 551–562.
mation Age. New York: McGraw Hill.
doi:10.1108/17410390610703675
Harikanth, S. (2009). Maruti Value Chain. OP
Crosby, T. (2009). How Inventory Management
Papers.com. Retrieved May 20, 2009, from
Systems Work. Retrieved May 4, 2009, from
http://www.oppapers.com/essays/Maruti-Value-
http://communication.howstuffworks.com/how-
Chain/167129
inventory-management-systems-work.htm
Harland, C. M. (1996). Supply chain manage-
D’Avanzo, R., Starr, E., & Lewinski, H. V. (2004).
ment: relationships, chains, and networks.
Supply chain and the bottom line: A critical link.
British Journal of Management, 7, 63–80.
Outlook: Accenture, 1, 39–45.
doi:10.1111/j.1467-8551.1996.tb00148.x
Description, U. Discovery and Integration. (2009).
Johnson, A. H. (2002 September 30). Wal-Mart
Retrieved May 12, 2009, from http://uddi.xml.org/
put intelligence in its inventory and recognized
the value of sharing data. Computerworld.
614
Kannabiran, G., & Bhaumik, S. (2005). Corpo- Rushton, A., Oxley, J., & Croucher, P. (2000). The
rate Turnaround through effective supply chain Handbook of Logistics and Distribution Manage-
management: The case of a leading jewellery ment. Glasgow, UK: Bell & Bain Ltd.
manufacturer in India. Supply Chain Management,
Sadagopan, S. (2009). Freqently asked ques-
10(5), 340. doi:10.1108/13598540510624160
tions on ERP in India. ERP Study centre, IIM-B.
Krotov, V., & Junglas, I. (2008). RFID as a dis- Retrieved April 17, 2009, from http://www.iiitb.
ruptive innovation. Journal of Theoretical and ac.in/ss/erp-faq/main4pg1.htm
Applied Electronic Commerce Research, 3(2),
Sammon, D., & Hanley, P. (2007). Case Study:
44–59. doi:10.4067/S0718-18762008000100005
Becoming a 100 per cent e-corporation: ben-
Lee, C., Lee, K. C., & Han, J. H. (1999). A Web- efits of pursuing an e-supply chain strategy.
based Decision Support System for Logistics Supply Chain Management, 12(4), 297–303.
Decision-Making. Paper presented at the meeting doi:10.1108/13598540710759817
of the Academy of Information and Management
Schneider, M. (2003). Radio Frequency Identifica-
Sciences, Myrtle Beach, SC.
tion (RFID) Technology and its Applications in the
Macleod, M. (1994, June). What’s new in supply Commercial Construction Industry. Unpublished
chain software? Purchasing & Supply Manage- doctoral dissertation, University of Kentucky,
ment, (pp. 22-25). Lexington, KY.
Nallayam, R. (2009). RFID Reality Check. Chan- Shore, B. (2001). Information sharing in global
nelworld.in. Retrieved May 16, 2009, from http:// supply chains. Journal of Global Information
www.channelworld.in/specialreports/index.jsp/ Technology Management, 4(3), 27–46.
artId=5013515
Simchi-levi, D., Kaminsky, P., & Simchi-levi, E.
Pantaloons - Information Technology in the Supply (2003). Managing The Supply Chain: The Defini-
Chain. Case Study. (2009). ECCH Case Collec- tive Guide For The Business Professional. New
tion. Retrieved April 16, 2009, from http://www. York: Irwin /Mc Graw Hill.
icmrindia.org/casestudies/catalogue/Supply%20
Srinivasan, K., Kekre, S., & Mukhopadhyay, T.
chain%20Management/CLSCM002.htm
(1994). Impact of Electronic Data Interchange
Report of the CEFACT Rapporteur for Asia. Technology on JIT Shipments. Management
(2009). Retrieved April 19, 2009, from http:// Science, 40(10), 1291–1304. doi:10.1287/
www.unece.org/trade/untdid/download/99cp3.rtf mnsc.40.10.1291
RFID market to reach $7.26Bn in 2008. (2009). Success stories: Mahindra and Mahindra. (2009).
IDTechEx. Retrieved May 27, 2009, from www. Retrieved May 17, 2009, from http://www.skan-
idtechex.com/products/en/articles/00000169.asp dsoft.com/successstories.html#1
RFIDSystems. (2009). Retrieved May 10, 2009, Supply Chain Management in L&T ECC Divi-
from http://epic.org/privacy/rfid sion. (2009). ECCH Case Collection. Retrieved
April 29, 2009, from http://www.icmrindia.org/
Ross, J. (2006). Enterprise Architecture As
casestudies/catalogue/Operations/OPER045.htm
Strategy: Creating a Foundation for Business
Execution. Cambridge, UK: Harvard Business
School Press.
615
Supply Chain Management Solution for Hindu- True, M., & Izzi, C. (2002). Collaborative product
stan Lever. Case Study. (2009). Retrieved May commerce: creating value across the enterprise .
3, 2009, from http://fmcg-marketing.blogspot. Ascet, 4, 27–35.
com/2007/10/supply-chain-management-solu-
Van Oldenborgh, M. (1994). Distribution super-
tion-for.html
highway. International Business, 7(6), 80–84.
Thongchattu, C., & Buranajarukorn, P. (2007). The
Workforce, R. F. I. D. Mining the RFID Technol-
Utilisation of e-Tools of Information Technology
ogy Gold Rush. (2009). RFID Tribe. Retrieved
Towards Thorough Supply Chain Management.
May 27, 2009, from www.rfidtribe.com/news-
Paper presented at the Naresuan University Re-
05-05-11.html
search Conference, Thailand.
Towill, D. (1997). The seamless supply chain -
the predator’s strategic advantage. International
Journal of Technology Management, 13(1), 37–56.
doi:10.1504/IJTM.1997.001649
616
617
Chapter 3.4
Transforming Compensation
Management Practices
through Web-Based
Enterprise Technologies
Xiaoya Liang
Fudan University, China
IntroductIon developed using Internet technology, and to offer

some insights on understanding emerging HRIS
Human resources management system (HRMS) is trends and new strategic roles compensation
one of the core components in any major enterprise professionals play. To do this, I will start with an
resource planning (ERP) application suite since its overview of key features of a good Web-based
earliest development. As information communica- compensation program and main benefits of such
tions technologies progress, ERP vendors are able program. I compare common and unique features
to incorporate more sophisticated HR functions in performing compensation functions offered by
into their products in an increasingly user-friendly both major ERP vendors. Implications for HR
manner. HRM modules in ERP suites cover a full professionals and researchers will be discussed
spectrum of conventional HR functions ranging in the concluding section.
from recruitment, selection, compensation, and
so on. Meanwhile, there is also an explosion of
growth in the number of specialized HR software bAckGround
products and application service providers in the
market place. Compensation management programs were
The goal of this chapter is to provide a current, among the earliest HR applications of ERP. As
non-technical review and comparison of major IT advances, compensation software has evolved
compensation software products and services rapidly in the past decade. Recently, major ERP
vendors brought in the Internet technology in their
DOI: 10.4018/978-1-59904-883-3.ch129
Transforming Compensation Management Practices through Web-Based Enterprise Technologies
latest products. This Web solution is revolution- tem. Customized reports are also available to meet
izing how compensation systems are managed in specific needs. Managers can view aggregated
organizations now and in the foreseeable future reports or drill down the reports by department,
(Meade, 2003). Here are some key features and by work groups, or by individual employees
benefits of such a program within an ERP system. (Dulebohn & Marler, 2005). Users can conduct
24/7 accessibility: A Web solution means what-if scenario analyses and simulations in plan-
users can access the program using a standard ning compensation budgets which improve the
Web browser wherever and whenever they wish quality of strategic compensation decision making.
to. Compensation professionals no longer need User-friendly interface: Built on relational
to sit in front of their desktops at central offices data base structure, the Web-based compensa-
in order to process information. Employees can tion software leaves the control of the system in
view their pay and benefits information and update hand of end-users with minimal technical savvy.
their personal profile at home or even at vocation. Unlike prior software systems, it requires no
Integrated functionality: Compensation sys- system-specific training and keeps IT support and
tems are integrated with some other HR and non maintenance at a minimum.
HR systems within the organization. For example, Real-time accurate data: The HR department
a compensation professional can track up-to-date is no longer the sole party responsible for enter-
employee attendance information or performance ing all the data. Employees and line managers
reviews to make corresponding changes in com- are empowered to enter and update data on their
pensation. side. Thus, information is being updated on the
Data automation: Automation of existing Web as it occurs. All the analyses are conducted
processes and procedures has cost saving benefits. using real-time data (Brink & McDonnell, 2003).
Significant data automation relieves HR profes- One stop information center: Human resources
sionals from tedious routine tasks such as data home pages accessible via Internet or company
entry, filing, and report writing. Data importing intranet provide links to Web sites of outsider
and exporting are made easy between other office service providers such as employee stock ad-
applications software. ministrators or health care providers (Gherson
Streamlined workflow: The Web solution & Jackson, 2001).
streamlines all aspects of compensation planning Add-on applications: Self-service through
and implementation, including plan configura- ERP applications enables HR professionals,
tion, modeling and budgeting, plans review and employees, and line managers to focus on their
approval, and data exporting to payroll. In addi- primary value, adding tasks and spending less
tion, by posting current information regarding time on administrative tasks. Line managers can
compensation policies, program description, online view salary budgets, compare budgeting
eligibility explanation, new compensation forms, against actual spending, and take various sal-
and frequently asked questions and answers on the ary actions easily with self-service functions or
Web, that individual employee can access the HR corporate portals (Adamson & Zampetti, 2001).
department and saves a substantial portion of time, Communications between management and em-
which the HR staff used to spend on responding ployees are made easier. MSS allows managers
to requests from employees. take ownership in making compensation decisions
Flexible analytical tools: New compensa- with easy-to-use analytical tools (Gueutal, 2003).
tion software has strong analyzing capabilities. Cedar’s third annual self-service survey shows a
It provides users with online reviews of various continued expansion of self service applications
compensation reports that are pre-built in the sys- on the Web, and most surveyed companies report
618
business success with HR self service implemen- Internal structure

tation (Cedar Survey, 2001).
An internal structure includes a hierarchy of job
levels, pay differences among the job levels, and
comPensAtIon FunctIons the criteria used to determine the pay differences
comPArIson (Milkovich & Newman, 2005). To assure internal
equity in its compensation system, the organization
The level of sophistication and the speed of de- needs to conduct job evaluations, and compile
velopment of compensation software products pay grade and performing competency analysis.
indeed mirror the rapid-changing, dynamic, and Job evaluation: A systematic job evaluation
complex business reality of nowadays compensa- starts with job analysis. With the Web-based
tion management. compensation software, an in-house job analysis
Individual pay is no longer just monthly cash is usually performed using an online questionnaire
salary. Instead, it takes various forms, including to collect information directly from employees,
stock ownership and flexible benefits. Firms supervisors, HR, and outside subject matter
have to be innovative in developing employee experts. After the surveys are administered, the
compensation packages to attract talents. An effec- software analyzes the data with pre-built statisti-
tive compensation system is designed to address cal techniques and automatically generates a job
business objectives and align with key business description per job surveyed. Such job descrip-
operations (Gerhart, 2000). To stay competitive, a tion/analysis is used as bases for job evaluation.
firm needs to constantly compare its compensation The next step is to compute the relative value of
structure with major competitors in the market each job. Although Web-based programs support
places. There are also ever-changing external fac- multiple job evaluation methods by user’s defini-
tors such as legal regulations and labor economy tion, the point method is the most commonly used
that a firm has to closely monitor. Milkovich and one. Users need to input a detailed organizational
Newman’s (2005) four-component model summa- chart to clearly define job hierarchies and report
rizes key functions within a compensation system. relationships among them. At this point, market
The four components include internal structure, salary information can be incorporated as point
external structure, pay for performance tools, and of references in deciding job worth. Once users
administration tasks. We adopt these four broad define compensable factors with scoring rules, the
categories to compare major Web-based software actual computation is executed by the software.
on specific compensation functions. Below I will Pay grade: The system allows users to define
review how major ERP vendors and specialized grading structure and create multiple grades for
software vendors incorporate these functions un- multiple jobs across the organization. Users can
der the four categories in their product offerings. define pay range for each pay grade using imported
The discussion focuses on the latest products from external market salary data or existing internal
major software vendors including PeopleSoft®, data to set up a pay structure for employee base
Oracle®, and SAP. Frantzreb (2004) offers a salary. Rewards and salary scales are linked to
comprehensive guide of compensation software employee grading structure.
products in the market. Competency analysis: It is critical to keep
the competency analysis current given the fast
changing nature of work. The Web-based program
enables HR professionals to keep track of compe-
tency sets of current jobs and current employees.
619
Standardized functions are part of the new system rable compensation questionnaire which allows
to record knowledge, skills, and abilities (KSAs) customization. Participating organizations can fill
by job, by employee, or by job family. To utilize out the surveys online. Data are entered online and
integrated functionality, competency analysis is are processed to generate wide-range statistics and
tied with performance appraisal, training, recruit- reports. The system can import salary data from
ment, knowledge management, and personnel external data source for comparison and analysis or
administration. For example, an employee’s export internal compensation and benefits details
competency profile will be updated when she/he to spreadsheets for survey compassion.
accomplishes new training or obtains new certifi- Within MySAP HR suite, a module called
cations. Ability to track individual competencies SAP Benchmark Infocube is specially designed
progress is very important for organizations to for market salary data processing. This program
base their pay structure on competency developed performs comparative analysis of compensation
in a competency-based pay plan. packages (including base salary, variable pay, and
PeopleSoft8® includes comprehensive knowl- benefits) using internal and external data with a
edge base for tracking, understanding, and devel- variety of custom report options.
oping employees’ skills base. Examples of this Pay lines represent market pay rates for an
feature are unlimited competencies and review array of benchmarked jobs. An organization uses
ratings. The new system improves performance pay line information to create ranges for each pay
appraisal process to accommodate the needs for grade reflecting its pay policy.
conducting 360 degree job evaluation online. Regression analysis is commonly used to
This process provides multi-source assessment construct a pay policy line with average market
of employee’s skills and job performance and pay rates. Pay lines then define a mid-point base
invaluable feedbacks for future improvement. salary for each pay grade. Depending on how an
Oracle® records employee skill qualification, organization wants to match with market rates, the
competencies, and experiences from their hiring range of minimal and maximal pay can readily
date. Performance appraisal is planned and con- identified. Any Web-based compensation program
ducted online so that the system tracks recent and should include a standard wage/salary table. Users
historical performance appraisal records. set up standard wage/salary tables per job with
Standardized online employee reviews and minimal, maximal, and mid-point base salary
appraisals are offered by My SAP HR. based on pay line information. Standard salary
tables, individual skills and competency profiles,
external structure and job evaluation points make up the manager’s
workbook to determine individual pay levels.
To maintain external equity of its pay practices,
an organization needs to gather market salary data Pay for Performance Programs
and construct pay lines for common job categories.
Salary survey data gives the individual organi- To attract and retain top talents, organizations
zation an estimate of market price of certain jobs. invented various pay for performance programs
Conducting a salary survey requires a great deal such as merit pay, bonus pay, stock options, profit
of time and resources input from compensation sharing, and a few. These pay-for-performance
professionals with the conventional pencil-and- programs are presumably able to differentiate
paper based method. between top performers and bottom performers by
New Web-based compensation programs speed tying their pay levels with individual performance
up collecting salary data with a built-in configu- or their contributions toward organizational goals.
620
The basic logic seems simple, but designing and eligibility rules and allocation rules. Managers
implementing such a plan is never easy. To accom- can review and approve bonus pay plans. Data are
modate multiple pay-for-performance plans in one exported to payroll within the integrated system.
integrated compensation system is challenging. Stock options: While managers can grant em-
In designing software programs, venders had to ployee stock options as part of their compensation
strive to balance level of specificity and breath. packages using either managerial self-service
ERP applications are kept flexible and let users application tools or user-defined rewarding plans
define their own variable pay plans. functions within standard compensation systems,
The variable compensation module in People- employee or executive stock option administration
soft® suite can be administered by group, by is not a covered function in common compensa-
individual, or on an ad-hoc basis. Users define tion modules. PeopleSoft® is one exception. The
standard reward rules and identify eligibility PeopleSoft® stock administration module allows
rules to link rewards to jobs, grade, groups, or employees to view their personal stock option
departments. The system supports multiple plans information online, model future stock earnings,
including commission sales plans. Users define check vesting periods, and exercise options. The
commission sales plans with multiple quotas, module helps managers to design and implement
revenue targets, and rate tables per employee. employee stock grants or purchase plans.
Payment methods and schedules, incentives by Incentive plans: Oracle® Incentive Compen-
specific types or products, or services are avail- sation allows users to define revenue classes,
able in the program. Oracle offers similar general compensation terms, rate tables, and quotas.
variable pay functions as Peoplesoft® does. Special features include sales person subledger,
Compensation applications in My SAP HR al- rule-based collection and revenue classification,
lows managers to design and implement innovative credit receiver, and manual adjustment.
reward plans including performance-based pay,
competency-based pay, and various short-term salary Planning
and long term incentives.
Merit pay: The common practice is to tie merit Annual salary planning is the single most important
increase with performance appraisal. New online compensation function and it involves not only
performance appraisals save HR professionals HR professionals, but also line managers and
time from compiling data and coordinating re- other departments such as finance and accounting.
view processes. Managers can view individual Hours of time are spent each year in designing and
performance ratings online and allocate merit administering compensation planning and often
increases using tools such as the performance/ times the plans did not work well.
increase percentage matrix. The matrix provides Web-based compensation software is designed
recommended percentage increase quartile by to streamline the entire compensation planning
rating scores. Managers make informative deci- process as best practices are modeled in the system
sions on individual merit increases. The system to share the success. As an example, Web-based
calculates pay increases and generates reports and compensation tools enable Dell™ to reduce
graphical charts. Data are shared with all related planning cycle from 8 weeks to 3 (Gherson &
systems such as payroll. Promotion pay increase Jackson, 2001).
is handled in the system in a similar manner. PeopleSoft® supports budgeting and salary
Bonus pay: Unlike merit pay, bonus pay is planning by groups and allows multiple budgets.
one-time lump sum money paid to individuals for The system includes employee review functions to
reasons such as recognition rewards. Users define accommodate the needs of employee participation
621
in the planning process. Roll-up budget reports conclusIon

and trend reports help executives to view overall
compensation budgets. Managers can monitor total Compensation expenditure is a firm’s largest
HR costs by headcount or by other user-defined operating expense (Lawler, 2001). An effective
factors. Budget reports can be viewed at all levels compensation program can turn such expenses
from individual jobs, pay grades, teams, depart- from the cost of doing business into a power-
ments, to organizational levels. Interactive review ful strategic lever which will be translated into
gives users flexibility to get information that meets competitive edge.
individual needs. The system is able to perform The pace of technological changes is inevitably
compa-ratio penetration analysis by grade or job challenging the traditional roles of compensa-
and to generate various legal compliance reports. tion professionals. New compensation software
Global payroll engine is a unique feature of took away routine administrative tasks like data
PeopleSoft® offering which enhances interna- entry and compiling and automated these tasks
tional payroll capabilities. It delivers predefined with greater efficiency. More sophisticated ad-
country specific compensation rules and offers ministrative tasks like stock options administra-
extensive expatriate compensation management. tion and salary surveys are turned into hands of
Group build module provides a centralized area of outside specialists. In addition, with Web-based
functionality that enables users to define a group’s self-service tools, line managers are empowered
membership based on any user defined criteria. to make compensation decisions for their staff.
This function facilitates the administration of Employees have online access to most personal
team-based compensation plans. information with self service functions. What are
Oracle® offers some similar budgeting func- left with compensation professionals to perform
tions such as supporting multiple budgets, reports in their jobs? Here are some proposed new roles
rollups, and drill-downs. A unique feature of the for e-compensation professionals.
Oracle product is the simulated what if planning Strategic compensation innovator: A central
analysis. This function helps managers understand role e-compensation professional will play is to
the consequences of various planning scenarios. design and modify compensation plans to assure
The system can generate reports to track budget the alignment between compensation strategy
fulfillment. and business strategy. They will be focusing on
MySAP HR supports personnel cost planning optimizing compensation management processes.
and simulations. When planning, managers can Any advanced computer software is no more than
take compensation relevant data on organization an application tool and it by itself does not auto-
objectives into account. Compensation budgets matically generate business value. Some process
are generated based on input from line managers, reengineering is necessary in order to fully utilize
accountings with integrated headcount planning software features.
capabilities. A centralized employee database is HRIS strategist: E-compensation professionals
integrated with performance management, train- are the ones in an organization who understand
ing, and staffing and recruiting and data are entered the evolutions of compensation-related software
once and shared with all other relevant system. and appreciate present-day and future products
development (Martin, 2001). More importantly,
they will be in charge of phasing proper time
lines of major system application reinvestment
and updates.
622
In-house compensation expert and consultan:. Gueutal, H. G. (2003). The brave new world of
E-compensation professionals will convert their HR. In D. L. Stone et al. (Eds.), Advances in
compensation expertise into strategic initiatives by human performance and cognitive engineering
providing executives and line managers with the research (vol. 3, pp. 13-37). Elsevier.
latest global and industry-specific best practices,
Keebler, T. (2001). HR outsourcing in the Internet
current research, and survey findings. They work
era. In A. J. Walker (Ed.), Web-based human re-
together with executive teams on new business
sources (pp. 259-276). New York: McGraw Hill.
solutions (Ulrich, 2000).
Lawler, E. (2000). Rewarding excellence. San
Francisco: Jossey-Bass.
reFerences
Martin, T. (2001). Leveraging technology to
Adamson, L., & Zampetti, R. (2001). Web-based communicate total rewards. The next frontier:
manager self-service: Adding value to the work. Technology and total rewards. Burlington, MA:
In A. J. Walker (Ed.), Web-based human resources IHRIM/World at work.
(pp. 24-35). New York: McGraw Hill. Meade, J. G. (2003). The human resources software
Anderson, L. (2001). Understanding PeopleSoft handbook: Evaluating technology solutions for
8. Sybex. your organization. San Francisco: Jossey-Bass.
Brink, S., & McDonnell, S. (2003). IHRIM go-to- Milkovich, G., & Newman, J. M. (2005). Compen-
guides: e-compensation, the emerging technology sation (8th ed.). New York: McGraw-Hill Irwin.
series. Burlington, MA: IHRIM. Peterson, W. J., Gelman, L., & Cooke, D. (2001).
Cedar (2001). Human resources self-service sur- ERP trends. New York: The Conference Board.
vey. Baltimore: Cedar Inc. Ulrich, D. (2000). From e-Business to e-HR. Hu-
Dulebohn, J., & Marler, J. (2005). E-compensa- man Resource Planning, 23(2), 12–21.
tion: The potential to transform practice? In H. Walker, A. J. (Ed.). (2001). Web-based human
Gueutal & D. Stone (Eds.), The brave new world resources. New York: McGraw Hill.
of e-HR. San Francisco: Jossey-Bass.
Frantzreb, R. B. (2004). Compensation admin-
istration software directories. Roseville, CA: key terms And deFInItIons
Advanced Personnel Systems.
Competency Analysis: People-based ap-
Gerhart, B. (2000). Pay strategy and firm perfor-
proach of job evaluation. It identifies key skills,
mance. In S. R. B. Gerhart (Ed.), Compensation
abilities, and knowledge associated with a par-
in organizations: Current research and practice.
ticular job.
San Francisco: Jossey-Bass.
Employee Self Service: An application
Gherson, D., & Jackson, A. P. (2001). Web-based designed for employees to access related HR
compensation planning. In A. J. Walker (Ed.), information. The ESS system allows employees
Web-based human resources (pp. 83-959). New to input and update personal data, manage benefit
York: McGraw Hill. packages, view compensation information, sign
up for training programs, and review internal job
postings.
623
HR Portal: The integrated Web site for em- Pay Lines: Market pay rates for an array of
ployees to visit for HR information and services. benchmarked jobs. An organization uses pay line
It allows employees to customize the appearance information to create ranges for each pay grade
and the content of their own HR portal. reflecting its pay policy.
Job Evaluation: Defines the content, respon- Manager Self Service: An application de-
sibilities, and skill requirement of a particular job signed to assist line managers in HR related
position. Methods used to conduct job evaluation responsibilities. Major functions covered in MSS
are ranking, classification, and point method. are compensation administration, performance
Pay Grade: Grouping jobs with similar value management, recruitment and staffing, time and
based on job evaluation into the same pay range. attendance, and training.
This work was previously published in Encyclopedia of Human Resources Information Systems: Challenges in e-HRM, edited
by Teresa Torres-Coronas and Mario Arias-Oliva, pp. 877-882, copyright 2009 by Information Science Reference (an imprint
of IGI Global).
624
625
Chapter 3.5
Business Process Management
as a Critical Success Factor
in EIS Implementation
Vladimír Modrák
Technical University of Košice, Slovakia
Nowadays, the implementation of business process It is no doubt that the business process management
management modern tools in companies becomes (BPM) gives companies the fundamentals they
a mater of acceptation of an effective organization need to continuously improve enterprise processes
management. The first ultimate precondition for and structure and get more from their existing ERP
achieving this goal is a properly structured company. investments. It also known, that many approaches
An attention in the study is placed on business pro- have been evolved from research on BPM. On the
cess reengineering due to preparing preconditions other hand there is agreement that business process
for smooth implementation of enterprise information modeling must be an integral part of the methods
system (EIS). Since there are differences between for business process management (Hess & Brecht,
tools of business processes redesign and informa- 1995; Scheer & Nüttgens, 2000; Scheer, 1998). The
tion systems development, then a main focus was effective business process management depends on
on overcoming existing semantic gaps. With aim how well it defines responsibilities and forces an
to solve this problem the specific modeling method employee to take control of their own performance.
has been used that was clear for company’s staff and The first ultimate precondition for achieving this
usable for EIS designers. Used modeling approach goal is a properly structured company.
was supported by QPR software. Since management of Original Equipment
Manufacturing (OEM), where presented study was
conducted, was not confident whether company’s
DOI: 10.4018/978-1-60566-892-5.ch003
Business Process Management as a Critical Success Factor in EIS Implementation
internal and external activities are univocally organization models and IS architecture-designing
defined, this reality motivated given company to issues are described in the following section. Cur-
change this situation. Particularly, in coherence rent directions in business process modeling are
with ISO 9000:2000 their focus had been oriented dealt in section 3. Then selected methodological
on gradual transformation of functionally oriented aspects of the process modeling technique used
management to process-oriented management for the creation of business process models in
system. In this effort they went through more real case are presented. The final sections discuss
developmental periods that are described in this future trends and some decisive findings from
study. A concurrent attention had been placed on the case study.
business process redesign due to preparing precon-
ditions for smooth implementation of enterprise
information system. bAckGround
The Company pays special attention to the
technical innovations by promoting a modern It is common knowledge that the introduction of
quality management style. Similarly as many other Enterprise Information Systems in SMEs requires
small and medium enterprises (SMEs), company a principal restructuring of the Business processes
started with the traditional vertically arranged (BP) and management systems of an organization,
organizational structure that was represented in a simply Enterprise Reengineering with its typical
graphical form by an organizational chart. It was a steps (see figure 1).
hierarchical structure with a chief executive officer Usually, reengineering project in organization
and other executive at the top, small number lay- presents a complex task, normally assigned to
ers of management below this, with the majority external consultants, with the aim to transform
of employees at the bottom of the pyramid. The the current functionally-oriented organizational
jobs in given organizational structure was grouped model to process-based organizational pattern.
by function into departments such as accounting, Greasley (2004) states that a process-based ap-
sales, human resources, and so on. proach plays crucial role in the implementation
The chapter is structured in the following way. of an information system and claims that process
A theoretical background on an interrelation of mapping and business process simulation support
Figure 1. Phases of enterprise reengineering
626
of current process design. When we want to get the between the system and its environment (Castro et
pertinent information elements into the informa- al., 2002). Zhuang et al. (2007) define a semantic
tion system, we should name the unnamed, make gap in an information system as ‘the inefficacy in
them explicit, and organize them into a simple information transmission through representation,
structure and process. Accordingly, it could be especially when an information bearer is unable
seemed that process based approach is implicit to carry the information that it is supposed to.
in each IS designing technique. In reality, pro- Accordingly, semantic gaps between IS users
cess based approach in IS design rather means a and designers is one of the potential realm in the
transformation of process-oriented organizational deployment of the enterprise information systems
model into a IS design model. applications. Basically, this phenomenon results
Accordingly, organizational structure models from different views of users and designers to IS
might correspond with the architecture of an EIS and their different knowledge levels. The users of
as much as possible. According to Vasconcelos et IS understand it in simplified way as an operation
al. (2005), application architecture of enterprise automation tool, mainly used in data processing
IS has to be a part of an enterprise organizational and accessing. Designers of IS or its part under-
model. In reality, a consistency of enterprise or- stand it principally as HW/SW subsystems and
ganization models and architectures of enterprise their interaction. These contradictory notions are
information system is more or less unusual. In this shown in Figure 2, in which chosen components
connection Kalnins et al.(1998) compared tools are used as an example (Modrák, 2007).
and languages for business process reengineering It is obvious that outlined semantic gap cannot
and found out that ‘higher level business goals be replaced completely. The goal should be to
have not got a real semantic link with other parts reduce it to a level allowing relieving the shortage
of modeling’. The reason of inconsistency of en- made by it, when creating and deploying IS. The
terprise organization models and architectures of way of minimizing this semantic gap is possible
enterprise information system is that information through process modeling approach leading up
system development methodologies have tradi- to merge of process models to process maps in
tionally been inspired by programming concepts, the phase before IS design and development. A
not organizational ones, leading to semantic gaps precondition for exploitation of such process maps
Figure 2. Different views of users and designers to IS
627
is understandability for users and simultaneously current dIrectIons In

usability for IS developers (consultants). In op- busIness Process modelInG
posite way, even though the EIS is sophistically
designed, possible deficiencies are practically According to Farnken et al. (1997) models of
anticipated. In other words, to ensure better oper- business processes can be dividing into exten-
ability of enterprise activities through IS without sive and intensive model structures. In extensive
detailed knowledge of process relations in various model structures of the business processes are the
levels of decomposition of enterprise activities processes described as an integrated whole. With
is rather complicated. This potential risk is not a such a model, investigation is not concentrated
crucial issue for designers of IS, moreover if they on internal process structure and its behavior.
know that client/user is responsible for informa- An extensive model describes the process from
tion requirement determination. But companies the viewpoint of its environment, which usually
are not fully known of this reality. It confirms the consists of external or internal suppliers and
fact that companies mostly use simple models of customers. On the contrary, intensive structures
organization structures without formal determina- models of business processes are used to describe
tion of work and information flows and reflects the entities interactions inside the part of the
a lack of process management in organization. system that is being investigated. Such models
By this thesis, IS architecture and subsequently describe business processes from the viewpoint
effectiveness of EIS strongly depends on maturity of their internal objects, sources and other enti-
of process management in organization. This is a ties such as the staff, technological components,
role of Business Process Management that pres- protocols, etc..
ents systematically managing all of a company’s Intensive models of processes structures can
business processes and attempts to continuously be additionally divided into:
improve processes in organization.
A successful transformation of process-orient- • Workflow-oriented model structures,
ed organizational model into a IS design model which represent the behavior of a business
requires being accepted two basic conditions: process from the perspective of a single
item that crosses through the process.
• process based model of organization has to • Functional-oriented model structures that
be continuously optimized and sufficiently reveal the behavior of functional units
unambiguous, (departments) from the viewpoint of sub-
• notation of process based model of orga- sequent business functions. These models
nizational will be equally familiar for two represent the obsolete functional approach
sides, of organization structuring.
• involved company’s staff and designers of
IS. Technically, horizontal structuring and vertical
structuring of business process models are mostly
Then such models form the basis of compre- recognized (Jonkers, 1997). Horizontal structuring
hensive business reference, detailing how the helps to handle each customer engagement with
entire operation in company fits together (White guaranteed service. Vertical structuring serves to
& Miers, 2008). distinguish different levels of details, or abstrac-
tion levels.
In developing business process management
notation by the Object Management Group were
628
differentiated the following levels of process 5. Training,

modeling (White & Miers, 2008): 6. Framework for metrics,
7. Compliance, audit, and assessments,
• Process maps: Simple flow charts of the 8. Program execution.
activities; a flow diagrams without a lot of
details others than names of the activities In the framework of this chapter it is focused
and the broad decision conditions. on process based models of organization that helps
• Process descriptions: Provide more ex- organizations in transitioning to the advanced
tensive information on the process, such as and efficient BPM practices. Accordingly, it is
the roles of involved actors, the data, infor- pertinent to mention Process Maturity Model
mation and so on. (PMM) approach, which was built on success
• Process models: Detailed flow charts en- with software Capability Maturity Model (CMM).
compassing sufficient information such A modified version of PMM describing main
that the process is amenable to analysis and development phases is shown in Figure 3.
simulation. For the Level 1 is symptomatic that processes
are usually ad hoc and chaotic. At this stage, busi-
Process models can be created or presented ness processes need to be studied identified and
using many different methodologies. Basically, catalogued. Work unit management (Level 2)
each business process model is built for a different assumes that basic standards and tools in business
purpose. In this context can be identified many po- process management are adopted. Key attributes of
tential uses of process models (Browning, 2002): Process Maturity Level 3 include that all manage-
ment processes are well defined, established and
1. Program planning, controllable. At maturity Level 4, processes are
2. Baseline for continuous improvement, quantitatively managed and the non-conformances
3. Knowledge retention and learning, of process are known. According to Reitzig et al.
4. Process visualization, (2002) ‘A critical distinction between a defined
Figure 3. Process maturity model
629
process and a quantitatively managed process is • so called Integrated Processes (IP),

the predictability of the process performance’. • so called elementary processes (EP).
Organization at maturity Level 5 is expected to
document continuous improvement action plans Smaller objects than Elementary Processes
for both innovation and process improvement. are hierarchically (top to bottom) divided to two
Moreover, it is assumed to employ the continuous groups:
transfer of work experiences across the organiza-
tion (Kováč & Svač, 2007). • Complex Tasks (CT),
• Activities (A).
tHe cAse on busIness Substantial problem concerning originally

Process modelInG applied approach to process-oriented model of or-
ganization lied in practical impossibility to create
Due to the company’s focus on ISO 9000 series overall network of interaction between connected
standards company’s quality management team processes. It was due to a lack of unequivocal meth-
originally prepared first version of process-ori- odological rules. In further presented approach is
ented model of organization based on a gradual applied the hierarchical classification framework
decomposition, which is shown in figure 4. to built up process based organizational model
It copies a typical known approach to the cat- that will allow previously segregated business
egorization of business processes into operational, units to work in synthesis and enable business
support, and managerial processes. units to integrate processes with external enti-
In a proposed approach a hierarchical classi- ties and consolidates redundant processes across
fication framework for the systematic rebuilding functional areas. It closely corresponds to business
of business processes was used. The classifica- process integration approach (Ulrich, 2001), which
tion of processes is based on splitting up them to results in an enterprise that is more efficient and
three hierarchical levels, which are represented effective, which in turn drives up revenues and
by (Modrák & Manduľák, 2008): drives down costs. It was a ground of motivation
to develop the following method for business
• so called Unified Enterprise processes processes mapping and modeling that is based
(UEP), on a process decomposition resulting in a set of
Figure 4. Process-oriented model of organization
630
business structure models, which are represented Processes (UEP) from the original arrangement of
by diagrams in the order given (Modrák, 2005): processes. Subsequently, relations between them and
the environment of the enterprise are specified. The
• System diagram, environment is represented in a System diagram by
• Context diagrams, External Entities, with which the system communi-
• Commodity flow diagrams of the first and cates, while their content is not a subject of analysis in
second decomposition stage, the following steps. They usually represent the initial
• State transition diagrams of the first and source of commodity flows, or their end consumer.
second decomposition stage. In fact it represents the starting base of modelling
processes, from which other diagrams are derived
In the sense of the outlined procedure of reusing the principle of process decomposition. Based
designing business processes, the first step of this on the transformation of previous process-oriented
method is the creation of a System Diagram. Its model of organization (showed in figure 4) has been
purpose is to separate so-called Unified Enterprise created System diagram (see figure 5).
Figure 5. System diagram
631
Figure 6. Context diagram of the process A2

express relations only of the given UEP with its
environment. All surrounding elements of the
given UEP in Context diagram, irrespective of
whether they represent objects outside the en-
terprise or internal processes, are considered as
External entities. Supplier/customers rules might
be the same as for external as for internal subjects.
An example of the Context diagram of the process
A2 is depicted in figure 6.
Consecutively, Commodity flow diagrams of the
first stage are designed for A1, A2 and A3 processes,
which describe relations usually between integrated
processes. Two of them for the A2 and A3 processes
are shown on Figure 7. The purpose of the Commod-
ity flow diagrams is gradual decomposition of UEP,
up to the level of Elementary Processes. Commodity
flow diagrams of the second stage are constructed in
an analogous way as Commodity flow diagrams of
Subsequently are created the Context diagrams the first stage. It is the last stage of Commodity flow
for each Unified Enterprise Process depicted in diagrams because the Elementary Processes, which
a System diagram. Individual Context diagrams present the objects of modeling in this diagram’s
class, are considered to be the primitive processes.
Figure 7. (a) Commodity flow diagram (CFD) for the process A2; (b) CFD for the process A3
632
The objective of the state transition diagrams the level of Commodity flow diagrams of the first
(STDs) is the description of an internal structure stage, up to the level of state transition diagrams.
and mutual relations of tasks and/or activities of An example of the creation of the process map
Elementary Processes. State transition diagrams from two Commodity flow diagrams of the first
describe all of the states that an object can have, stage specified in the previous charts (showed in
the events under which an object changes state Figure 6) is depicted in Figure 9.
and the conditions that must be fulfilled before the Such maps are defining where the business
transition will occur. State Transition Diagrams process starts and ends and are identifying the
are basically used for describing the behavior of specific activities that need to be performed by the
individual objects. Hoverer, STDs have limited process owners. From those maps the managers
possibilities for describing the collaboration be- can easily define critical processes and analyze
tween objects that cause the transitions. This level them in detail. This approach evidently differs
of modeling is also pertinent for IS designers, from the method with which the company ac-
through which proper form modelling technique complished requirements of standard for quality
is selected. The most popular variety of STD in management system.
the programming area of OOD (Object Oriented
Design) is the Harel state chart (Harel, 1987).
Example of one STD drawn by QPR software in Future trends
this study is described in Figure 8.
In analogical way, “STDs of the first stage” According to Van der Aalst (2003) in BPM using
are sequentially decomposed to the level of State information technology dominated data-driven
Transition Diagrams of the second stage. approaches during the last decades. Consequently,
An important condition in the designing and the modeling of business processes was often ne-
subsequent modeling of process diagrams is to glected and existing process structures had been
maintain the consistency of inputs and outputs, so directly adapted to information technology. In the
that it is possible to create process maps, starting at present time, more and more company’s processes
Figure 8. STD of the first decomposition stage for the process A211
633
Figure 9. Process map fragment by the merging of two CFDs at the first level of decomposition.
are being conducted under the external supervisors ness methodology identifying nine different con-
of information systems that are driven by process cerns that anyone trying to change processes in an
models. However, the role of IS supervision does organization must consider. By Harmon (2007)
not aim at taking on the duties of company man- ‘software architects today would probably refer
agement alike as IS supervision can not guarantee to it as framework’. The combination of BPM
avoiding problems arising from organizational and service-oriented architecture (SOA) style is
changes. When spiking about process based IS, considered as effective methodological approach
one of the most important aspects is the selection for the future, especially for companies with ma-
of techniques for process architecture. White & jor ERP systems. This combined approach can
Miers (2008) states that functional decomposition bring greater flexibility than ERP systems since
should be definitely avoided and recommend integration of these tools with other enterprise
thinking of processes as a dynamic network of applications and business partner applications
interacting process instances. Rummler & Brache can be achieved easier.
(1990) introduced important framework of busi-
634
dIscussIon And conclusIon presumes that key managers are not only aware of
their role in the specification of requirements on
Competition and changes in the business envi- information system functionalities, but they have
ronment induce the need to constantly adapt and better understanding of organizational context of
improve the company’ business processes. The information system.
solution of this problem requires implementing
an effective business process modeler tool for
visualizing, understanding, analyzing, improv- reFerences
ing and documenting business processes. For the
beginning phase of BPM work it is sufficient to Aggarwal, R. (2004). Making BPM work.
model processes in the manner illustrated by the DM Review, 14(9), Retrieved October 1,
schemes in figure 5 or figure 8. However, software 2008 from http://www.dmreview.com/is-
is often considered the most important in success- sues/20040901/1009167-1.html
fully implementing BPM initiatives (Aggarwal, Browning, T. R. (2002). Process integration using
2004). For this purpose we implemented QPR the design structure matrix. Systems Engineering,
software that supports all key elements of busi- 5(3), 180–193. doi:10.1002/sys.10023
ness processes management - from modeling and
documentation, communicating, measurement and Castro, J., Kolp, M., & Mylopoulos, J. (2002).
analysis, to continuous process management and Towards requirements-driven information sys-
improvement. It served us to create comprehensive tems engineering: Rhe Tropos project. Informa-
business process models with accordance to the tion Systems, 6, 365–389. doi:10.1016/S0306-
presented approach illustrating the current status 4379(02)00012-1
of operations. Due to the fact that QPR software Chrissis, M., Konrad, M., & Shrum, S. (2003).
enables to generate process maps, equally as CMMI: Guidelines for process integration and
was shown in figure 8, it opens a further door to product improvement. Boston: Addison-Wesley.
modern business process management. Moreover,
this software tool can be integrated with other Davenport, T. H., & Short, J. E. (1990). The new
systems using a web-based portal that provides an industrial engineering: Information technology
effective means for communicating, monitoring and business process redesign. Sloan Manage-
and continuously improving business processes. ment Review, 11–27.
Nevertheless, for a successful BPM imple-
Deepak, V. (2007). Simplifying BPM implementa-
mentation, organizations might look beyond the
tions: Business process management. Executive
technology piece of the project and focus on the
Insights Gordian Transformation Partners Inc.
“human-side” of the project as well (Deepak,
Retrieved October 1, 2008 from http://www.
2007).
bptrends.com/resources_publications
Another important finding in this work has
been that enterprise process modeling; analysis and Franken, H. M., de Weger, M. K., & Jonkers, H.
implementation are at the core of business process (1997). Structural and quantitative perspectives
management and optimization of EIS designing. on business process modelling and analysis. In
The gradual transition from traditional hierarchical Proceedings of the 11th European Simulation
function based organization to the process-oriented Multiconference, Istanbul, June 1-4, 1997 (pp.
organization envisages a noticeable change in 595-599). Ghent, Belgium: Society for Computer
company manageress’s perception of business Simulation International.
process redesign in a scope of reengineering. It
635
Greasley, A. (2006). Using process mapping Modrák, V. (2007). Bridging organizational struc-
and business process simulation to support a ture and information system architecture through
process-based approach to change in a public process (LNCS 4537, pp. 445-455).
sector organization. Technovation, 26(1), 95–103.
Modrák, V., & Manduľák, J. (2008). LPH Vranov
doi:10.1016/j.technovation.2004.07.008
n/T-Case study. In L. Fischer (Ed.), BPM excel-
Harel, D. (1987). A visual formalism for complex lence in practice 2008 – Using BPM for competi-
systems. Science of Computer Programming, 8(3), tive advantage (pp. 51-62). Lighthouse Point, FL:
231–274. doi:10.1016/0167-6423(87)90035-9 Future Strategies.
Harmon, P. (2007). Business process change: Reitzig, R. W., Miller, J. B., West, D., & Kile, R.
A guide for business managers and BPM and L. (2003). Achieving capability maturity model
Six Sigma (2nd ed). San Francisco, CA: Morgan integration maturity level 2 using IBM Rational
Kaufmann Publishers Inc. software’s [White Paper]. Retrieved October
1, 2008 from http://www.cognence.com/pdfs/
Hess, T., & Brecht, L. (1995). State of the art des
CMMI_ProcessAndRequirementsManage-
business process redesign. Wiesbaden, Gabler-
ment_WhitePaper%20v1.0.pdf
Verlag.
Rummler, G., & Brache, A. (1990). Improving per-
Jonkers, H. (1997). The application of hybrid
formance: How to manage the white space on the
modeling techniques for business process per-
organization chart. San Francisco: Jossey-Bass.
formance analysis. In Kaylan & Lehmann (eds.),
Proceedings of the 11th European Simulation Scheer, A.-W. (1998). ARIS – Business process
Multi-conference, Istanbul, Turkey, 1-4 June (pp. frameworks (2nd ed.). Berlin.
779-786).
Scheer, A. W., & Nüttgens, M. (2000). ARIS
Kalnins, A., Kalnina, D., & Kalis, A. (1998) architecture and reference models for business
Comparison of tools and languages for business process management (LNCS 1806, pp. 301-304).
process reengineering. In Proceedings of the Third
Ulrich, M. (2001). Critical success factors in a
International Baltic Workshop on Databases and
business process integration initiative [Tactical
Information Systems, Riga, July, 24-28.
Strategy Group White Papers]. Retrieved October
Kováč, M., & Švač, V. (2007). Knowledge sharing 1, 2008 from http://www.ebizq.net/topics/ bpm/
in project Equal. In EQUAL for Automotive (pp. features/2602.html.
10-12). Technical University of Košice.
Van der Aalst, W. M. P. (2003). Business process
Lopez, H., Massacci, F., & Zannone, N. (2007). management: Past, present, future [BPTrends
Goal-equivalent secure business process re- White Papers]. Retrieved October 1, 2008 from
engineering for e-health. In J. Sztipanovits, et http://www.bptrends.com/resources_publications
al. (eds.), Workshop on model-based trustworthy
Vasconcelos, A., Sousa, P., & Tribolet, J. (2005).
health information systems. September 28 - Oc-
Information system architecture evaluation:
tober 3, Toulouse, France.
from software to enterprise level approaches.
Modrák, V. (2005). Business process improvement Paper presented at 12th European Conference
through optimisation of its structural properties. on Information Technology Evaluation (ECITE
In L. Fischer, (Ed.) Workflow handbook 2005 (pp. 2005), Turku, Finland.
75-90). Lighthouse Point, FL: Future Strategies.
636
White, S., & Miers, D. (2008). BPMN Modeling Zhuang, Q., Feng, J., & Bao, H. (2007). Mea-
and reference guide. Lighthouse Pt, FL: Future suring semantic gap: An information quantity
Strategies Inc. perspective. 5th IEEE International Conference
on Industrial Informatics, 2, 669-674.
637
638
Chapter 3.6
E-Markets as Meta-Enterprise
Information Systems
Martin Grieger
Accenture, Germany
Evi Hartmann
SMI Supply Management Institute, Germany
Herbert Kotzab
Copenhagen Business School, Denmark
IntroductIon systems such as supply chains (Grieger, 2004)

where borderless organizational structures occur
Electronic marketplaces are inter-organizational (see Picot, Reichwald, R., Wigand, 2001).
information sharing systems that enable virtual
business transactions and allow the exchange of
price and product information between buyers and bAckGround
sellers (Kollmann, 2000). For many decades, such
information systems were solely utilized for buying The term “electronic marketplace” is a widely
and selling of goods under “market conditions.” accepted name for information systems that cross-
With the emergence of advanced IT, academia, organizational boundaries, link one or more compa-
and also businesses have begun to use electronic nies to their customers and/or suppliers, and involve
marketplaces for information sharing system and resources shared between these organizations such
it is no longer only a medium for electronic buying as the exchange of information about prices and
and selling processes. In addition, electronic market- product offerings. A firm that utilizes an electronic
places are predicted to be a key-stone for integrating market place is known as the intermediary. This
the information flows within inter-organizational can be either one of the market participants such
as a buyer or a seller, an independent third party,
DOI: 10.4018/978-1-59904-885-7.ch063 or a multi-firm consortium (Bakos, 1991, p. 296).
E-Markets as Meta-Enterprise Information Systems
The application of the Internet technology tHe 6c busIness model

allowed further differentiations in this approach
and electronic marketplaces became Internet- Figure 1 illustrates the “6 Cs” business models
based electronic marketplaces (IEMPs). Such that can be applied to meta-enterprise information
marketplaces are also labeled as online exchanges, in order to characterize those systems.
e-hubs, electronic exchanges, online marketplaces, The “content” model covers the selection,
net-markets, e-auctions, or e-markets interchange- systematizing, and preparation of data according
ably (see Lipis, Villars, Byron, & Turner, 2002). to the user’s needs. It includes all information
Electronic markets can be virtual, public, necessary for the commerce transaction, for ex-
digital, neutral, or open (see Grieger, 2003). Our ample the electronic catalogue containing product
definition however shows that the phenomenon specifications, supplier data, and customer data
can be examined from a broader perspective as (e.g., credit history). Content is the foundation for
the following comparison between traditional context and commerce business models.
and internet-based electronic markets emphasizes The “context” business model copes with
(see Table 1). finding, organizing, and systematizing informa-
In an IEMP-setting, “buyers and sellers” be- tion sources. On the Internet, browser platforms
come “members” and/or large communities. The (also called “navigation systems”) such as Google,
markets can be located anywhere in the world, Lycos, or Altavista have adopted the “context”
and the new medium also brings new trading business model. By using the context business
partners together. Not only buying and selling models, firms can easily scan their environment
related information is exchanged, but also prior for all relevant information (Liu, 1998).
and later transaction phases such as bringing the The “commerce” model deals with enabling
trading partners together, negotiations, settlement, and supporting business transactions by facilitat-
and after sales are covered by the IEMP. And not ing the commercial frame. In most cases, com-
only trading transactions are offered, but also not merce models are supplemented by value-added
buying/selling related functions. services (see Essig & Arnold, 2001). eBay, Let’s
buy it, and Amazon are examples of firms that have
built their business around the commerce model.
Table 1. Comparison between traditional electronic marketplaces and IEMPs
Bakos’ (1991) electronic

IEMP-specific extensions/variations
market definition
Members
Large communities
Participants Buyers and sellers
Firms anywhere in the world
New trading partners
Focus on industry verticals or specific business functions
Buying and selling
Meeting and trading
Information exchange related to price
Functions Conducting transactions
and product offerings
Various mechanism (matching, auctioning, RFQ, exchange,
catalog aggregation)
Additional functions
Virtual place or forum
Other features available through Global
the Internet Public
Online, open 24 hr a day, 7 days a week
639
Figure 1. The six Cs business models for meta-

stage (extraction), through the end user, as well
enterprise information system
as the associated information flows” (Handfield
& Nichols 1999, p. 2).
The management of such chains requires an
accurate adoption of information technology (IT)
that encompasses a whole range of business pro-
cesses. Owing to the generally close connection
between information flows and material flows,
conflict situations in IT adoption can lead to great
losses as goods cannot flow without information
as Alshawi (2001) points out. Information-flow
related activities within supply chains refer to
information allowing, forbidding, directing physi-
cal flows, information checking and confirming,
The “community” models focus on bringing as well as information to provide proof and audit
people together with common interests in a virtual trails for taxes and billing (Alshawi 2001, p. 238).
place or forum. Shared interests have made it pos- We have chosen the supply chain management
sible to build communities that generate a certain (SCM) perspective that can be seen as a hybrid
level of trust among their respective participants organizational governance structure where in-
(Armstrong & Hagel, 1996, Figallo, 1998; Iacono dependent actors decide to collaborate based on
& Weisband, 1997) thus inspiring a productive relational norms and standards. This understand-
IEMP environment. ing of the SCM concept is illustrated in Figure 2,
We add a sixth “C” for the cost aspect of which depicts a simplified supply-chain network
IEMPs in the future. With this business model we structure, the information, and product flows, and
emphasize on the cost model for IEMP provider the key supply-chain business processes penetrat-
(e.g., cost of goods sold, operating expenses, sales ing functional silos within the company and the
and marketing costs, general and administrative) various corporate silos across the supply chain.
and how those costs are passed on to the IEMP Thus, business processes become supply-chain
customer using an underlying revenue model (see business processes linked across intra- and inter-
Amit & Zott 2000). company boundaries. We suggest here that the
information flow is managed by IEMP.
mAIn Focus oF tHe cHAPter

crItIcAl lIterAture revIeW
Electronic marketplaces on the Internet operate
24 hours a day, 7 days a week. This has certain In the late 1990s, numerous investigators began
consequences for the organization of the involved to study IEMPs and a large percentage of research
partners where Picot et al. (2001) would argue on IEMPs has been on the basis of conceptual
to identify a borderless organization where the papers. Examples of conceptual papers that focus
management of information plays a key role. on presenting different definitions, forms, and
One example of such a borderless organization characteristics of IEMPs are the contributions of
can be found in supply chains, which are defined as Bakos (1998), Grieger (2003), Archer and Gebauer
the sum of “all activities associated with the flow (2000), Kaplan and Sawhney, (2000), Wise and
and transformation of goods from raw materials Morrison (2000), or Korff (2001).
640
Figure 2. A general supply chain management model (Source: Cooper, Lambert, & Pagh 1997)
From a supply chain management point of artificial adaptive agents by using a “genetic
view, one can identify that the number of articles algorithm-based learning technique.”
on the topic IEMP is very rare (see Gibson, Hanna, Baldi and Borgman (2002) provide a short
& Menachof, 2001). Some contributions solely discussion on different theoretical perspectives
cover buying and selling aspects of IEMPs like on the study of automotive IEMPs (the consortia
auctioning and negotiating (see e.g., Emiliani & Cosivint and Volkswagen’s private market place)
Stec, 2002; Sashi & O’Leary, 2002; Wang, Wang, though their transaction cost economics approach
& Tai, 2002) or the use of electronic market places can be seen as rather superficial as they focus
for logistics integration (see Gudmundsson & on the cost aspect of the theory instead of incor-
Walczuk, 1999). porating other important transaction attributes
Wildemann (2002) has presented more exam- such as uncertainty, asset specificity, frequency,
ples of IEMPs application within the supply chain opportunism or bounded rationality (Williamson
area such as using electronic auctions, online bro- 1975, 1985).
kers, electronic catalogs, or electronic freight stock Especially the contribution of Malone, Yates,
exchanges for electronic procurement, electronic and Benjamin’s (1987)Electronic Markets and
logistics, and customer relationship management. Electronic Hierarchies, shows the importance of
Lee and Whang (2001) have presented a model transaction cost economics for IEMP where they
of an information hub—an approach to Internet- propose the electronic market hypothesis saying
based supply chain integration. The information that by reducing the costs of coordination, IT will
hub approach, which is more commonly known lead to an overall shift toward proportionately
as an e-hub, seems to be a key function that an more use of markets—rather than hierarchies—to
IEMP can provide from a supply chain manage- coordinate economic activity (Malone, Yates, &
ment perspective. Benjamin 1987, p. 484, 496).
Grewal, Comer, and Mehta (2001) investigated
the antecedents of organizational participation in
IEMPs theorizing that ability results from the in- IemP In scm busIness PrActIce
fluence of organizational learning and information
technology capabilities. Alternatively, Heezen and The coupling of IEMP with supply chains shows
Baets (1996) investigate IEMPs’ organizational the necessity to advance the IEMP business model
impact on firms. Oliver (1996) even focuses on from a pure buying-selling platform toward a meta-
641
enterprise information system that all involved transaction reductions can be realized and most
supply chain partners can use for information important efficient monitoring and management
exchange and knowledge sharing. The following of all processes.
business cases demonstrate this idea by a review
of how this business model is applied in five se- chemconnect and elemica:
lected IEMPs that serve different industries and IemPs for the chemical Industry
still exist after the e-market-hype disappeared.
However, consolidation in this market is still ChemConnect, founded 1995, is an independent
ongoing and IEMPs need to further improve their IEMP that aggregates supply and demand of
value-added service portfolio. This includes the chemical products (“match-making”). The goal
development of a sustainable business model that of this IEMP is to minimize the time and the costs
ensures their existence. of negotiating prices and conditions for chemical
Our analysis helps to validate the trend of raw materials, usually on spot-basis. ChemCon-
advancing the IEMP-business model towards a nect facilitates different price mechanisms that
meta-enterprise information sharing system. is (reverse-)auctions and an exchange floor.
More than 40 of the leading chemical companies
cc-hubwoo: A cross-Industrial IemP have invested in ChemConnect. In June 2001
ChemConnect added the Envera network, which
Cc-hubwoo was formed by the 2004 merger of the provides connectivity to exchange transactional
French company Hubwoo-Avisium and its German information about long-term contractual business.
competitor cc-chemplorer. In 2005, Trade-Ranger Furthermore, ChemConnect acquired CheMatch
in the U.S. was acquired. This independent IEMP in January 2002, its only online commodities
is the leading global provider for source-to-pay trading rival in the chemical industry, in a private
electronic solutions and supplier network manage- stock trade. The combined companies handled
ment. The company manages the largest business- $8.8 billion in transactions in 2002 (see “news”
to-business eProcurement community in the world at www.chemconnect.com).
with more than 60 buying corporations from all Elemica provides a neutral, independent, and
business sectors ranging from chemical to retail- open network for secure information sharing
ing, construction to aeronautics, and over 12,000 within pre-established chemical trading partner-
connected suppliers in 44 countries worldwide. ships (contracts are already negotiated off-line).
Cc-hubwoo delivers a full range of solutions The users can electronically connect to the Elemica
and services covering the whole procurement network via a single connection, using either estab-
process from sourcing to invoicing to supplier lished industry standards or a series of proprietary
enablement. The product portfolio allows large translations designed around their ERP systems
companies to automate purchase and procure- (www.elemica.com). This single connection lets
ment processes, connect with suppliers, access them share information with trading partners in the
customized electronic catalogues, and execute Elemica network in real time. Currently, Elemica’s
electronic transactions. Cc-hubwoo’s trading hub main function is to offer order processing capa-
processes 2 million purchase orders representing bility directly connected to the companies’ ERP
€5 billion in customer spend value annually. Cc- systems, or for businesses without ERP systems,
hubwoo enables companies to get transparency, Elemica offers the same services through a Web
consistency and control in their purchasing activi- browser. With the formation of Elemica in August
ties. Besides significant efficiency improvement 2000, the 22 founding companies felt they had
in the order-to-pay cycle, substantial cost per to transform the industry, rather than let industry
642
outsiders attempt to provide a “one-size-fits-all” equipment suppliers (currently 150), and ship own-
solution (www.elemica.com). ers in a comprehensive global network. According
to the tribon.com Web site, the core functionality
WWre: the IemP solution for of Tribon.com is a global database with more
the Grocery retail Industry than 70,000 components that can be accessed by
shipyards, agents and suppliers around the world.
In March 2000, 17 international retailers founded The database contains technical information (e.g.,
the Worldwide Retail Exchange (WWRE) to 3D models, design information, pictures, specifi-
enable participating retailers and suppliers to cations and commercial information provided by
simplify and automate supply chain processes, the suppliers who subscribe to the service). By
thereby eliminating inefficiencies in the supply subscribing to the service, shipyards can access,
chain. The founding members committed equity download and integrate information directly to
and provided the basis for the WWRE to be their design model (e.g., Tribon’s design software).
securely launched. Today, there are a total of 61 Tribon.com is supplemented by e-procurement
member companies worldwide that support the tools such as auctions (www.tribon.com).
IEMP (www.worldwideretailexchange.org).
The WWRE is governed by and for its mem- comparison of the examples
bers and is an independently managed, neutral from an scm Point of view
company. The WWRE enables retailers and sup-
pliers in the food, general merchandise, textile/ Table 2 illustrates the previously described char-
home, and drugstore sectors and aims to reduce acteristics of the five IEMPs—cc-hubwoo (across
costs across product development, electronic industries), ChemConnect (including Envera)
procurement, and other supply-chain processes and Elemica for the chemical industry, WWRE
(www.worldwideretailexchange.org). for the retail industry, and Tribon.com for the
maritime industry.
tribon.com: the IemP solution We also see how well the various SCM busi-
for the maritime Industry ness processes are integrated within the IEMP
solutions allowing meta-enterprise information
In October 2000, Tribon.com was launched as a exchange (see Table 3).
neutral IEMP for the maritime industry. It links The cases also show that there are different
together shipyards (currently 100), maritime approaches of the IEMPs to provide specific
Table 2. Classification of the IEMP-cases into the general IEMP characteristics
IEMP Characteristics
IEMP
cc-hubwoo X X X X X X X X x
ChemConnect x x X X X x
(Envera) x x x x X
Elemica x x x
WWRE x x x x x x x
Tribon.com x x x (x) (X) x
Legend: 1: vertical; 2: horizontal; 3: buy-side; 4: sell-side; 5: neutral; 6: private; 7: consortia; 8: independent; 9: fix; 10: variable; 11: without;
12: open; 13: closed; (x): fix and variable pricing is available at Tribon.com but not used by the industry
643
Table 3. SCM processes and their IEMP-integration
IEMP and SCM processes

Across Industry Retail industry Shipping industry
Chemical Industry IEMP
SCM-process IEMP IEMP IEMP
cc-hubwoo ChemConnect Elemica WWRE Tribon.com
Product flow x x x
Customer relationship
x x
management
Customer service
management
Demand management x x x
Order fulfillment x x (envera) x x
Manufacturing-flow
management
Supplier-service manage-
ment (including procure- x x x (x)
ment)
Product development and
x x
commercialization
Return management
services related to the SCM processes. Tribon. After the dot-com crash, the few surviving elec-
com, for example, is specialized in the product- tronic marketplaces redefined themselves as value
development process, the WWRE, on the other added service providers. They have focused more
hand, offers a broad choice of functions related on operational excellence, transitioning from IT
to different SCM processes. Elemica is special- shops to services-led business partners. Today, we
ized in the order-fulfilment process but currently can observe the third stage of digital exchanges.
broadens its portfolio. ChemConnect, on the Leading IEMPs, such as cc-hubwoo, are preparing
other hand, has extended its SCM-process range the way for true business process outsourcing. The
by acquiring other IEMPs—the electronic hub vision is to leverage cross-company, cross-industry
Envera, with similar functionality as Elemica, and strengths, and competencies to enable complete
Chematch, with its commodity-trading floor and outsourcing of non-strategic core services, e.g. the
cc-hubwoo’s product offering covers the entire entire procurement function. To make this happen,
source-to-pay process. common business processes need to be identified
and defined and a unified, Web-based architecture
needs to be developed to enable the process and
conclusIon And outlook system integration (Gass, 2005).
The chapter described the phenomenon of IEMP

as meta-enterprise information systems. IEMPs reFerences
went through an initial boom of investment and
expectations, and a few years of hard and pre- Alshawi, S. (2001). Logistics in the Internet age:
dicted consolidations. The first-generation IEMPs Towards a holistic information and processes pic-
focused on the idea that they enlarge the supplier ture. Logistics Information Management, 14(4),
base leading to product offerings at lower price. 235–242. doi:10.1108/EUM0000000005718
644
Amit, R., & Zott, C. (2000). Value drivers of e- Figallo, C. (1998). Hosting Web communities:
commerce business models. INSEAD Working Building relationships, increasing customer loy-
Paper 2000/06/ENT/SM, 6. alty, and maintaining a competitive edge. New
York: Wiley Computer Publishing.
Archer, N., & Gebauer, J. (2000, Jan. 19-21).
Managing in the context of the new electronic Gass, S. (2005). Reaching For the Real-Time En-
marketplace. In Proceedings of the 1st World terprise: Optimizing the Source-to-Pay Process.
Congress on the Management of Electronic Com- A Brief History of Timing. White Paper.
merce, Hamilton, Ontario, Canada, 2000.
Gibson, B. J., Hanna, J. B., & Menachof, D. A.
Armstrong, A., & Hagel, J. III. (1996). The real (2001, Sept. 13-14). An international analysis of
value of on-line communities. Harvard Business the value of logistics periodicals for teaching,
Review, 74(3), 134–141. research, and outreach purposes. In A. McKin-
non (Ed.), Proceedings of the Logistics Research
Bakos, Y. (1991). A strategic analysis of electronic
Network (LRN) Conference (pp. 149-159). Heriott-
marketplaces. MIS Quarterly, 15(3), 295–310.
Watt University, Edinburgh. London: Institute of
doi:10.2307/249641
Logistics and Transport.
Bakos, Y. (1998). The emerging role of electronic
Grewal, R., Comer, J. M., & Mehta, R. (2001). An
marketplaces on the Internet. Communications of the
investigation into the antecedents of organizational
ACM, 41(8), 35–42. doi:10.1145/280324.280330
participation in business-to-business electronic
Baldi, S., & Borgman, H. (2002). Ownership- markets. Journal of Marketing, 65(3), 17–33.
structures of electronic b2b marketplaces—A doi:10.1509/jmkg.65.3.17.18331
multi-perspective analysis. In H. Buhl, A. Huther
Grieger, M. (2003). Electronic marketplaces: A
& Reitwiesner (Eds.), Proceedings of the Informa-
literature review and a call for supply chain man-
tion Age Economy, Proceedings 5th International
agement research. European Journal of Opera-
Conference on Information Management (pp.
tional Research, 144(2), 280–294. doi:10.1016/
589-603). University of Augsburg.
S0377-2217(02)00394-6
Cooper, M. C., Lambert, D. M., & Pagh, J.
Grieger, M. (2004). Internet-based electronic
D. (1997). Supply chain management: More
marketplaces and supply chain management.
than a new name for logistics. International
Copenhagen: Samfundslitteratur.
Journal of Logistics Management, 8(1), 1–14.
doi:10.1108/09574099710805556 Grieger, M., & Kotzab, H. (2002a, June 13-14).
Supply chain management beyond electronic mar-
Emiliani, M. L., & Stec, D. J. (2002). Realizing
ketplaces—insights from the chemical industry. In
savings from online reverse auctions. Supply
Proceedings of the 14th International NOFOMA
Chain Management: An International Journal,
Conference. Trondheim, Norway.
7(1), 12–23. doi:10.1108/13598540210414355
Grieger, M., & Kotzab, H. (2002b, Oct. 23-27).
Essig, M., & Arnold, U. (2001). Electronic pro-
The use of electronic marketplaces for managing
curement in supply chain management: An infor-
supply chains. The Use of Electronic Marketplaces
mation economics-based analysis of electronic
for Managing Supply Chains. In Proceedings of
markets. Journal of Supply Chain Management,
the 5th International Conference on Electronic
37(4), 43–49. doi:10.1111/j.1745-493X.2001.
Commerce Research. Montreal, Canada.
tb00112.x
645
Gudmundsson, S. V., & Walczuk, R. (1999). The de- Lipis, L. J., Villars, R., Byron, D., & Turner, V.
velopment of electronic markets in logistics. Inter- (2002). Putting markets into place: An e-market-
national Journal of Logistics Management, 10(2), place definition and forecast IDC. Retrieved from
99–113. doi:10.1108/09574099910806021 http://www.idc.com
Handfield, R. B., & Nichols, E. L., Jr. (1999). Liu, S. (1998). Strategic scanning and interpreta-
Introduction to supply chain management. Upper tion revisiting: Foundations for a software agent
Saddle River, NJ: Prentice Hall. support system—Part 1: understanding the con-
cept and context of strategic scanning. Industrial
Heezen, J., & Baets, W. (1996). The impact of
Management & Data Systems, 98(7), 295–312.
electronic markets: The case of the Dutch Flower
doi:10.1108/02635579810241764
Auctions. The Journal of Strategic Information
Systems, 5(4), 317–333. doi:10.1016/S0963- Malone, T. W., Yates, J., & Benjamin, R. I. (1987).
8687(96)80032-1 Electronic markets and electronic hierarchies.
Henderson, D. R. (1984). Electronic marketing
doi:10.1145/214762.214766
in principle and practice. American Journal
of Agricultural Economics, 66(5), 848–853. Oliver, J. R. (1996). Market process reengineering
doi:10.2307/1241012 through electronic market systems: Opportunities
and challenges. Journal of Management Informa-
Iacono, S., & Weisband, S. (1997). Developing
tion Systems, 13(3), 113–126.
trust in virtual teams. The 30th Annual Hawaii
International Conference on System Sciences Picot, A., Reichwald, R., & Wigand, R. T. (2001).
(Vol. II, pp. 412-420). Die grenzenlose Unternehmung—Information,
Organisation, und Management (The borderless
Kaplan, S., & Sawhney, M. (2000). E-hubs—The
organization). (4 ed.) Wiesbaden: Gabler.
new B2B marketplaces. Harvard Business Review,
(May-June): 97–104. Sashi, C. M., & O’Leary, B. (2002). The role of
Internet auctions in the expansion of B2B mar-
Kauffman, R. J., & Dai, Q. (2000). Business
kets. Industrial Marketing Management, 31(2),
models for Internet based e-procurement systems
103–110. doi:10.1016/S0019-8501(01)00189-4
and b2b electronic markets: An exploratory assess-
ment. In Proceedings of the 34th Annual Hawaii Wang, K., Wang, E. T. G., & Tai, C. F. (2002). A
International Conference on System Sciences study of online auction sites in Taiwan: Product,
(HICSS-34). IEEE Computer Society. auction rule, and trading type. International Jour-
nal of Information Management, 22(2), 127–142.
Kollmann, T. (2000). Competitive strategies for
doi:10.1016/S0268-4012(01)00050-0
electronic marketplaces. Electronic Markets, 10(2),
102–109. doi:10.1080/10196780050138155 Wildemann, H. (2002). Supply chain management
mit e-Technologien (supply chain management
Lee, H. L., & Whang, S. (2001). E-business and
with e-technology). [Journal for Business Ad-
supply chain integration Stanford global supply
minsitration]. Zeitschrift für Betriebswirtschaft,
chain management forum. Retrieved from http.//
67(3), 1–20.
www.altosvc.com
Williamson, O. E. (1975). Markets and hierarchies:
Analysis and antritrust implications. New York:
The Free Press.
646
Williamson, O. E. (1985). The economic institu- mainly concerned with the sharing of information
tions of capitalism: Firms, markets, relational among members of the supply chain, for example,
contracting. New York: The Free Press. demand information, inventory status, capacity
plans, production schedules, promotion plans,
Wise, R., & Morrison, D. (2000). Beyond the
demand forecasts, and shipment schedules (infor-
exchange: The future of b2b. Harvard Business
mation integration). The second element refers to
Review, 78(6), 86–96.
tightening organizational relationships between
firms (organizational linkage), and the third ele-
ment refers to the redeployment of decision rights,
Web sItes used cc- work, and/or resources to the “best-positioned”
supply chain member (coordination).
Chemconnect: http://www.chemconnect.com Internet-Based Electronic Marketplace
(IEMP): Internet-based meta-enterprise informa-
Elemica: http://www.elemica.com/
tion system that allows participating firms (B-to-B)
Hubwoo: http://www.cc-hubwoo.com/ to interact and to accomplish one or more of the
following market-making activities/transactions:
Tribon.com. http://www.tribon.com
identifying potential trading partners, facilitating
Worldwide Retail Exchange. http://www.world- the exchange of information, goods, services, and
wideretailexchange.org payments and providing an infrastructure that
enables an efficient functioning of the market.
Meta-Enterprise Information System: An
information system that crosses organizational
boundaries and involves resources shared between
6C Business Model: A system-oriented view two or more enterprises.
or concept for analyzing meta-enterprise informa- Supply Chain Management: The integration
tion systems. The six Cs are abbreviations for six of key business processes from end user through
subsystems or business models: context, connec- original suppliers, which provides products, ser-
tion, content, commerce, community, and cost. vices, and information that add value for customers
Electronic Marketplace or E-Market: A type and other stakeholders.
of an inter-organizational information systems that Virtual Supply Chain: A supply chain (net-
enables virtual business transactions and allows work) that is connected through electronic links
the exchange of price and product information and characterized by its inherent flexibility to
between buyers and sellers. quickly adopt and adapt to changes in the business
Integrated Supply Chain/Supply Chain environment. Meaning that, new members can be
Integration: A “seamless supply chain” where continually added and old members deleted or
territorial boundaries between trading partners are have roles reassigned to them within the chain. As
eliminated allowing them to operate effectively a result, a virtual supply chain is a combination
as if they were part of one organization. Supply of a computer-based structure and a temporary
chain integration typically encompasses three organized supply chain network.
distinct, albeit related, elements: One element is
This work was previously published in Encyclopedia of Networked and Virtual Organizations, edited by Goran D. Putnik
and Maria Manuela Cruz-Cunha, pp. 475-483, copyright 2008 by Information Science Reference (an imprint of IGI Global).
647
648
Chapter 3.7
Enterprise Systems as an
Enabler of Fast-Paced Change:
The Case of Global B2B
Procurement in Ericsson
Oswaldo Lorenzo
Instituto de Empresa Business School, Spain
Angel Díaz
Instituto de Empresa Business School, Spain
AbstrAct light of the above, this investigation studies how

and why enterprise systems are able to support
This chapter studies the deployment of the SAP fast-paced changes on a global scale. In other
B2B (business-to-business) procurement appli- words, this chapter presents enterprise systems as
cation in Ericsson between 1999 and 2003, and flexible and responsive infrastructures that enable
argues that it enabled complex organizational organizational change.
change in a three-phase process: the implemen-
tation of said application in Spain; the evolution
of the application into a regional B2B procure- IntroductIon
ment platform; and its final transformation into
a global, pan-European B2B procurement unit. This chapter studies the deployment of the SAP
As described in the chapter, the enterprise sys- B2B (business-to-business) procurement appli-
tem allowed the company to flexibly support cation in Ericsson between 1999 and 2003, and
the majority of changes that took place during a argues that it enabled complex organizational
period of explosive growth of mobile phone sales change in a three-phase process. During this period
followed by an unexpected market downturn. In the telecommunication industry in Spain went
through a period of explosive growth of mobile
DOI: 10.4018/978-1-59904-531-3.ch008 phone sales followed by an unexpected market
Enterprise Systems as an Enabler of Fast-Paced Change
downturn. This triggered revolutionary changes and centralized e-procurement model, designed to
in company processes as they prepared first for reduce costs worldwide. This entailed outsourcing
growth, then for market deceleration, and finally services to a Nordic marketplace for the time-
for the market downturn. For three years Ericsson consuming activities of supplier activation, content
reported heavy operating losses and the company management, and integration with suppliers. This
undertook radical restructuring (e.g., offshoring, phase allowed the company to develop a network
outsourcing, and downsizing). During this period of global suppliers and to obtain greater visibility
the information system infrastructure, based on throughout the whole process.
SAP applications, served as a key enabler of these This study uses Pettigrew’s (1990, 1997)
changes. As described in the chapter, the enterprise framework to structure and analyze fieldwork
system allowed the company to flexibly support factors from the perspective of the Spanish sub-
the majority of actions throughout the growth sidiary. From this analysis emerges a model of
and downturn phases. In light of the above, this evolutionary implementation of B2B procurement
investigation studies how and why enterprise at different organizational stages (local, European,
systems are able to support fast-paced changes. and global). It is based on an in-depth examination
In other words, this chapter presents enterprise carried out from different angles, which include
systems as flexible and responsive infrastructures market and industry, local and global organization,
that enable organizational change. and inter-organizational relations with suppliers
The transformation process comprised three and technological partners, while taking into ac-
phases. Firstly, the Spanish subsidiary implement- count change processes occurring in parallel in
ed the system as an enabler for agility and control other areas of the organization. For each period
in the MRO (materials, repairs, and operations) of study, the research investigated how these
materials procurement process. This was a con- levels interact and how these forces influenced
sequence of the explosive growth of mobile sales the B2B procurement implementation process.
up to December 1999, when the legacy system Furthermore, the study includes a longitudinal
was no longer able to support the unexpected high analysis in order to understand how the results of
demand for materials and services from the sales previous stages influenced the subsequent stages.
and project areas, resulting in a lack of control The relevant success factors inferred by a
over MRO materials and low service levels. The comparison of these experiences are: enterprise
implementation of B2B procurement allowed the systems (ESs) and business alignment, the sup-
company to reduce the number of suppliers, the port of senior management, ES specialists, project
number of employees in the purchase area, and management, change agents, communication
reduce maverick spending1, as well as to develop strategies, organizational commitment, win-win
a process that was both flexible and controlled. relationship with suppliers, and information tech-
In the second phase, in response to the market nology (IT) compatibility.
downturn in 2000, the system was rolled out into Finally, the study links its results to previous
the Iberia Market Unit (Portugal and Spain) as an theory in order to explain the findings, and analyze
enabler of shared services for the purchase-to-pay said findings from the perspective of information
process. This centralization allowed the company system infrastructure and organization flexibility.
to further reduce the number of purchasing em- Based on Anthony and Turner’s (2000) dimensions
ployees throughout the area, to reduce costs of the of information systems flexibility (integration,
purchase-to-pay process by an average of 30%, modularity, and personnel), the study argues that
and to build up a network of regional suppliers. enterprise systems like SAP can be considered
In the last phase Ericsson implemented a global flexible infrastructures for organizations. This
649
finding contributes to the current debate on from that of the internal processes developed by
enterprise systems flexibility. In actual fact, the the company in response to these drivers.
study supports the rejection of the myth that en-
terprise systems render the business model rigid, external context: market drivers
as it shows how the enterprise system chosen by
Ericsson allowed the purchase organization to Between 1999 and 2000 the telecommunication
evolve quickly from a local-based organization, sector underwent huge growth. Two causes un-
to a regional and finally a global organization in derlie this phenomenon. First was the expansion
a highly dynamic environment. of new telecommunication technologies and
This research was based on the empirical services such as mobile phones, as companies in
evidence and then analyzed using Pettigrew’s the sector competed fiercely to gain market share
(1990, 1997) framework and linked to IT infra- for their mobile phones in different regions. As a
structure flexibility literature. The methodology consequence there were huge investments in the
used was that of in-depth single case study. This facilities and infrastructures required to support
methodology is recommended for the study of mobile networks. Second, this period coincides
“how” questions, and for the understanding of with the technological bubble and the “dot.com”
phenomena in early stages of research (Miles & enterprises market, a phenomenon that also af-
Huberman, 1994; Yin, 1994). Data were collected fected the telecommunications sector, given that
using a variety of techniques: unstructured inter- it provided many of the services demanded by
views, semi-structured interviews, and internal and the new economy.
external documentary review. This triangulation
permits cross-checking. The analysis consisted external context:
of data reduction, data display, and conclusion technology drivers
drawing (Yin, 1998; Miles & Huberman, 1994;
Eisenhardt, 1989). Even though the methodology Although ERP systems had been implemented
used in this study makes it difficult to generalize by the majority of large companies worldwide,
the findings and conclusions, we believe these e-procurement functionalities were initially very
findings can be helpful to practitioners and schol- limited. Hence many companies had decided at the
ars. For practitioners, the study presents ES as a time to develop their own custom-made applica-
flexible infrastructure to enable change. This goes tions, Ericsson’s head office in Sweden among
against common belief and introduces learning to them. Later however, many Internet-based appli-
practitioners involved in the deployment of ES. cations appeared on the market designed to give
For scholars, this study opens up debate vis-à-vis support in specific areas, including e-procurement,
traditional research that views ES as inflexible. to companies using ERP systems. Although a
Further research will doubtlessly enrich the debate. world leader in ERP systems, SAP arrived late
to the boom of Internet-based applications, but
then moved quickly to upgrade its programs
PHAse one: deAlInG and adapt them to a Web-enabled infrastructure.
WItH GroWtH When Ericsson Spain decided to purchase an ef-
fective e-procurement application rather than use
This section describes the first stage in the trans- that developed earlier by Ericsson’s head office,
formation of Ericsson, that of accelerated growth SAP was searching for companies that wanted to
in 1999 and 2000, from the perspective of the implement the SAP e-procurement functionality as
external drivers that generated the growth and a pilot test, in order to fine-tune the first version.
650
Internal context: corporate b2b Procurement

strategy and business model Implementation Process
Before 2000 Ericsson’s corporate strategy had The implementation process analysis was adapted
been one of a decentralized global company. from Cooper and Zmud (1990) and Lorenzo and
This meant that Ericsson España (EE) could plan, Diaz (2005). From the analysis of data emerged
execute, and control many of its activities on a an implementation process of six stages: (1) ini-
stand-alone basis. For example, EE had procure- tiation, (2) process design and gap analysis, (3)
ment contracts negotiated locally. At that moment implementation, (4) mutual adaptation of orga-
“each subsidiary concentrated on its own business, nization and technology, (5) supplier adaptation,
and it was unusual to collaborate among differ- and (6) establishment of a routine.
ent countries,” pointed out Rafael Berriochoa,
director of IS and sourcing manager. On the Initiation
other hand, the local IS department developed
and bought software independently, according The implementation of the B2B procurement be-
to its own needs, given that Ericsson’s business gan in 1999 when senior management identified
model permitted internal capabilities in each area the need for a better system for procurement of
or business process. Thus Ericsson owned facto- MRO materials. Explosive market growth and the
ries and R&D centers in different countries. The internalized service culture of “always arriving
company had also deployed an Excellence Center first” had encouraged the need for a faster and
(Ericsson Business Consulting), composed of 160 more flexible execution of business processes.
consultants in Spain. Although customers were satisfied with the high
level of quality and speed of Ericsson’s services,
Internal context: structure, the company had sacrificed efficiency and control
culture, and Processes of its back-office processes. Procurement of MRO
materials exhibited the following shortcomings:
As a consequence of its corporate strategy, EE
developed many core capabilities (e.g., manufac- • Widespread maverick spending:
turing, R&D, and network installation) locally, “Everyone was placing orders to every sup-
along with support processes (e.g., purchase, plier,” pointed out one of the interviewees.
administration, and systems). Creativity, inde- • Invoices without purchase orders.
pendence, and an entrepreneurial spirit were all • Lack of standardized processes: “There
part of the company culture. was a different process for almost every
EE implemented SAP R/3 between 1997 and purchase,” pointed out the project manager.
1998 through its Excellence Center, having been
encouraged by the company’s head office. This In light of the above, the senior management of
project meant having to embark on the standard- EE identified the need to change the MRO procure-
ization of processes and systems throughout the ment process in order to have more control, but
organization. After a period of time, the company without sacrificing the speed and agility required
had extended SAP into many areas or processes. to fulfill market growth. According to the sourcing
The implementation of SAP B2B procurement and business information manager, “The only way
formed part of this diffusion strategy. to work at this market speed was to implement a
B2B procurement application.” At the end of 1999,
EE began implementing SAP B2B procurement.
651
The aim was to design an efficient, controlled, and • IT capacity and plans for the future, also
flexible process. The company decided to choose based on examinations
SAP because of ensured integration between the
new functionality and the ERP system. For SAP Process design was performed using gap analy-
it was an exceptional opportunity to test and fine- sis of system and process requirements. Although
tune its B2B procurement application. some gaps resulted from this analysis, they were
eventually conciliated (see below, Mutual Ad-
Process Design and Gap Analysis aptation of Organization and Technology). This
exercise was used to establish the to-be process.
The purchase of MRO materials accounted for
24% of annual purchasing costs (about 100 mil- Implementation
lion euros). One of the main shortcomings of
the legacy process was the fact that orders were Implementation of the new functionality took three
placed through “free text,” which meant that the months, from December 1999 to March 2000.
procurement department had to rewrite them The implementation consisted of the creation of
one by one. At that moment the department staff catalogs, the design of workflows, system cus-
comprised 22 people. “Perhaps the purchasing tomization, pilot testing, and training. The project
process was more expensive than the cost of the team was composed of key users, Ericsson’s own
actual product,” the project manager commented. consultants, and the SAP consultants. For SAP it
The process requirements were: was essential to ensure the success of its first B2B
implementation in Spain.
• Simple and user friendly
• Decentralization of the order process to Mutual Adaptation of
employees Organization and Technology
• Use of catalogs with price and predefined
supplier As a result of the gap analysis, adjustments were
• Flexibility for managing the purchasing made to the organization and to the system, in-
of non-codified materials (e.g., marketing cluding:
campaigns)
• Workflows • Change in the procurement depart-
• Automatic conciliation of invoice and ment’s role: Purchasers were to spend
order more time on strategic tasks (e.g., negoti-
• Centralization of negotiation with suppliers ating with suppliers) than on transactional
purchasing tasks.
There was also an evaluation and selection • Decentralization of the order process to
of the suppliers that would continue delivering employees.
materials to the company. The key selection • Clear definition of order approval roles.
variables were: • New role of cataloging in the procure-
ment department: This role was respon-
• Price and quality of the delivered materials sible for the joint design and management
• ISO 9000 of catalogs with suppliers.
• Delivery capacity, based on an examina-
tion of processes at the supplier’s plant
652
Supplier Adaptation PHAse tWo: PrePArInG For

mArket decelerAtIon
The new MRO materials procurement process in
Ericsson brought important changes for many sup- This section describes the second stage in the
pliers. First, a number of suppliers were eliminated transformation process that took place in 2001,
from the Ericsson list. Ericsson España reduced when it became apparent that the bubble was go-
the number of suppliers from 4,000 to 160 in the ing to burst. In 2001 the company started to lose
first year. The group of remaining suppliers (many money for the first time in its history, and by the
of which were SMEs) had to introduce the use of year 2003 sales were half those of 2000.
e-mail accounts to exchange information with EE.
external context: market drivers
Establishment of a Routine
In 2001, the dot.com bubble burst, an event that
Once the implementation process was complete, can be attributed to unfulfilled expectations. It
the company and its suppliers began to interact also affected telecommunication companies that
through the new process and technology. The main had invested heavily in new licenses for the new
benefits achieved were as follows. third-generation services that were supposed to
be launched in 2001. Investments had to be cut
consequences dramatically in the face of new market realities.
These changes brought a number of benefits for external context:

EE and for its suppliers: technology drivers
• Personnel reduction—the number of pur- In 2001, although ERP suppliers offered a greater
chasers was reduced from 22 to 4 people. range of e-procurement functionalities, in-house
• Zero stocks of MRO materials, because custom-made applications were the most common
suppliers delivered directly to users within option (AMR Research, 2001). That same year
two days. SAP launched an upgraded and enhanced B2B
• Cost reduction—the cost of an order was procurement functionality called Enterprise Buyer
reduced from €140 to €14. Professional (EBP).
• Reduction of the number of suppliers from
4,000 (listed) to 160 (actives). Internal context: structure,
• 10% reduction in the cost of MRO culture, and Processes
materials.
• Standardized control and discipline Although Ericsson continued to pursue a global
processes. decentralized strategy, the company began to
• SAP programmed new functionalities for consolidate some of the support processes in
its B2B application to improve its products shared services. The aim was to create a more
and satisfy company requirements (e.g., adaptable and flexible business model that per-
free-text catalogs). mitted economies of scale and scope. Execution
• The remaining suppliers achieved high- of the abovementioned strategy entailed creating
er sales volumes and more standardized business support centers (BSCs) for the countries
relationships. in Western Europe, a decision that affected 17
653
subsidiaries. The BSCs used shared services for ad- in Spain. This reference process took into ac-
ministrative, financial, and purchasing processes. count minor requirements from other countries
The new business model brought significant like France and Holland and was redesigned to
changes in EE’s structure and culture. The admin- create a core model that became the standard for
istrative, financial, and purchasing processes were countries participating in BSC initiatives.
consolidated in two hubs for Western Europe. Sub-
sidiaries defined new roles and were responsible Implementation
for the internal customer relationship management.
This model also meant that EE developed a service The implementation of ep2p and the creation of
culture in back-office departments. the BSCs took six months in all. The ep2p project
Given that the expectations in the telecom consisted mainly of the migration of the former
market changed from an optimistic scenario to a functionality to the new version (EBP). In Spain
conservative one, Ericsson wanted to create a more a good communication campaign was required
flexible and adaptable business model that would to explain to users the need to migrate from one
be able to react to any external contextual change. version to another, even though the former ver-
A study by Cap Gemini Ernst & Young revealed sion had only recently been implemented. The
high operating costs due to a decentralized model project was led by the local purchase area and was
that duplicated resources in all subsidiaries. The supported by Cap Gemini consultants, who then
company subsequently developed the market unit rolled out the system to the other European BSCs.
solution whereby it unified applications in all the
subsidiaries of a market unit. The market unit solu- Mutual Adaptation of
tion was based on a former SAP implementation, Organization and Technology
and incorporated the implementation of the SAP
B2B functionality as the enabler of consolidated This initiative focused on the transactional ele-
purchase-to-pay processes. ments of the purchasing processes. For EE there
were no significant changes in the way the com-
the Implementation of the electronic pany carried out the purchasing process. The main
Purchase-to-Pay Process change was the introduction of a new role called
‘service manager’ in the subsidiaries, responsible
Initiation for the service management of the purchase-to-pay
process. Given that Spain was one of the places
The implementation of the SAP B2B functional- were a BSC was established, the purchasing
ity into the purchase-to-pay process was a Euro- personnel remained the same as before. From a
pean initiative. The project was called electronic technological perspective, this initiative consisted
purchase-to-pay, or just ep2p, and it became a key of merely migrating from one version to another,
enabler of the BSC initiative. One of the BSCs the second of which included a better materials
was established in Madrid to manage Portugal classification functionality.
and Spain. The upgraded functionality (EBP) was
used for this initiative. Supplier Adaptation
Process Design and Gap Analysis From a process and technological perspective,
there were no significant changes in the way
The modeling of the to-be purchase-to-pay process suppliers interacted with EE. However, the new
borrowed from the process already implemented initiatives did imply some major changes in or-
654
ganizational structure. Ericsson asked suppliers moment, telecom companies (e.g., Telefónica in
to take on the responsibility of the cataloging Spain) began to reduce their investments and to
process. Training programs were needed to achieve restructure their business models.
this objective.
external context:
Establishment of a Routine technology drivers
Once implementation was complete, the company From 2001, there was a significant increase in
and suppliers adapted quickly to the new processes. the offering of B2B procurement functionalities
and marketplace solutions. In 2002, the diffusion
consequences of packaged software for the B2B and electronic
marketplace reached 28% of the market, with
• For EE this initiative meant that it began SAP, Ariba, Commerce One, Oracle, and i2 (AMR
to lose independence in the design and ex- Research, 2001) as main suppliers. Important
ecution of local back-office processes. The electronic marketplace initiatives were adopted in
new consolidated purchase-to-pay process different sectors by firms like Covisint (automo-
allowed the corporation to have more vis- tive), Aero Exchange International (aerospace),
ibility of spending and more control over and ChemConnect (chemical).
the operation.
• This initiative allowed the company an ad- Internal context: structure,
ditional 30% process costs reduction. culture, and Processes
• Suppliers had to catalog all their materials
into the system. The telecom crisis resulted in huge losses for
• Suppliers developed internal cataloging companies. For example, between 1999 and 2002
capabilities. Ericsson’s operating losses totaled €2,000 mil-
• Some local suppliers began to develop lion. As a consequence the company had to make
the capacity to deliver in more than one some hard decisions, for which purpose a crisis
country. committee was set up. One of the first decisions
was to develop a global company, reducing local
PHAse tHree: coPInG WItH capabilities.
tHe mArket doWnturn In 2002, Ericsson carried out a total overhaul
of its business model. The company had to adapt
This section describes the third stage in the quickly to new market conditions. “Our new orga-
transformation process that took place after 2001 nization will put even more emphasis on serving
and entailed the consolidation of back-office our customers, reducing complexity, and applying
processes. our resources more efficiently,” said president
and CEO Kurt Hellstrom (Wireless Newsfactor,
external context: market drivers 2001). One of the main lines of action was to
group subsidiaries located in 140 countries into
Between 2002 and 2003, the telecom market 31 market units. Subsequently EE became part of
downturn decelerated. Recovery remained slow, the EMEA (Europe, Middle East, and Africa) unit.
however, because of the terrorist attacks in New This restructuring meant downsizing employees
York, which extended the economic crisis to other by 24% worldwide. Further measures included
sectors such as airlines and insurance. At that offshoring (e.g., closing down the factory in
655
Zamudio, Spain), process centralization (e.g., Implementation

procurement), and outsourcing (e.g., IS global
outsourcing to IBM). The rollout project lasted about four months in each
The main change in the structure was the cen- country. The project was supported by Accenture
tralization of back-office processes like IS, which and by the IBX specialists. The IBX specialists
meant that some IS activities (e.g., programming worked directly with suppliers to update catalogs.
and maintenance) were outsourced to IBM. The IBX was a marketplace created by three partners
same occurred with IT services, which were out- (Ericsson, SEB, and b-business) and was a key
sourced to HP. New roles in Ericsson were cre- agent in the implementation of the global B2B
ated for managing service-level agreements with procurement initiative. In Paulson’s words:
the outsourcing supplier. With regard to culture,
the company focused on creating the concept of We have two types of relations with IBX. The com-
global employees. pany is an e-procurement implementation partner.
In short, the company was involved simultane- When the system goes into operation, IBX becomes
ously in many projects that included centralization, a supplier that must (1) deliver operation services
outsourcing, offshoring, and downsizing, and all for application availability and for supplier and
of which were connected to one another. content management and (2) work continuously
on improvements.” (IBXNewsletter, 2003).
the Implementation of a
Global Purchasing Process
Mutual Adaptation of
Initiation Organization and Technology
Hence Ericsson decided to create a common For Ericsson España this new project brought
global purchasing process for MRO materials. The small changes in relation to the former purchasing
company opted for EBP (SAP’s B2B procurement processes. For example, one of the changes was
tool) for this global initiative. “We work strategi- the simplification of workflows. The centralization
cally to minimize the number of local, unique, and standardization meant that EE lost power over
homegrown solutions so that we can cut costs and local purchasing processes. Where technology was
groom the business support system landscape,” concerned, EE had to migrate to the new version
pointed out Anders Paulsson, purchasing director of SAP’s B2B procurement functionality and had
(IBX Newsletter, 2003). to connect to the IBX marketplace.
Process Design and Gap Analysis Supplier Adaptation
The design of the new process was based on a This new project brought the following changes
centralized structure. Some members of this proj- for suppliers: (1) use of the IBX marketplace to
ect team visited EE in order to learn from local interact with Ericsson, (2) catalog management
experience in terms of both implementation and was moved to the IBX marketplace, and (3) the
use of the system. The implementation of the new payment of fees for interacting with Ericsson
process only allowed small deviations from the through the IBX marketplace.
established design.
656
Establishment of a Routine Even though this conceptual model provides

an in-depth comprehension of the transformation
Once the project was completed, Ericsson España of relations with suppliers at EE, it is important
began to execute the purchasing process accord- to identify key findings in the transformation pro-
ingly. Users from EE recognized that resistance cess that could be applied to different enterprises.
to change was higher than in previous projects. Hence an additional analysis was performed to
identify the key success factors in the transforma-
consequences tion process.
These relevant factors, as presented in Figure
The change initiatives that Ericsson implemented 1, are:
worldwide resulted in a sharp reduction in the
number of employees (from 103,000 to 51,000). 1. Alignment
Although it is difficult to estimate how much each 2. Support of senior management
project had contributed to achieving the aim of 3. Project management
reducing operating costs, Ericsson’s management 4. Change agents
acknowledged that the B2B procurement project 5. Communication
worldwide played a pivotal role in this change 6. Commitment
process. As a consequence of the global B2B 7. Support to suppliers
procurement project, the company managed to 8. Technological compatibility
build a network of global suppliers that were able
to negotiate global contracts with local deliver- Alignment
ies (e.g., Sun, HP, IBM, and Corporate Express).
This centralization rendered the purchasing costs From the data analysis it can be inferred that the
of MRO materials worldwide completely visible. transformation process of MRO procurement at
On the other hand, EE lost control over its local EE, at every stage, was a key business project used
purchasing process. to support core business objectives. Hence the
project was well aligned with business objectives:
lessons leArned • During the first phase (1999-2000), the

transformation process was key for the
Four key concepts emerge from the analysis: con- improvement of efficiency and control of
text, transformation processes, processes in other the MRO procurement process. The tech-
areas, and consequences of the change process. nology implemented therefore acted as an
Contextual variables helped identify drivers for enabler of the change process.
the transformation of relations with suppliers. A • During the second stage (2001), the trans-
detailed description of the transformation process formation of MRO purchases at a European
helped to identify key factors, in particular the level was a key business project designed
process of adoption of the technological enabler. to consolidate the purchases of the coun-
Understanding processes realized at other levels of tries associated to each BSC. A key task
the organization served to understand how paral- for this European objective was the migra-
lel events influenced the transformation process. tion of processes and technologies in EE
Finally, the analysis identified the organizational towards the new process and the new tech-
consequences of the transformation process for nology suggested for the BSC, based on
both the company and suppliers. existing technology.
657
Figure 1. Relevant factors affecting the transformation process
• During the third stage (2002-2003), the change projects were led by senior management
transformation process of MRO purchases at EE:
at a global level was key for the conver-
sion of the company to a global centralized • During the first stage, senior purchasing
enterprise. A key task for this global ob- management at EE was the main sponsor
jective was the migration of processes and of the transformation process. The pur-
technologies in EE towards the new pro- chasing department assumed responsibil-
cess and the new technology suggested by ity for the change process at all times. This
head office. role included decision making on process
design and the selection of suppliers, proj-
support of senior management ect monitoring, and support for change
management.
As a consequence of the alignment of transfor- • During the second stage, sponsoring the
mation processes to business objectives, all three transformation process was the respon-
658
sibility of BSC and the local purchasing experience in activity consolidation proj-
units. Subsequently, close relations were ects in global organizations with complex
established between them. A joint monitor- structures. According to the country man-
ing group was made responsible for close- ager for Portugal, this firm also played a
ly controlling migration, change manage- key role in local change management and
ment, and communications. in facilitating good relations between BSC
• Sponsoring the transformation process on and local companies.
a global level was the responsibility of the • During the third stage, the transformation
CEO and of the local purchasing units. A process was supported by the consulting
detailed control was required of head of- firm Accenture and by the implementation
fice for on-time fulfillment of each project team of the IBX portal. Accenture guided
in the rollout. the rollout project in all countries, and IBX
guided the process of catalog updates and
specialists managing suppliers.
Due to the complexities of organizational and tech- Project management

nological change associated with each project, Er-
icsson’s management decided to recruit the support Each transformation process had a project direc-
of specialists who could bring industry-specific tor with a clear role in planning, deployment,
knowledge of business processes or of technology. and monitoring. Thus projects were carefully
These internal and external consultants brought structured. For example the first transformation
the resources and experience required to reach the process was directed by Cesar Barba, then an
objectives in the fastest time possible: internal consultant experienced in IT implemen-
tations and who had previously participated in
• During the first stage, the transforma- the company’s ERP implementation. The second
tion process had enjoyed the support of project was directed by Julio Gil, a member of
the internal consulting office of Ericsson. the Iberian (Spain and Portugal) BSC, and highly
This unit was created to offer consulting experienced in procurement and fluent in several
services to other companies and to sup- languages. For the last stage, a single project direc-
port Ericsson with internal implementation tor for the integration of MUS-EBP-HRMS was
that required specialized personnel. This is appointed. As the head office required very fast
considered a key success factor in terms of rollouts and that a defined standard be followed
available resources. The project also had to the letter, the leaders of these projects used
support from a team of SAP consultants an aggressive executive style in the fulfillment
who had participated in a similar project of objectives. The majority came from the sales
in France. For SAP the success of this first areas of Ericsson.
implementation of the B2B functionality in
Spain was critical for its market expansion change management
plans, and the system was customized by
a SAP partner with experience in this kind In each phase of the project, directors identified
of project. the involvement of users as a key success factor.
• During the second stage, the transforma- Hence key users became change agents, supporting
tion process was guided by consulting firm training and communication in the management
Cap Gemini Ernst & Young, which had of change:
659
• During the first stage, 10 key users were objective was to raise awareness among
designated as project champions and as employees of the economic circumstances
training facilitators. These key users were of the organization and of the importance
secretaries and those responsible for pur- of moving toward centralized processes
chasing in the projects department. and systems in order to reduce costs. In ad-
• During the second and third stages (2001- dition to electronic means of communica-
2003), the change agents (or key users) tion (e-mail, Web sites, and intranet), there
were also designated as project champions was direct communication with key users.
and training facilitators. For the project
leader it was very important to transmit to commitment
these champions the reasons and drivers
that made EE leave a local-sourcing model The commitment of all stakeholders to the trans-
for one of the back-office activities central- formation process was obtained by demonstrating
ized first at the BSC and subsequently at its benefits. The departments of procurement and
the head office in Sweden. finance participated proactively in the definition of
requirements. These were subsequently negotiated
communication with the SAP consultants, who in many cases had
to customize the system to meet organizational
Communication of project objectives and the rea- needs. For the department of systems and IT,
sons for embarking on the project played a pivotal standardizing procurement was critical, as was
role in change management throughout all three compatibility with the existing ERP system. Even
stages. This was achieved through clear intranet, if final users may have perceived the project as an
e-mail, and notice board messages: increase in their day-to-day responsibilities (and
hence may have been prone to resist implementa-
• During the first stage, the objective of the tion), the proposed transformation promised high
communication campaign was to explain service levels, which resulted in commitment.
to all employees the need for better control The explosion of the technological bubble, and
and efficiency in the procurement of MRO its impact on Ericsson, was clearly understood by
materials and services, and to expound the the employees, many of whom accepted early re-
basic characteristics of the technological tirement or negotiated leaves of absence, facilitat-
enabler. ing the commitment to changes that could enable
• During the second stage of transformation, the survival of their business. Comprehending the
the objective of the communication cam- reasons and drivers behind this change process
paign was to explain to all employees the brought commitment from key users and project
need for migration and adjustment of pro- participants. In addition to bonus schemes, there
curement to a new model that could con- was an implicit psychological contract with the
solidate purchases at a European level. It success of the project.
was considered very important that these
changes be related to the transformations support for suppliers
occurring in the marketplace.
• During the third stage, information was At each stage of the transformation process, EE
communicated in progressive fashion, giv- helped suppliers adapt to changes:
en that subsequent changes were to have a
profound impact on the organization. The
660
• During the first stage, the main support • During the second stage, the selection of
activity was training. This was basic and the new version of SAP’s B2B (EBP) to
only took half a day; and as technological support the purchase-to-payment process
requirements were few (just an e-mail ac- at BSC simplified the new transformation
count), the need for support was limited. process at EE, given that from a techno-
• During the second stage, the main change logical perspective the new project was
for suppliers was the assignment of re- merely a question of migrating to another
sponsibility for the catalog, which had to version of the technology.
include all items purchased. EE supported • During the third stage, the selection of a
its suppliers at the beginning of this pro- new version of EBP to support a central-
cess through training and a support group. ized procurement process at the head office
• During the third stage, the main change for level facilitated the third transformation
suppliers was their integration into the IBX process at EE, as again, from a techno-
marketplace. EE arranged for head office to logical perspective, the new project was
suspend connection fee charges to the mar- merely a question of migrating to another
ketplace portal for a one-year period while version of the technology.
suppliers learned to appreciate the advan-
tages of this new means of communica- dIscussIon on tHe FlexIbIlIty
tion. Also, to reduce language barriers, EE oF enterPrIse systems
supported connection to Spanish suppliers
through a Spanish portal called Opciona, This study links its results to previous theory in
which had alliances with IBX. Opciona order to explain the findings. Hence the findings
and IBX worked closely with suppliers to were analyzed from the perspective of information
streamline the process of updating catalogs. system infrastructure and organization flexibility.
Anthony and Turner’s (2000) study took the first
technological compatibility step toward creating a valid IT infrastructure
flexibility construct. Their findings reveal that
Another relevant factor for each transformation IT infrastructure flexibility can be expressed in
process was the selection and implementation of three factors: two related to technical issues and
technologies that were compatible with or easily one to human issues. They labeled these factors
integrated into existing ones. Ericsson preferred as “integration,” “modularity,” and “IT personnel
to maintain a single-system platform to simplify flexibility,” respectively. The integration factor is
technology management and to ease users’ learn- a merger of the dimensions of IT connectivity and
ing curve: IT compatibility. Connectivity is the ability of any
technology component to attach to any other com-
• During the first stage, the selection of SAP ponent inside or outside the company. Compatibility
for the B2B functionality facilitated the is the ability to share any type of information with
integration of the new MRO procurement any technology component. This factor suggests
process into the existing ERP application. that transparent access into all platforms contributes
It also eliminated the need for middleware, to IT flexibility. The modularity factor is a merger
aided technology management, and meant of the dimensions of application functionality and
that the company could use the learning database transparency. Modularity is the ability to
curve of the previous SAP implementation add, modify, and remove any component of the
as a base. infrastructure with ease. IT personnel flexibility
661
is related to the depth and breadth of four types systems flexibility. In fact, this study supports
of knowledge and skills: (1) technology manage- the refutation of the myth that enterprise systems
ment knowledge and skills, (2) business functional make for a rigid business model. The study shows
knowledge and skills, (3) interpersonal and man- how the enterprise system chosen by Ericsson
agement skills, and (4) technical knowledge and permitted the purchase organization to evolve
skills. This implies that technical skills alone are quickly from a local-based organization to first
not enough to implement and use IT. a regional and then to a global organization in a
If one evaluates EE’s B2B implementation highly dynamic environment.
experience using Anthony and Turner’s frame-
work, one can argue that enterprise systems such
as SAP are flexible information technologies. The conclusIon
three factors mentioned by Anthony and Turner
are present in the case of EE. These three factors This chapter provides a detailed analysis of
allowed SAP to evolve flexibly according to the organizational transformation of Ericsson
Ericsson’s needs and challenges. First, the inte- in Europe, from the perspective of the Spanish
gration factor allowed the SAP B2B procurement subsidiary. It focuses on how said organizational
functionality to connect smoothly to the SAP transformation was enabled by enterprise systems.
back-office functionalities, such as the financial Specifically the chapter describes this process as
and materials management modules. Addition- taking place in three stages: the implementation
ally, the integration factor allowed the SAP B2B of a B2B procurement application in Spain; the
procurement functionality to share information evolution of this application into a regional B2B
with other organizational departments. procurement platform; and its final transformation
Second, the modularity factor allowed the into a global, pan-European B2B procurement unit.
SAP B2B procurement functionality to add new We argue that this successful and rapid process,
versions to previous implementations with ease. driven by drastic market changes, was partially
In fact, users considered that all migrations were enabled by a B2B procurement enterprise system.
transparent processes. Furthermore, the modular- This process in turn leads us to examine key
ity factor allowed the SAP B2B functionality to success factors in ES-supported global organiza-
modify previous configurations with no problems. tion transformation processes. These factors are:
Finally, the IT personnel flexibility factor allowed alignment with business objectives, support of
the SAP B2B functionality to be implemented senior management, project management, change
and diffused quickly throughout the corporation. agents, communication, commitment, support
Enterprise systems from leading suppliers like given to suppliers, and technological compatibil-
SAP are common tools used by many technical ity. Finally, this analysis suggests that enterprise
specialists and functional users worldwide. There systems can be considered enablers of evolution-
is therefore a reliable supply of SAP knowledge ary change, and that they are more flexible than
around the world that allows companies to rollout managers realize.
the system quickly. Although the study generated concepts and
One can use Anthony and Turner’s (2000) relations shown by the transformation model for
dimensions of information systems flexibility relations with suppliers (Figures 2, 3, and 4), they
(integration, modularity, and personnel) to argue are not generalizable to similar processes in other
that enterprise systems like SAP are flexible companies. Further cases and cross-analysis with
infrastructures for organizations. This finding the current results are required to build a more
contributes to the current debate of enterprise robust model.
662
Figure 2. Phase one: Managing growth
Figure 3. Phase two: Preparing for market deceleration
Future reseArcH dIrectIons In particular, further research should attempt

to find more evidence that helps support the
A future research agenda should include the em- claim that enterprise systems can provide flexible
pirical validation and elaboration of global B2B information infrastructures for businesses. This
procurement adoption processes in other settings. could include searching for evidence in traditional
The theoretical models were generated by ana- enterprise system infrastructures (e.g., SAP R/3
lyzing only one enterprise, albeit in depth. More or MySAP) coupled with a further search for evi-
empirical comparisons will enrich the constructs dence in new service-oriented architectures (e.g.,
developed here and yield more understanding of enterprise service architecture of SAP). It could
the global B2B procurement phenomenon. also be interesting to investigate if the customers’
663
Figure 4. Phase three: Coping with the market downturn
Figure 5. ES flexibility at Ericsson
perceptions of enterprise systems as inflexible all players in this supply chain seemed to help
tools have varied between the implementation of Ericsson achieve its objective of changing how
a traditional platform and the implementation of it manages relations with its providers. Several
service-oriented infrastructure. questions emerge, including: What level of sup-
With regard to the key factors affecting the port should a company give to its supplier when
transformation process at different levels in a implementing a B2B procurement system? What
global company, it may be interesting to deepen the factors affect the decision pertaining to the level
understanding of how some of these factors work. of support afforded to suppliers?
For example, support for suppliers to implement Two initial strategies for further research can
a B2B procurement may vary depending on the therefore be proposed. The first strategy, aimed at
internal and external contextual characteristics of theory building, is to compare emergent concepts
the client company. For example, in the case of and constructs with extant literature (Eisenhardt,
Ericsson, the strong position of the client com- 1989). The second strategy, designed to develop
pany in an unfavorable business environment for generalizable theory, is to study a reasonable
664
number of additional cases (e.g., between 4 and Duncan, N. B. (1995). Capturing flexibility of IT
10, according to Eisenhardt, 1989). Finally, further infrastructure: A study of resource characteristics
research can be carried out to test hypotheses that and their measure. Journal of Management Infor-
can emerge from the previous qualitative studies. mation Systems, 12(2), 37–57.
Eisenhardt, K. (1989). Building theories from case
study research. Academy of Management Review,
AcknoWledGment
14(4), 532–550. doi:10.2307/258557
We would like to thank Ericsson España for al- Fan, M., Stallaert, J., & Whinston, A. (2000). The
lowing us to carry out this study. Special thanks adoption and design methodologies of component-
to Rafael Berriochoa (IS and sourcing manager) based enterprise systems. European Journal of
and Cesar Barba (internal consultant and project Information Systems, 9, 25–35. doi:10.1057/
leader). palgrave/ejis/3000343
Please note: This chapter is based on a prelimi-
Lorenzo, O., & Díaz, A. (2005). Process gap analy-
nary study published in the conference proceedings
sis and modelling in enterprise systems. Interna-
of the 2006 European Operations Management
tional Journal of Simulation and Process Model-
Association (EUROMA) conference. It differs
ling, 1(3/4). doi:10.1504/IJSPM.2005.007642
from the original conference proceedings, how-
ever, in two aspects: (1) this chapter extends the Miles, M., & Huberman, A. (1994). Qualitative
analysis to European and global experiences, data analysis, an extended sourcebook (2nd ed.).
and (2) it includes an in-depth discussion of the Thousand Oaks, CA: Sage.
lessons learned through these three experiences
IBX Newsletter. (2003). Global e-procurement a
(local, European, and global).
winner for Ericsson. IBX Newsletter, (3).
Pettigrew, A. (1990). Longitudinal field research
reFerences on change: Theory and practice. Organization
Science, 1(3), 267–292. doi:10.1287/orsc.1.3.267
Allen, B. R., & Boynton, A. (1991). Information
architecture: In search of efficient flexibility. MIS Pettigrew, A. (1997). What is a processual analy-
Quarterly, 15(4), 435–445. doi:10.2307/249447 sis? Scandinavian Journal of Management, 13(4),
337–348. doi:10.1016/S0956-5221(97)00020-1
Anthony, T., & Turner, D. (2000). Measuring the
Flexibility of Information Technology infrastruc- AMR Research. (2001). Application spending
ture: Exploratory analysis of a construct. Journal and penetration report 2002-2003.
of Management Information Systems, 17(1). Wireless Newsfactor. (2001). Ericsson addresses
Cooper, R., & Zmud, R. (1990). Information tech- globalization with sweeping restructuring plan.
nology implementation research: A technological Wireless Newsfactor.
diffusion approach. Management Science, 36(2), Yin, R. (1994). Case study research. Design and
123–139. doi:10.1287/mnsc.36.2.123 methods. Thousand Oaks, CA: Sage.
Davenport, T. (2000). Mission critical: Realizing
the promise of enterprise systems. Boston: Harvard
Business School Press.
665
AddItIonAl reAdInG Johnson, M., & Whang, S. (2002). E-business

and supply chain management: An overview and
Ash, C., & Burn, J. (2003). Assessing the benefits framework. Production and Operations Manage-
from e-business transformation through effective ment, 11(4).
enterprise management. European Journal of
Information Systems, 12, 297–308. doi:10.1057/ Kimberly, J. (1981). Managerial innovation. In
palgrave.ejis.3000476 P. Nystrom & W. Starbuck (Eds.), Handbook of
organizational design. Volume 1: Adapting orga-
Bhattacherjee, A. (2000). Beginning SAP R/3 nizations of their environments. Oxford: Oxford
implementation at Geneva Pharmaceuticals. Com- University Press.
munications of the AIS, 4, Article 2.
Laseter, T. (2003). Whirlpool Corporation global
Boyler, K., & Olson, J. (2002). Drivers of Internet procurement. Case UVA-OM-1071, Darden Busi-
purchasing success. Production and Operations ness Publishing, University of Virginia, USA.
Management, 11(4).
Lee, H., & Whang, S. (2001). E-business and
Christopher, M. (1992). Logistics and supply chain supply chain integration. Proceedings of the Stan-
management: Strategies for reducing costs and ford Global Supply Chain Management Forum
improving services. London: Pitman. (SGSCMF-W2-2001).
Cowan, E., & Eder, L. (2003). The transformation Leonard-Barton, D. (1988). Implementation as
of AT&T’s enterprise network systems group to mutual adaptation of technology and organization.
Avaya: Enabling the virtual corporation through Research Policy, 251–267. doi:10.1016/0048-
reengineering and enterprise resource planning. 7333(88)90006-6
Journal of Information Systems Education, 14(3).
Lorenzo, O., Kawalek, P., & Wood-Harper, T.
Diaz, A. (2000). E-business: From demand net- (2005). Embedding the enterprise systems into
works to techno logistics. Supply Chain Forum— . the enterprise: A model of corporate diffusion.
International Journal (Toronto, Ont.), 1. Communications of the AIS, 15, 609–641.
Ettlie, J. (1992). Organization integration and Magnusson, M. (2004). Managing the knowledge
process innovation. Academy of Management landscape of an MNC: Knowledge networking at
Journal, 15(4), 795–827. doi:10.2307/256316 Ericsson. Knowledge and Process Management,
Fraser, P., & Klassen, R. (2005). E-procurement. 11(4). doi:10.1002/kpm.210
MIT Sloan Management Review, 46(2). Malone, T., & Crowston, K. (1994). The interdisci-
Frohlich, M. (2002). E-integration in the sup- plinary study of coordination. ACM Computing Sur-
ply chain: Barriers and performance. Decision veys, 26(1), 87–119. doi:10.1145/174666.174668
Sciences, 33(4). doi:10.1111/j.1540-5915.2002. Malone, T. W., Crowston, K., Lee, J., Pentland,
tb01655.x B., O’Donnell, E., & Dellarocas, C. (1999). Tools
Galliers, R. (1992). Choosing information sys- for inventing organizations: Toward a handbook
tems research approaches. In R. Galliers (Ed.), of organizational processes. Management Science,
Information systems research. Oxford: Blackwell. 45(3). doi:10.1287/mnsc.45.3.425
Ives, B., Jarvenpaa, S., & Mason, R. (1993). Global Markus, M. (2000). Paradigm shifts—e-business
business drivers: Aligning information technology and business/systems integration. Communica-
to business strategy. IBM Systems Journal, 32(1). tions of the AIS, 1(10).
666
Markus, M., Tanis, C., & Fenema, P. (2000). Multi- Taylor, D. (1997). Global cases in logistics and
site ERP implementations. Communications of the supply chain management. Thomson Press.
ACM, 43(4), 42–46. doi:10.1145/332051.332068
Tyre, M., & Orlikowski, W. (1994). Windows of
Segev, A., & Gebaur, J. (2001). B2B procure- opportunity: Temporal patterns of technological
ment and marketplace transformation. Infor- adaptation in organizations. Organization Science,
mation Technology and Management, 2(3). 5(1), 98–117. doi:10.1287/orsc.5.1.98
doi:10.1023/A:1011442008251
Westerman, G., & Cotteleer, M. (1999). Tektronix,
Smith, H. L., Dickson, K., & Smith, S. L. (1991). Inc: Global ERP implementation. Boston: Harvard
There are two sides to every story: Innovation Business School Publishing (Case 9-699-043).
and collaboration within networks of large and
small firms. Research Policy, 20(5), 457–469.
doi:10.1016/0048-7333(91)90069-3
endnote
Soh, C., Siew Kien, S., & Tay-Yap, J. (2000).
Cultural fits and misfits: Is ERP a universal solu-
1
Maverick spending refers to purchases made
tion? Communications of the ACM, 43(4), 47–51. outside established processes for ordering
doi:10.1145/332051.332070 goods and services.
Stefanou, C. (1999). Supply chain management

and organizational key factors for successful
implementation of ERP systems. Proceedings
of the 5th Americas Conference on Information
Systems, Milwaukee, WI.
of IGI Global).
667
668
Chapter 3.8
Extending Enterprise
Application Integration
(EAI) With Mobile and Web
Services Technologies
Abbass Ghanbary
Bhuvan Unhelkar
AbstrAct collaborative business requires us to consider the

business paradigm in terms of policies-processes-
Web Services (WS) technologies, generally built standards. This chapter, based on experimental
around the ubiquitous Extensible Markup Lan- research carried out by the authors, demonstrates
guage (XML), have provided many opportunities how the technologies of WS open up the doors to
for integrating enterprise applications. However, collaborative Enterprise Architecture Integration
XML/Simple Object Access Protocol (SOAP), (EAI) and Service Oriented Architecture (SOA)
together with Web Services Definition Language resulting in Business Integration (BI). The chapter
(WSDL) and Universal Description Discovery and also provide a quantitative investigation based on
Integration (UDDI), form a comprehensive suite of organization’s adaptation to mobile and Web Ser-
WS technologies that have the potential to transcend vices technologies.
beyond mere application integration within an orga-
nization, and to provide capabilities of integrating
processes across multiple organizations. Currently, INtrODUctION
the WS paradigm is driven through parameters
however; the paradigm shift that can result in true This chapter describes how WS can be used in
order to align and integrate business processes of
DOI: 10.4018/978-1-60566-330-2.ch013 organizations (internal and external processes) to
Extending Enterprise Application Integration (EAI) With Mobile and Web Services Technologies
satisfy the needs of Enterprise Architecture (EA). the needs of business, technology, and data systems
Thus far, the concept of Business Integration (http://www.integrationconsortium.org).
(BI) has been mainly focused on integrating the Based on Chase (2006), originally designed
business processes internal to an organization; as a way to develop the technology architecture
however this chapter is an investigation to identify for an organization, TOGAF has evolved into a
how the organizations can extend this integration methodology for analysing the overall business
with those business processes belonging to other architecture. The first part of TOGAF is a meth-
enterprises and how they adapt mobile and Web odology for developing the architecture design,
Services technologies in order to integrate with which is called the Architecture Development
those business processes. Method (ADM). It has the following nine basic
According to Finkelsteing (2006) Enterprise phases:
Architecture (EA) builds on business knowledge
and allows business specialist experts to apply • Preliminary phase: Framework and
their respective knowledge to determine the most principles. Get everyone on board with
effective technology and process solutions for the plan.
the business. • Phase A: Architecture vision. Define
Information and Communication Technology your scope and vision and map your over-
(ICT) architectures have not paid enough attention all strategy.
to integration of the services in the past. Service • Phase B: Business architecture. Describe
Oriented Architecture (SOA) is an architecture your current and target business archi-
that makes the services of a system to interact tectures and determine the gap between
and perform a task supporting a request. SOA them.
is classified as sub-architecture of Enterprise • Phase C: Information system architec-
Architecture. tures. Develop target architectures for
Based on Barry (2003), a Service Oriented your data and applications.
Architecture (SOA) is a part of an EA and can • Phase D: Technology architecture. Create
be viewed as “sub-architecture” of an Enterprise the overall target architecture that you will
Architecture. SOA existed before the advent of implement in future phases.
Web Services. Technologies such as Common • Phase E: Opportunities and solutions.
Object Request Broker (CORBA) and Distributed Develop the overall strategy, determining
Component Object Model (DCOM) afforded what you will buy, build or reuse, and how
the opportunity to create SOA. Web Services is you will implement the architecture de-
ideal technology for developing sophisticated scribed in phase D.
architecture. • Phase F: Migration planning. Prioritize
The Open Group Architecture Framework projects and develop the migration plan.
(TOGAF) is a critical architecture for the effective • Phase G: Implementation governance.
and safe construction of business and information Determine how you will provide oversight
systems. TOGAF provides the TOGAF Architec- to the implementation.
ture Development Method (ADM). TOGAF ADM • Phase H: Architecture change manage-
is a comprehensive, detailed, industry standard ment. Monitor the running system for nec-
method for developing Enterprise Architectures essary changes and determine whether to
Integration (EAI), and related information, appli- start a new cycle, looping back to the pre-
cation, and technology architectures that address liminary phase.
669
These phases provide a standardised way of LItrAtUrE rEVIEW

analysing the enterprise and planning and man-
aging the actual implementation. The Service The increase in the demand of the management
Oriented Architecture is considered in Phase of the Information and Communication Technol-
D: Technology architecture where the TOGAF ogy (ICT) has caused the research to focus their
defines the services and their relationship with efforts on integrating of business processes and
each other and define how the services could be data. The term Enterprise Integration (or System
invoked by different requesters. Integration) reflects the capability to integrate a
Service Oriented Architecture (SOA) describes variety of different system functionalities.
how the service could be invoked and how the Traditionally, information systems were
service attributes are implemented. The concepts implemented to support specific functional areas.
of SOA and TOGAF relate to each other when However, the advancement of information tech-
Technology Architecture is invoked by different nology enables new forms of organizations and
requesters. TOGAF contains two reference models facilitates their business processes to collaborate
that can be used in this way: a platform-centric even when these organizations are not necessarily
Technical Reference Model that focuses on the known to each other. As organizations become
services and structure of the underlying platform more complex and diverse in the collaborative
necessary to support the use and reuse of applica- context, it becomes nearly impossible for them to
tions, and an Integrated Information Infrastructure implement their collaborative business concepts
Reference Model that focuses on the applications without enterprise integration.
space, and addresses the need for interoperability, New technology seems to suggest that mo-
and for enabling secure flow of information where bile services will be the greatest opportunity for
and when it is needed (http://www.ebizq.net). businesses to develop richer and more profitable
The highly competitive nature of the current relationships with individual customers by giv-
business environment creates tremendous pressure ing them what they actually want (Falcone and
for organizations to collaborate. It is essential Garito, 2006)
for companies to understand rapidly changing According to Jostad, et al, (2005) the demand
business circumstances. The rapidly changing for flexible, efficient and user-friendly collabora-
environment encourages the enterprises to inte- tive services is becoming more and more urgent
grate their business functions into a system that as competition in the current market oriented
efficiently utilises ICT. arena is becoming more intense. Enterprises have
The chapter proposes a theoretical model as to be more dynamic in terms of collaboration
the recommended implementation of the integra- with partners and even competitors. The Service
tion that requires a substantial amount of time Oriented Architecture is a promising computing
and financial commitment. The supplemental paradigm offering solutions that are extendible,
technologies of SOA, EAI and TOGAF automate flexible and compatible with legacy systems.
the integration process with the collaborative en- This chapter proposes and investigates the use of
vironment of the business processes of multiple SOA in the construction of collaborations across
organizations. multiple organizations.
Harrison and Taylor (2005) define an SOA that
builds on the concept of a service. It is a collection
of services capable of interacting in three ways,
commonly referred to as ‘publish, find and bind’.
In other words, a service must be able to make
670
its interface available to other services (publish), • The Platform-Specific Model (PSM) adds
other services must be capable of discovering to the PIM the details of a specific comput-
the interface (find), and finally services must ing platform on which the business solu-
be able to connect to one another to exchange tion will be deployed.
messages (bind). The loose coupling of an SOA • Transformations (mappings) are performed
is achieved firstly through the separation of data on these models to progress from a higher
exchange from the software agents involved in level of abstraction to a lower level of
the exchange, and secondly through the discrete abstraction.
nature of the service. • All of this activity is based on internation-
The biggest challenge may be the behaviour ally accepted standards.
of the users to adapt to the developed system. Ac-
cording to Chen, et al, (2006) the consumer of a Businesses that aim to support mobile work-
service is not required to have a detailed knowledge ers and enhance process effectiveness will need
of implementation, implementation language, or to consider extending their process and systems
execution platform of the service. The only concern beyond the workplace (Alag, 2006). According
of the consumer is how a service can be invoked to Godbole, (2006) Mobile Commerce is best
according to the service interface. suited where the consumer is driven by a “sense
Change management and transformation of an of urgency” when they need to have their goods
organization can be very difficult and sensitive is- and services immediately for upcoming functions
sues. Conversely, it can be argued that behavioural and events.
integration is critical to the success of enterprise Every organization on the planet consisting
integration. The technical integration can be a of more than one person has already realised
success but if the organization is not going to in- that their information technology infrastructure
ternalise the enterprise system, the entire project is is effectively a distributed computing system.
a failure. As such, to achieve the maximum benefit To integrate information assets and use infor-
and impact from enterprise integration, we need mation effectively, it must be accessible across
to have both successful technical and behavioural the department, across the company, across the
integration (http://delivery.acm.org). world and more importantly across the service-
The successful architecture confirms that busi- or supply-chain from the supplier, to one’s own
ness requirements and information technology organization, to one’s customers. This means that
design are captured in models. The modelling CPUs must be intimately linked to the networks
technique of abstraction to separate business of the world and be capable of freely passing
concerns from technology concerns (what the and receiving information, not hidden behind
business system needs to do, versus its underlying glass and cooling ducts or the complexities of the
computing platform) is also an important aspect software that drives them. www.omg.org/docs/
of the success of the architecture. omg/03-06-01.pdf
The following issues also could be classified
as the critical factor for the success of the Service
Oriented and Enterprise Architecture: UNDErstANDING sOA
AND WEb sErVIcEs
• To capture business requirements.
• Platform-Independent Model (PIM) by Service-Oriented Architecture is architecture
promoting designing a business solution based on internal and external processes of an
prior to selecting how it will be deployed. organization. Web Services technology is the
671
Figure 1. Internal and external impacts of WS

of large-scale electronic publishing, XML is also
playing an increasingly important role in the ex-
change of a wide variety of data on the Web and
elsewhere (http://www.w3.org/XML/).
XML schemas associated with SOAP message
payloads often need to be designed with some of
the more advanced features of the XML Schema
Definition Language. Specifically, the use of
extensible or redefined schemas may be required
when building documents that represent multiple
data contexts. See Figure 2.
most appropriate technology to develop SOA. WsDL AND sOA

Curbera, et al, (2003) states that Web Services
provide generic coordination mechanisms that can WSDL is an XML format for describing net-
be extended for specific protocols and Ghanbary work services as a set of endpoints operating on
(2006) extends this expression by stating that WS messages containing either document-oriented
represent the applications that organizations pub-
lish/locate on unknown and disparate platforms.
Figure 2.
According to Unhelkar & Deshpande (2004), Web
Services based technologies enable applications
to “talk” with one another even across organiza-
tional firewalls, resulting in an opportunity for a
cluster or group of organizations to simultaneously
transition to Web-based entities.
Barry (2003) clearly states that the use of Web
Services appears to be the missing puzzle piece in
creating a complete picture of a service oriented
architecture work. The statement given by Barry
(2003) identifies the importance of universal
adoption of Web Services by software vendors.
Figure 1 illustrates the importance of the
adoption of the Web Service by internal as well
as the external architecture. The following is the
explanation of the functionality of the Web Ser-
vices that could create successful service oriented
architecture.
XML/sOAP AND sOA
Extensible Mark-up Language (XML) is a simple,

very flexible text format derived from SGML (ISO
8879). Originally designed to meet the challenges
672
or procedure-oriented information. The opera- UDDI AND sOA

tions and messages are described abstractly, and
then bound to a concrete network protocol and Universal Description, Discovery and Integration
message format to define an endpoint. Related (UDDI) specifications define a registry service
concrete endpoints are combined into abstract for Web services and for other electronic and
endpoints services (http://www.w3.org/TR/wsdl). non-electronic services. A UDDI registry service
See Figure 3. is a Web service that manages information about
WSDL is extensible to allow description of service providers, service implementations, and
endpoints and their messages regardless of what service metadata. Service providers can use UDDI
message formats or network protocols are used to to advertise the services they offer. Service con-
communicate. Web Service Definition Language sumers can use UDDI to discover services that
SOA starts with the design of a service. Building suit their requirements and to obtain the service
software services start with the definition of what metadata needed to consume those services (http://
the service is and what the service does. SOA pro- www.uddi.org/faqs.html).
vides a standardised means of building software Universal Description, Discovery, and Integra-
services that can be accessed, shared, and reused tion (UDDI) discover the prospective requester
across a network. While SOA is a well-established from the directory that is also an integral part of
concept, it has become increasingly popular with an organization. This specification allows for
the emergence of Web Services. The starting point the creation of standardised service description
in developing SOA services is the Web Services registries both within and outside of organization
Description Language (WSDL). boundaries. UDDI provides the potential for Web
Figure 3.
673
services to be registered in a central location, According to Erl (2004) an SOA is a design

from where they can be discovered by service model with a deeply rooted concept of encapsulat-
requestors. Hence SOA services should be ac- ing application logic within services that interact
cessed, shared, and reused across a network. The via a common communication protocol. When
UDDI directory provides the channels of access Web Services are used to establish this commu-
across the network. nication framework, they basically represent a
web-based implementation of Service Oriented
Architecture. Business process integration is part
tHE UsE OF sOA IN of enterprise integration solutions, which is why
cOLLAbOrAtIVE OrGANIZAtIONs coordination services for business activities are
utilised exclusively for the management of long
One of the key challenges in modern day busi- running business activities.
ness is the pressing need to integrate their wide Based on Chung (2005), Web Services inte-
and varied software systems and applications. gration enables a dynamic e-business model that
Furthermore, large organizations such as banks fosters collaboration with heterogeneous business
and insurance companies have vast amount of services and opens the door for new business op-
data that is embedded in their legacy systems. portunities. A service-oriented architecture (SOA)
They have a need to expose those data and the is an application framework that takes everyday
corresponding applications in a ‘unified’ view to business applications and breaks them down
the customer on the Internet - resulting in what into individual business functions and processes,
is known as ‘business integration’. However, as called services.
a result of this integration, and technical ability Figure 4 will clearly explain how Service Ori-
of applications to transact over the Internet, ented Architecture will impact the requirements
businesses are now readily able to offer and of collaborative Organizations.
consume ‘services’ across the Internet. Cur-
rently, there is a limited of literature on mod- Figure 4. The role oF SOA in collaborative or-
elling and managing the challenges emanating ganizations
from collaboration between varied businesses
and applications.
Based on Pasley (2005), service interoper-
ability is paramount. Although researchers have
proposed various middleware technologies to
achieve SOA, Web services standards better sat-
isfy the universal interoperability needs. In order
for multiple organizations to collaborate many
challenges were identified as such: technologi-
cal, methodological and social factors resulting
rational interactions between businesses. The
good architecture takes place when the services
of different applications have the capability to
communicate. The previous statement leads us
to the concept of Service Oriented Application
departing beyond the boundary of standard com-
munications framework.
674
Figure 5. Percentage and size of the organiza-

Figure 4 shows the importance of SOA in
tions
developing the applications of Collaborative Or-
ganizations. The technology and the architectural
aspects of this integration based on collaboration
have also been demonstrated. The requirements of
collaborations as far as the multiple organizations
are the required technology, required methodology,
social threats, how to implement the integration,
how to architect the integration and investigate
the structural changes to the organization after
the integration.
ADAPtING MObILE AND WEb

sErVIcEs tEcHNOLOGIEs
tions. According to the ABS report, the organiza-
The interoperation amongst multiple organizations tions with the number of 10-200 employees are
needs a technology to support the collaboration classified as medium size and the organization
across their business process especially when with the number of more than 200 employees are
the participated organization are not necessarily classified as the large size organizations (Trevin,
known to each other, and have never collaborated 2001). Figure 5 illustrate the demographic of the
previously. organization based on their organizational size.
According to Barry (2003) the main driving The organizations span across different indus-
forces for adopting Web Services are classified tries, as listed in table below.
as interoperable network applications, emerging
industry-wide standards, easier exchange of data, the Organization category
reduced developing time, reduced maintain costs,
availability of external services and availability This section is related to the category of the
of training and tools. participated organization. The study identifies
The main restraining forces are also classified the importance of the organization’s category as
as different semantics in data source, semantic to reach different industries in order to evaluate
translation effect on operation systems for up- the general technological adaptation in different
to-moment data request, standards evolving not organization. Table 1 demonstrate the organi-
fixed and mergers and acquisition. zational categories in which responded to the
Based on our survey, we asked 60 different distributed survey.
organizations in the Sydney metropolitan area to The majority of the participants are from the
inform us about their ideas about the adoption of Information Technology sector of the industry. The
the Web Services from technical, methodological Government departments, education and banking
and social issues. The following Figures present are following in order. The study is able to proceed
the result of the survey which was already ap- hence the distribution of the questionnaire has been
proved by the Ethics committees. The 43% of the correctly allocated and the study can evaluate the
organization amounting 26, were medium sized result achieved based on the different category of
and 57% amounting 34, were large sized organiza- the organizations.
675
Table 1. The category of the organizations

management holds 13.3% of the respondents. The
Organizations Categories Number Percentage remaining of 48.3% of the respondents holds the
Information Technology 20 33.3%
key role positions in their organizations. These
Government Departments 14 23.3%
people are the decision makers in the organizations.
Education and Training 7 11.7%
Banking, Finance and Insurance 7 11.7%
MObILE tEcHNOLOGy
Professional Services (Legal and 5 8.3%
Accounting) INFOrMAtION
Retailing 3 5%
Health and Community Services 2 3.3% This section of the chapter demonstrates the result
Utility Services and Equipment 1 1.7% of the survey in regards to the respondents general
Manufacturing and Processing 1 1.7%
thought about mobility in business.
Total 60 100%
Importance of the Mobility
in the Organization
the Position of the Participant
This section is evaluating the use of mobile
in the Organization
technology (use of mobile devices) in the daily
activities of the business. The query further in-
This part is related to the position of the individuals
vestigates whether the organizations are already
in the organization who has actually responded to
using mobile technology, or are planning to use
the questions. This section is also very important
it in the near future. The responses are detailed
since the research can understand the role of the
in Figure 6.
respondent and their decision making power to
A substantial 87% (63% already using, and
change the technology of the organization. The
23% that plan to use in the near future) of the
positions of the respondent are presented in Table 2.
organizations responded in the affirmative to this
The participants who held the general man-
question, which verifies the fact that the key per-
agement positions in their organizations formed
sonnel in the selected sample are very much aware
the 21.7% while marketing manager and senior
of the value of mobility and mobile technology
for their organizations. Thirteen per cent of the
Table 2. The position of the respondent in the respondents said that they do not have a plan to
organization use mobile technology in the near future.
Position Number Percentage
Used Mobile Devices in
General Management 13 21.7%
the Organization
Marketing Manager 8 13.3%
Senior Management 8 13.3%
The question is to identify what kind of the devices
Systems Analyst/Programmer 8 13.3%
are in use in order to gain an insight into the current
IT/MIS Manager 6 10%
use of mobile technology in the organizations. The
Technical Support 6 10%
responses for this question are listed in Figure 7.
Executive Manager 5 8.3%
The main issue to consider in this section is that
Sales Officer 4 6.7% the organizations can use different devices in order
Customer Care 2 3.3% to proceed with their daily business activities. The
Total 60 100% survey identified that 40% of the organizations
676
Figure 6. The use of mobile technology in the organizations
Figure 7. Type of mobile devices in organizations
are currently using mobile devices while 30% of the scenario in Which
the respondent use mobile enabled laptops. The Mobile Devices are Used
Figure 8 clearly demonstrates that organizations
have realised that they need to take advantage of The respondents were queried about the use of
mobile technologies. these devices in their business activities. In this
section of the research is identifying the reason
for the use of mobile devices in order to identify
Figure 8. Typical reason for the use of mobile devices
677
the use of this technology in the collaborative 1. Mobile technology as a special technology
environment. The Figure 8 depicts the typical improve efficiency in customer meetings.
reasons for the mobile devices in the daily busi- 78.7% of the respondents either agreed or
ness activates. strongly agreed with this proposition.
Figure 8 presents the result by stating that the 2. Mobile technology has been used as a spe-
majority of participants believe that the use of cial tool to advertise in a captured market.
mobility has increased the performance of their 70.4% of the respondents agreed with this
business activities enabling them to have greater proposition.
access to the employee, as well as being accessible 3. Mobile technology as a tool has enabled
by their customers. The use of mobile devices has people to contact with the office for em-
created flexibility, availability and better access. ployees engaged in official travel. 63.8%
Questions 7, 8, 9, 12, 21, 22 and 23 have been agreed to this proposition.
designed to rate the answers on different method 4. Mobile technology has enabled the business
of evaluation. The legends are as followings:VSA to track goods in transit. About 70.5% of the
– Very Strongly AgreeSA – Strongly AgreeAg - responses agreed with this proposition.
AgreeDA – DisagreeSD - Strongly DisagreeVSD
– Very Strongly Disagree The result achieved states that the mobile tech-
Based on these ratings, the following results nology is a major technology which could improve
are extracted from the survey of organizations. In their business activities. Therefore, this technology
the followings the letter P stands for the Point as it could also facilitate this research to enter the new
was presented in the survey called Point number. proposed collaborative environment.
current Application of Advantages of Mobile technology

Mobile technology
The question establishes the advantage of using
The question asked whether there are any new mobile technology in organizations. Whilst the
applications and areas in which mobile technology question queried the availability of applications
could be included in the daily business activities of mobile technologies that can be included in
of the organization, under four propositions, as the organization, the objective of next question
listed in Table 3. is to re-emphasise this question in an alternate
way, by probing the advantages of using mobile
Table 3. New applications/areas for use of mobile technology
New Applications TOTAL VSA SA AG DA SD VSD

1- Special Technology-Improve Efficiencies 47 6 15 16 8 1 1
PERCENTAGE 100 13 32 34 17 2 2
2-Advertise in Captured Markets 44 2 9 20 8 2 3
PERCENTAGE 100 5 20 45 18 5 7
3-Contacting Office (Any where/Any time) 47 13 15 12 3 1 3
PERCENTAGE 100 28 32 26 6 2 6
4-Track Goods in Transit 51 8 12 16 8 3 4
PERCENTAGE 100 16 24 31 16 6 8
678
Table 4. Main advantages of mobile technology for oganizations
Mobile Advantages TOTAL VSA SA AG DA SD VSD

1- Cost Saving 44 8 8 16 10 1 1
PERCENTAGE 100 18 18 36 23 2 2
2-Connect Employees 48 21 15 9 3 0 0
PERCENTAGE 100 44 31 19 6 0 0
3- Improve Productivity 55 13 26 10 6 0 0
PERCENTAGE 100 24 47 18 11 0 0
4-Flexibility of Employees 47 7 10 17 12 1 0
PERCENTAGE 100 15 21 36 26 2 0
5- Better Access for Customers 48 10 13 18 6 0 1
PERCENTAGE 100 21 27 38 13 0 2
technology, rather than directly asking about new disregard of their location and time. 72.3%
applications. The results of this question are listed agreed to this proposition.
in Table 4. 5. Mobile technology has created better ac-
cess method for the customer to contact the
1. Mobile technology as a special technology is organization. 85.4% of the responses agreed
very cost efficient. 72.7% of the respondents with this proposition.
either agreed or strongly agreed with this
proposition. Advantage of Mobile
2. Mobile technology has been connecting technology for the business
people while out of the office. 93.8% of the
respondents agreed with this proposition. The question asked about other factors that would
3. Mobile technology has been improving the enhance the demand in introducing or using mobile
business productivity. 71% of the respon- technology in the organization. Four propositions
dents agreed with this proposition. were presented to the respondents to choose from.
4. Mobile technology has enabled employees Table 5 lists the results for the question.
to be more flexible hence they can work
Table 5. Factors influencing the use of mobile technology in an organization
Factors Influencing Mobility TOTAL VSA SA AG DA SD VSD

1- Mobility Demand by Employees 52 7 12 20 13 0 0
PERCENTAGE 100 13 23 38 25 0 0
2- Mobility Demand by Customers 50 4 12 21 12 0 1
PERCENTAGE 100 8 24 42 24 0 2
3- Mobility Demand by Supply Chain 48 3 12 17 12 3 1
PERCENTAGE 100 6 25 35 25 6 2
4- Mobility Demand by Social-Psych Factor 47 4 8 19 14 1 1
PERCENTAGE 100 9 17 40 30 2 2
679
1. Employees demand and show interest in Problem/Difficulties of

using the mobile technology. 75% of the Mobile technology
respondents either agreed or strongly agreed
with this proposition. The question investigated the anticipated prob-
2. Customer demand and show interest in using lems, difficulties and complaints the respondents
the mobile technology. 74% of the respon- may have when using the existing mobile gadgets.
dents either agreed or strongly agreed with The results are listed in Figure 10.
this proposition. The results state that the most important
3. Supply chain sector is more interested and difficulty is the small screen while the limited
show interest in using the mobile technology. applications, battery life span and complicity of
66.7% of the respondents either agreed or mobile devices are classified as the remaining
strongly agreed with this proposition. problem of mobile devices.
4. Social-Psychological factors are influencing
the people to use the mobile technology. 66% Disadvantages of Mobile technology
of the respondents either agreed or strongly
agreed with this proposition. The question investigated the disadvantages of
using mobile technologies, as perceived by the
Improvement caused by respondents. The results are listed in Table 6.
Mobile technology
1. Queried whether the cost of establishment
The question investigated the perceived value of of mobile applications is a concern for
mobile technology in the daily activities of the the organization. 74% of the respondents
organization. This section allows the respondents either agreed or strongly agreed with this
to select more than one choice. The survey results proposition.
are listed in Figure 9. 2. Queried the recurring cost of using mobile
The most important benefit of using the mobile technology as a major tool. 83% of the re-
devices as predicted has been classified as the spondents agreed with this proposition.
availability to be contactable at any time, and 3. Queried whether technical drawbacks, which
anywhere. A fact revealed by this question is that are inherent in current mobile technologies,
cost savings are not the main driver for organiza- are a factor considered as a disadvantage
tions to use mobile technology. by organizations. 93% of the respondents
agreed with this view.
Figure 9. Mobile technology advantages to business activities
680
Figure 10. Problems faced by organizations using the mobile gadgets
Table 6. The Recognised disadvantages of mobile technology
Disadvantages of Mobile Technology TOTAL VSA SA AG DA SD VSD

1- Establishment Cost of Applications 56 6 15 20 11 2 2
PERCENTAGE 100 11 27 36 20 4 4
2-Recruitment Cost of Mobility 54 5 16 24 6 2 1
PERCENTAGE 100 9 30 44 11 4 2
3- Technical Drawback 57 12 16 25 2 2 0
PERCENTAGE 100 21 28 44 4 4 0
4-Legal and Privacy Issues 56 9 12 24 9 1 1
PERCENTAGE 100 16 21 43 16 2 2
5- Training and Adaptation Issues 55 5 24 14 9 1 2
PERCENTAGE 100 9 44 25 16 2 4
4. Queried legal and privacy concerns using technical Drawbacks of

mobile technology. Around 80% of the re- Adaptation Web services
spondents showed concern about the legal
and privacy issues with regard to mobile The question investigated the adaptation to WS
technology. technology from the technical perspective pre-
5. Queried with adoption and training issues in sented in Figure 11.
an organization with regard to mobile tech-
nology. 78% of respondents in the selected 1. Queried whether the unfamiliar concept of
sample agreed that such issues are a concern the Web Services technology is the great
for their organizations. concern for the organizations in order to
adapt web Services. About 70% of the re-
WEb sErVIcEs tEcHNOLOGy spondents either agreed or strongly agreed
with this proposition.
The following questions help the research to 2. Queried whether the limitation of the Web
evaluate the participant’s opinion in order to Services is important. 65% of the respondents
adapt Web Services technology from technical, agreed with this proposition.
methodological and social prospective.
681
Figure 11. The technical issues adapting the Web services
3. Queried whether the ambiguity of the Web 2. Queried whether the training of the em-
Services is the major concern (What it is ployees is the main concern. 61.5% of the
and what it does). 70% of the respondents respondents agreed with this proposition.
agreed with this view. 3. The concept of the competition while col-
4. Queried whether the participants understand laborating (how can you collaborate with
how WS could facilitate collaboration. your competitor). 86.6% of the respondents
Almost 80% of the participants agreed with agreed with this view.
this proposition. 4. The focus will shift to technology rather
than process. Almost 80% of the participants
Methodological Drawbacks of agreed with this proposition.
Adaptation Web services 5. How to manage the change when adapting
the WS technology. Almost 75% agreed with
The question investigated the adaptation to WS the proposition.
technology from the Methodological perspective
presented in Figure 12. social Drawbacks of
Adaptation Web services
1. Queried whether the impact on WS on ex-
isting business process is the main concern The question investigated the adaptation to WS
while adapting the WS. About 64.9% of the technology from the social perspective presented
respondents either agreed or strongly agreed in Figure 13.
with this proposition.
Figure 12. The methodological issues adapting the Web services
682
Figure 13. The social issues adapting the Web services
1. Evaluate the adaptation rate by customer activities while 23% has stated that they are plan-
and the employees. About 60% of the re- ning to adapt it in the near future. The total of 87%
spondents either agreed or strongly agreed of the organizations recognises the importance of
with this proposition. the mobile technology while currently the majority
2. How the competitor react to change. 75% of of these people mainly use their mobile phones
the respondents agreed with this proposition. and mobile enabled laptops.
3. How the technology provide support in order The respondents defined that mobility is a
to trust the competitor. Almost 75% of the great communication tool, make it easier to find
respondents agreed with this view. and locate personnel, create more flexibility and
4. What happens to the organizations already increases the general productivity. The major
in line of collaboration? Almost 65% of the advantage of mobile technology is providing
participants agreed with this proposition. availability to people disregards of their location
5. The legal issues involved in collaboration and time. The study revealed that the accessibility
(Government and the internal policies). is one of the greatest advantage of the collabora-
Almost 65% of the participant agreed with tive organizations therefore the advantage of the
the proposition. mobility (Anywhere – Anytime) could provide
benefit to collaborative organizations.
ANALysEs OF tHE DAtA The survey has also investigated the current
and potential application of mobility, advantages
This section describes the further analyses the and disadvantages of mobility and the improve-
overall assessment of the survey in regards to ments caused by mobility to provide a better
the adaptation of Mobile and Web Services tech- understanding of this technology.
nology as far as these technology are aiding the Interestingly, the survey identified the cost of
collaborative organizations. mobility is not classified as a big disadvantage in
comparison to the benefit it provides. All these
Mobile technology disadvantages and drawbacks seem to be due to
Information (Evaluation) the fact that the technology is new and still evolv-
ing. When there is more commercialisation of the
The mobility appears to be an important tech- technology, applications will become cheaper and
nology for the businesses to precede their daily recurring costs will be less. The decreasing cost of
activities. 63% of the organizations are already technology while the capabilities are improving
using their mobile devices to run their ordinary rapidly is highlighted in (Roth, 1998).
683
Web services technology of change management has been classified as

(Evaluation) important concepts. Almost 40% of the partici-
pants agree with these issues while close to 30%
The interoperation amongst multiple organizations strongly agree. Close to 20% of the respondents
needs a technology to support the collaboration classify the concepts of change management as
across their business process especially when their greatest concern for adopting WS technol-
the participated organization are not necessarily ogy for their organization. However, about 20%
known to each other, and have never collaborated and 30% of these organizations disagree that the
previously. affect of WS on existing business processes and
According to Barry (2003) the main driving training the employees as a great importance.
forces for adopting Web Services are classified There is no doubt that the methodological issues
as interoperable network applications, emerging play an important role in adopting Web Services
industry-wide standards, easier exchange of data, technology.
reduced developing time, reduced maintain costs, Based on the survey, the customer/ competi-
availability of external services and availability tor reaction and the impact on the organizations
of training and tools. that already in collaboration are classified by the
The main restraining forces are also classified participants as a strongly agree point. Almost
as different semantics in data source, semantic similar number is attracting the very strongly agree
translation effect on operation systems for up- comments however up to 35% of all participants
to-moment data request, standards evolving not classify all the issues identified by the study as
fixed and mergers and acquisition. the major drawback for adaptation of WS by
Based on our survey, all the issues identified organizations.
by the research such as unfamiliarity concepts of The research concludes that technical, meth-
Web Services, limitation of Web Services, how to odological and social factors identified by this
adapt the new technology and how the processes study are to impose the adaptation and adoption
collaborate are classified as the major concerns of of new technology by organizations. The busi-
the organization in order to adopt Web Services. ness opportunities resulting from WS have seen
Only the minority of 10% of the participants these technologies being rapidly adopted across
very strongly agree to understanding how Web the world. For example, an IDC Report in 2003
Services could help the collaboration while close revealed that 30% of Australian organizations
to 30% of the participants had the same concern are already using web services – although a large
with ticking the strongly agree box. The research number of these organization’s applications are
has concluded that the organization knowledge in behind the corporate firewall. Another survey
regard to the technical issues of WS is very limited. conducted by CSC found 105 of Australia’s largest
This lack of knowledge could be classified as the organizations are already using web services or
major drawback in adaptation of Web Services. planning to do so (Mackenzie, 2003).
More work is required to educate enterprises in In general, the issues of incompatible technol-
regards to the capability and functionality of Web ogy, competition, licensing agreement (legal is-
Services from technical point of view. sues) and mistrust are also classified as additional
Based on the result of the survey, all the issues major concern while adopting the new technology
identified by the study such as impact of Web to be discussed later in the thesis.
Services on existing processes, training of the em- The research concludes that technical, method-
ployees, concept of competition in collaboration, ological and social factors identified by this study
shifting the focus of technology and the concept are to impose the adoption of new technology by
684
organizations. In general, the issues of incompat- Chase, N. (2006, February 14). Introducing the
ible technology, competition, licensing agreement open Group Architecture Framework, Under-
(legal issues) and mistrust are also classified as standing TOGAf and IT Architecture in today’s
additional major concern while adopting the new World.http://www-128.ibm.com/developerworks/
technology. ibm/library/ar-togaf1/ Accessed: 3/04/2007
Chen, X., Wenteng, C., Turner, S. J., & Wang,
Y. (2006). SOAr-DSGrid: Service-Oriented
cONcLUsION AND DIrEctION
Architecture for Distributed Simulation on the
Grid. Proceedings of the 20th Workshop on Prin-
Service Oriented Architecture and Web Services
ciples of Advanced and Distributed Simulation.
were introduced in this chapter extending the men-
ISSN:1087-4097, 0-7695-2587-3
tioned architecture and technologies that support
the Collaborative Business Process Engineering. Chung, J. Y. (2005). An Industry View on Service-
The technical, methodological and social factors Oriented Architecture and Web Services. Proceed-
in order to adopt Web Services technologies by ings of the 2005 IEEE International Workshop on
organizations were also investigated. This chapter Service-Oriented System Engineering (SOSE’05)
has also described a survey carried out in the Syd- 0-7695-2438-9/05 © 2005 IEEE.
ney metropolitan area in large and medium-sized
Curbera, F., Khalaf, R., Mukhi, N., Tai, S., &
organisations, in order to assess the organisations’
Weerawarana, S. (2003, October). The Next Step
concerns, readiness for and adaptability to emerg-
in Web Services. Communications of the ACM,
ing technologies of Mobile and Web Services
46(10). doi:10.1145/944217.944234
technologies. The final result of the survey has
revealed (in fact within the selected sample) that Erl, T. (2004). Service-Oriented Architecture. A
the key personnel of the organisations agree with Field Guide to Integrating XML and Web Services.
the major concerns identified by the study and Pearson Education, Inc. ISBN: 0-13-142898-5
queried in the survey. The above 60% rate has been
Falcone, F., & Garito, M. (2006). Mobile Strat-
the result achieved for every individual question.
egy Roadmap. In B. Unhelkar (Ed.), Handbook
The chapter has presented pictorial illustrations
Resources of Mobile Business. Hershey, PA USA:
of the achieved results and the analyses provided
Idea Group. ISBN: 1591408172
of the collected data are also discussed.
Finkelsteing, C. (2006). Enterprise Architecture
for Integration. Rapid Delivery Methods and Tech-
rEFErENcEs nology. British Library Catalogue in Publication
Data. ISBN: 1-58053-713-8
Alag, H. (2006). Business Process Mobility. In B.
Unhelkar (Ed.), Handbook Resources of Mobile Ghanbary, A. (2006). Collaborative Business
Business. Hershey, PA, USA: IGI Global. USA. Process Engineering across Multiple Organiza-
tions. A Doctoral Consortium. Proceedings of
Barry, D. K. (2003). Web Services and Service- ACIS 2006. Australia: Adelaide.
Oriented Architecture. The savvy Manager’s
Guide. USA: Morgan Kaufmann Publishers. Godbole, N. (2006). Relating Mobile Computing
ISBN: 1-55860-906-7 to Mobile Commerce. In B. Unhelkar (Ed.), Hand-
book Resources of Mobile Business. Hershey, PA
USA: Idea Group. ISBN: 1591408172
685
Harrison, A., & Taylor, J. I. (2005). WSPeer - An The ACM Digital Library. http://delivery.
Interface to Web Service Hosting and Invoca- acm.org/10.1145/610000/606273/p54-lee.
tion. Proceedings of the 19th IEEE International html?key1=606273&key2=4006199511&col
Parallel and Distributed Processing Symposium l=GUIDE&dl=portal,ACM&CFID=11111111
(IPDPS’05). ISBN: 1530-2075/05 &CFTOKEN=2222222#lead-in. Downloaded:
5/10/2006.
http://www.ebizq.net/hot_topics/soa/fea-
tures/5857.html?&pp=1.Downloaded: The open Group Integration Consortium. http://
12/10/2006 www.integrationconsortium.org/docs/W054final.
pdf. Downloaded: 5/10/2006.
Jostad, I., Dustdar, S., & Thanh, D. V. (2005). A
Service Oriented Architecture Framework for Col- UDDI.org. http://www.uddi.org/faqs.html. Down-
laborative Services. Proceedings of the 14th IEEE loaded: 12/10/2006
International Workshops on Enabling Technolo-
Unhelkar, B., & Deshpande, Y. (2004). Evolv-
gies: Infrastructure for Collaborative Enterprise.
ing from Web Engineering to Web Services: A
ISBN ~ ISSN:1524-4547, 0-7695-2362-5
Comparative study in the context of Business
Miller, J., & Mukerji, J. (2003). Model Driven Utilization of the Internet. Proceedings of ADCOM
Architecture (MDA) Guide Version 1.0.1. http:// 2004, 12th International Conference on Advanced
www.omg.org/docs/omg/03-06-01.pdf. Down- Computing and Communications, Ahmedabad,
loaded: 5/10/06 India, 15-18 December
Pasley, J. (2005, May/June). How BPEL and SOA W3Consortium. http://www.w3.org/TR/wsdl.
are changing Web services development. Internet Downloaded; 12/10/2006
Computing, IEEE, 9(3), 60-67. Digital Object
W3Consortium. http://www.w3.org/XML/.
Identifier 10.1109/MIC.2005.56
Downloaded: 12/10/2006
of IGI Global).
686
687
Chapter 3.9
Towards a Model-Centric
Approach for Developing
Petraq Papajorgji
Center for Applied Optimization, University of Florida, USA
Panos M. Pardalos
Center for Applied Optimization, University of Florida, USA
AbstrAct ing, and managing different kinds of software

components without considering any particular
This chapter aims to present a new modeling implementation technology. At the center of this
paradigm that promises to significantly increase approach are models; the software development
the efficiency of developing enterprise informa- process is driven by constructing models repre-
tion systems. Currently, the software industry senting the software under development. Code
faces considerable challenges as it tries to build that expresses the implementation of the model
larger, more complex, software systems with in a certain underlying technology is obtained as
fewer resources. Although modern programming a result of model transformation. Thus, the intel-
languages such as C++ and Java have in general lectual investment spent in developing the business
improved the software development process, they model of an enterprise is not jeopardized by the
have failed to significantly increase developer’s continuous changes of the implementation tech-
productivity. Thus, developers are considering nologies. Currently there are two main approaches
other paths to address this issue. One of the poten- trying to implement MDA-based tools. One of
tial paths is designing, developing and deploying the approaches is based on the Object Constraint
enterprise information systems using the Model Language and the other on Action Language. An
Driven Architecture (MDA). MDA is a model- example of designing, developing and deploying
centric approach that allows for modeling the an application using this new modeling paradigm
overall business of an enterprise and capturing is presented. The MDA approach to software
requirements to developing, deploying, integrat- development is considered as the biggest shift
since the move from Assembler to the first high
DOI: 10.4018/978-1-61520-625-4.ch010 level languages.
Towards a Model-Centric Approach for Developing Enterprise Information Systems
INtrODUctION models should be provided with behavior to be

able to be executed and therefore tested. At the
Business environment is very dynamic. On the one center of this approach are models; the software
hand mergers and acquisitions impose changes development process is driven by constructing
in the business model and therefore changes in models representing the software under develop-
the information system. On the other hand, the ment (Papajorgji, Clark & Jallas; Pastor & Juan
introduction of new technologies forces constant Carlos Molina, 2007). Code that expresses the
change in the business environment. In this case implementation of the model in a certain under-
the same business model has to be rewritten us- lying technology is obtained as a result of model
ing a new implementation technology. In order transformation.
to survive, businesses must have the ability to The MDA approach is often referred to as a
thrive in a continuously changing and difficult model-centric approach as it focuses on the busi-
to predict environment. Therefore, it is important ness logic rather than on implementation techni-
to know what aspects of the business are more calities of the system in a particular programming
likely to change and what are not when dealing environment (MDA Guide V1.0.1, 2009). Thus,
with constant change. the focus of OMG is to take advantage of new
An important part of the efforts in developing technologies and make the application develop-
a new information system are spent in mastering ment process independent of the infrastructures
the technological complexity of the solution (a they use. Once the model is constructed, it can
particular implementation technology such as Java be transformed into code in several languages
or .NET or other technology) rather than focusing (Pastor & Juan Carlos Molina, 2007). The MDA
on understanding the problem to be solved (Pastor approach to software development is considered
& Juan Carlos Molina, 2007). Experience shows as the biggest shift since the move from Assembler
that most of the time it is the implementation to the first high level languages.
technology that changes rather than the business An information system is regarded in MDA
model. Therefore, it is desirable to have a software from three different perspectives: computation
engineering approach that separates the business independent, platform independent and a platform
knowledge from the implementation technology specific. The computation independent perspec-
so they can continue to develop without neces- tive focuses on expressing the business model
sitating a complete rework of existing systems (business processes, stakeholders, departments
(Papajorgji, Clark & Jallas, 2009; Pastor & Juan and the relationships amongst them) without any
Carlos Molina, 2007). consideration about the information system that
The Model Driven Architecture (MDA) is a will be used during the implementation phase.
framework for software development defined by Efforts to represent the business model indepen-
the Object Management Group (OMG) (Object dently of the computing platform are referred
management group [OMG], 2009; OMG Model to as Computational Independent Model (CIM)
driven architecture: How systems will be built). (Gasevic, Djuric & Devedzic, 2006; Kleppe,
The OMG defines MDA as “Fully-specified plat- Warmer & Blast, 2003; Object Management
form independent models can enable intellectual Group, 2009). A CIM represents a high level
property to move away from technology-specific model that describes the relationships amongst
code, helping to insulate business applications principal elements of the business model. As an
from technology evolution and further enable in- example, a simple model expressing relationships
teroperability” (OMG Model driven architecture: between the manufacturing sector and the sales
How systems will be built). As presented by OMG, department can be considered as a CIM. Usually
688
CIM’s are designed by analysts. Parts of a CIM pre-conditions and post-conditions for an opera-
may be supported by software, but in general a tion. The pre- and post-conditions are specified
CIM is software independent. using the Object Constraint Language (OCL),
The platform independent perspective focuses (Warmer & Kleppe, 1999) which is a formal
on the functional capabilities of a system without language for specifying assertions (formally, an
considering any specific platform (or set of plat- “expression language”). Although Meyer’s work
forms) that will be used for its implementation. has been well implemented in several program-
Such a model is referred to as PIM (Platform ming languages (one of these languages is his own
Independent Model) (Object Management Group, language Eifel (Meyer, 1992)), it is not sufficient
2009, OMG/MDA). PIM’s are constructed using to completely generate code from the pre- and
UML and they express the relationships amongst post-conditions of an operation. Even the most
concepts of the domain under study without ardent defenders of this approach (Kleppe, Warmer
referring to any particular computing platform & Blast, 2003) state that total code generation
(Kleppe, Warmer & Blast, 2003; Papajorgji & may not be possible and the corresponding code
Shatar, 2004). has to be added manually in later phases of the
A platform specific perspective enriches a implementation.
platform independent perspective with details
relating to the use of a specific computing plat- the Object constraint
form. In this case, details of the technology used Language (OcL)
for the implementation of the PIM will be part of
the model. A model that represents details of the OCL was first developed in 1995 by a group of
underlying technology is referred to as a Platform researchers at IBM Europe with the goal of creat-
Specific Model (PSM) (Object Management ing a formal language and yet simple to be used
Group, 2000, OMG/MDA, 2009). by software engineers (Warmer & Kleppe, 1999).
Currently the MDA market is evolving. Several Currently OCL is part of the UML standards and
vendors have implemented OMG guide lines using thus available to modelers to apply constraints to
one or the other approach such as Kennedy Carter objects (Kleppe, Warmer & Blast, 2003; Papa-
(Kennedy Carter Inc., 2009), Oliva Nova (Pas- jorgji, Clark & Jallas, 2009; Warmer & Kleppe,
tor & Juan Carlos Molina, 2007), ANDROMDA 1999). Constraints are necessary as they provide
(http://www.andromda.org/), Virtual Enterprise software engineers with a set of well-defined
(http://intelliun.com) (Visual Enterprise, 2009) rules in order to define and control the behavior
to name a few. of objects. Within UML, OCL is used to specify
After analyzing the current state of the MDA invariants, preconditions, postconditions, and
implementation, it is reasonable to say that there other kinds of constraint (Warmer & Kleppe,
are two main responses to the OMG’s defini- 1999). OCL extends the power of UML as it al-
tion: the elaborationist and the translationist lows software engineers to create more expressive
approach. and more precise models.
OCL is purely an expression language: it does
the Elaborationist Approach not have side effects as it does not change the
status of the model (Papajorgji, Clark & Jallas,
The approach, referred to as the elaborationist 2009, Warmer & Kleppe, 1999). OCL is not a
approach, is embraced by the followers of the programming language and thus, it does not al-
school of Meyer (Meyer, 1988). According to the low writing code. When an expression in OCL is
elaborationist view, behavior can defined using evaluated, it simply delivers a value.
689
Using OCL with UML element of the UML model (Kleppe, Warmer &
Blast, 2003) and the OCL expression expressing
Initially, OCL has been used to specify constraints a certain constraint is always evaluated from the
on objects while defining object’s behavior. The point of view of this particular context. UML pro-
behavior of an object can be expressed using vides a particular stereotype, noted <<invariant>>
preconditions and postconditions on its operations to indicate an invariant constraint. In a class dia-
(Kleppe, Warmer & Blast, 2003; Meyer, 1988; gram constraints are presented between brackets
Warmer & Kleppe, 1999). This approach was in a note as shown in Figure 1. This figure shows
introduced and largely used by Meyer (Meyer, that the context of the constraint is Employee, a
1988) and it is the basis of his “Design by Con- UML class, represented by self.
tract” philosophy of software development. Only
recently, has OCL been used to express other types
of expressions such as specifying initial values for PrEcONDItIONs AND
attributes, specifying derivation rules for attributes POstcONDItIONs
or associations, specifying guard conditions in
state charts and etc (Kleppe, Warmer & Blast, Preconditions and postconditions are an effective
2003). OCL constraints are always connected way to express the semantics of operations and
to an object-oriented model be it a UML model methods (Warmer & Kleppe, 1999). A precondi-
or other type of object-oriented model. In UML, tion is a constraint that must be true at the start of
constraints can be used in different places. the execution of the operation and a postcondition
is a constraint that must be true at the end of the
Invariants execution of the operation. UML provides the
stereotypes <<precondition>> and <<postcondi-
An invariant is a constraint that can be associated tion>> for expressing these kinds of constraints. In
to any UML modeling element. An invariant states a class diagram though, preconditions and postcon-
that the result of the expression must be true for all ditions can be presented using notes similarly as
instances of the UML modeling element any time. in the case of invariant as shown in Figure 1. The
The context of an invariant is always a specific precondition in Figure 1 shows that the amount
Figure 1. Example of constraints and preconditions
690
must be non negative and the postconditions shows Simons, 2003; McNeile & Simons, 2004). In
that the salary after the increase is equal to the general, the translationist approach is based on
salary before plus the amount. the idea that the behavior of an object can be ex-
pressed using state machines. The main tool for
the translationist Approach expressing object’s behavior in the translationist
philosophy is the Action Language.
The second approach, referred to as the trans-
lationist approach, embraces the early efforts of
Mellor’s state machine for describing behavior ActION LANGUAGE
(Mellor & Balcer, 2002). According to this ap-
proach, only a PIM is needed to represent the For quite sometime the followers of the transla-
conceptual model and it is translated directly to tionist approach have required OMG to launch
code (McNeile & Simons, 2003). This second a Request for Proposals (RFP) in order to define
approach has mostly been used in the embedded the standards for a language that is UML com-
system world, but lately has positioned itself as a patible, executable and complete, implementa-
viable approach to MDA under the name of “Ex- tion independent at a level of abstraction above
ecutable UML”. Executable UML combines an any implementation technology. In 2001 OMG
abstract program model with a platform-specific adopted the Precise Action Semantics for the
model compiler that outputs an executable ap- Unified Modeling Language specification. The
plication. Grady Booch in his talk The Limits of standard was the result of a collaborative work of
Software states: a large industry consortium comprising some of
the best-known companies in the field of software
Today, we’re at the beginning stages of the next engineering such as Rational Software (http://
level. Executable UML is the next logical, and ibm.com), Kabira Technologies (http://www.
perhaps inevitable, evolutionary step in the ever- kabira.com/), Mentor Graphics (http://www.
rising level of abstraction at which programmers mentor.com/), Kennedy-Carter (http://kc.com),
express software solutions. Rather than elaborate Oliva Nova (http://www.care-t.com) and others.
an analysis product into a design product and then The Precise Action Semantics for the UML
write code, application developers of the future standard provides a well-defined and an unam-
will use tools to translate abstract application biguous semantic set of the operations needed to
constructs into executable entities. Someday soon, specify object’s behavior in a UML model. The
the idea of writing an application in Java or C++ level of detail is such that a translation engine (or
will seem as absurd as writing an application in compiler) can completely generate an executable
assembler does today. And the code generated from application from that model. The semantic set
an Executable UML model will be as uninteresting includes operations that support the so-called
and typically unexamined as the assembler pass CRUD (create, read, write, delete) manipula-
of a third generation language compile is today. tion of objects, the generation and handling of
asynchronous events (signals), and all the logi-
Recently, under the same umbrella of the cal constructs to support the definition and the
translationist approach, it has appeared another specification of algorithms. The modeler defines
slightly different method referred to as the event the behavior of objects using a text-based ac-
based approach. This approach considers events tion language. The main characteristics of this
as the focus of the modeling effort (McNeile & language are:
691
• The action language allows modelers to parameter. In this example the

define behavioral specification at a higher association between Order and
level of abstraction. Item is one to many.
• The language is independent of any spe- AllOrders.select ([:order | or-
cific implementation technology in the ex- der.items.detect([:item | item.
ecution environment. itemNumber == $itemNumber])
!=null ])
In order to illustrate the idea of defining ob-
ject’s behavior using the Action Language let us The final product of the MDA approach is
refer to an MDA tool named Virtual Enterprise generating code to be executed in a defined plat-
(Visual Enterprise, 2009) that is used to develop form. How is code generated from models? This
the application presented in section MDA and is the subject of the next section.
Application Development. The model allows for
searching for a job description using the following MDA Transformations
expression in Action Language:
The central idea in the MDA approach is to start
Dialog.readln(first(TheJobBoard. first by developing a PIM. The PIM must represent
jobDescriptions[code == $code])) the main concepts of the domain under the study
As presented in the expres- and their relationships. The PIM may also include
sion, TheJobBoard.jobDescrip- some general rules for model transformation.
tions returns a collection of Note that the same PIM could be transformed in
job descriptions present in different PSM based on the specific computing
the system. first(TheJobBoard. platform. As an example, the same PIM could be
jobDescriptions[code == $code]) transformed into an EJB (Enterprise Java Beans)
returns the first job descrip- or into a relational-based (SQL-based) platform.
tion whose code is equal to the In this case the rules guiding the transformation
value of the input parameter process are different. Transformations in the MDA
$code. The execution of the en- approach are presented in Figure 2.
tire expression will display job The transformation rules are a set of unam-
descriptions details satisfying biguous specifications that will guide the trans-
the criterion. formation process with the goal of obtaining a
The following expression will target model. The target model could be another
display the item of a certain model (presented in UML) or code in a specific
number selected from the col- programming language (Kleppe, Warmer & Blast,
lection of items related to an 2003; Sewal, 2003).
order; $itemNumber is the entry
Figure 2. Producing code from a PIM model through transformations
692
PIM to Java transformation • A UML class will be transformed into a

Java class,
Let us consider a simple example to illustrate • An attribute in the UML model will be
this transformation process. Figure 3 shows a transformed into a private attribute in Java,
simple UML model with two classes related with • For each attribute in the UML class two
an association and our goal is to describe the public methods (a get and a set method) are
transformation of this simple model into code in generated in the corresponding Java class.
Java. The model shows that classes Student and • For each association end there is a private
Course are linked with an association named attribute of the same name in the opposite
takes, which models the fact that a student takes class of type equal to the type of the class
courses. Note that the association takes is one to in the case when the association is one to
one or many. Class Student contains three attri- one. In the event the association is of type
butes: firstName, lastName and dateOfBirth and one to many, as it is our case, then the type
class Course contains attributes courseName, of the private attribute is Set. In the case
description, and instructor. the association is navigable (directed as-
A UML class will be transformed into a Java sociation) then the private attribute will be
class, thus classes in the model presented in Figure added only to the class origin of the asso-
2 will be transformed into Java code as presented ciation (in our example, class Student).
in Figure 4. The transformation rules are:
Figure 3. Classes related with an association
Figure 4. PIM to Java transformation
693
Figure 5. PIM to relational transformation
Note that the principle of encapsulation is onto relationship takes as shown in Figure 5. Note
respected during the transformation process; val- that the association takes in Figure 5 is necessary
ues of attributes are accessed only by the object only to visually show that tables student and course
itself. Therefore, both classes Student and Course are linked using the foreign key studentId. It is
attributes are declared private and the correspond- important to note that non foreign keys attributes
ing get and set methods are defined. In a similar may contain the NULL value whereas attributes
way are defined the rules of transformation for that are foreign keys, such as studentId, may
obtaining a relational model from a PIM model not contain NULL values. This observation will
expressed in UML. be clear if the Entity-Relationship (ER) model
obtained by the transformation process is viewed
PIM to Relational Transformation with an ER tool.
Let us consider the same model presented in MDA and Application Development
Figure 4 and transform this PIM into a relational
model. Before presenting the transformation of Objects, Associations and
the model, let us introduce some rules that guide Code Generation
the transformation of attributes defined in classes
Student and Course. Some of these rules are de- Let us consider a simple example to demonstrate
fined as follows (Kleppe, Warmer & Blast, 2003): the power of this approach. Figure 6 shows a
simple model where the relationship between the
• A UML string will mapped onto a SQL objects Client and Address is modeled as a simple
VARCHAR. one-to-one association.
• A UML integer will be mapped onto a SQL The model execution causes the system to
INTEGER. automatically generate the web page as shown in
• A UML date will be mapped onto a SQL Figure 7. Note that there are two input fields for
DATE. both attributes of object Client and a hyperlink
A class in UML will be transformed onto a

table in the relational model. Thus, classes Student Figure 6. Client-address relationship modeled as
and Course will be transformed onto two tables a simple association
named student and course as shown in Figure 5.
Attributes of each class are transformed using the
above-mentioned rules.
Associations in UML are transformed into for-
eign keys in the Entity-Relationship (ER) model.
The relationship takes of Figure 4 is transformed
694
Figure 7. Web form representing the simple association
Figure 8. Client- Address relationship modeled owns the part (Booch, Rumbaugh & Jacobson,
as a composition 1999; Papajorgji & Pardalos, 2006). The model
execution will automatically generate the web
form shown in Figure 9. In this case, data entry
for objects Client and Address are presented in
the same web form; data entry fields for the ob-
ject Address (the part) are presented in a tab in
the form where data entry fields for object Client
named Address, that when used, it will display are presented. When the form is submitted, the
another page prompting to enter data for attributes submitted data are used to populate objects Cli-
of object Address. As it is shown in Figure 7, the ent and Address at the same time. The lifecycle
life-cycle of objects Client and Address are com- of both objects are synchronized, they are created
pletely independent: Instances of objects Client and deleted at the same time. Note that object
and Address can be created, modified and deleted Client (the whole) is responsible for creation,
independently. maintenance and destruction of object Address
Figure 8 shows the relationship between Client (the part) (Booch, Rumbaugh & Jacobson, 1999;
and Address modeled as a composition. A com- Papajorgji & Pardalos, 2006).
position is a strong relationship where the whole The above example shows the relevance of
understanding the relationships amongst concepts
Figure 9. Web form representing the composition association
695
Figure 10. Packages representing parts of the

to post openings, to view submitted applications,
application domain
and other necessary functionalities for managing
well the system. Three packages are created; each
holding a well-determined part of the application
is shown in Figure 10. Thus, package applicant
contains classes and relationships among them
for all objects necessary to cover functionalities
dealing with applicants. These functionalities
allow applicants to browse for jobs, apply for a
job, and so on.
Each package contains a UML model repre-
senting only a subset of the entire model. The
in the domain under study; different models gen- main reason for separating the entire domain in
erate different views. packages is that it helps reuse particular compo-
nents of the system. Figure 11 shows classes with
Example of Application Development their attributes, methods and relationships of the
core applicants package. The model shows that
In order to show issues that need to be addressed a job board contains many job openings; many
during the design and the implementation of a applicants may apply for a job opening and an
system using the MDA approach, let us consider applicant may apply for many job openings.
as an example, the system managing job offers and The job board contains information for many
job applications presented in (Visual Enterprise, applicants.
2009). The system should allow applicants to Figure 12 shows classes with their attributes,
browse job offers, apply for a job, browse his/her methods and relationships necessary to model an
application status, and so on. The system should applicant. An applicant has a profile expressing
provide the human resources with the capability the work history of the applicant, his/her educa-
to update different job categories, job descriptions, tion and technical skills.
Figure 11. Classes, attributes and relationships for the job package
696
Figure 12. The applicant model
The UML model shown in Figure 13 shows a collection of all job openings in the system. In the
classes with their attributes, methods and relation- event a particular posted job opening is needed it
ships contained in the package jobs. The black dot can be identified using the following instruction:
denotes the root object of the system; it serves as
the starting point for navigating the system. Object TheJobBoard.jobCategorie.select ([: j | j.status
JobBoard is linked to object JobOpening using == “Posted” && j.jobDescription.jobCategory.
the association jobOpenings of cardinality 0 to >name == name])
many. Thus, TheJobBoard.jobOpenings provides
Figure 13. UML model for the job package
697
Figure 14. Joining values of attributes
The MDA-based tool used (Visual Enterprise, list of potential values of job categories is created by
2009) has an interesting feature that links the value constraining the domain of association jobCategory
of an attribute to a set of potential values. As shown to contain only the job categories created in the sys-
in Figure 13, attribute category of class JobDescrip- tem. Thus, the values for this domain are defined by
tion is prefixed by the sign “/”. This sign shows that TheJobBoard.jobCategories that returns a collection
the value of attribute category can be selected from of job categories created in the system.
a list of all instances of class JobCategory already Use cases are implemented as processes. Figure
created in the system. Figure 14 shows the formula 15 show how processes can be ensemble to create
for calculating the value of attribute category of class a navigation model that will allow users to launch
JobCategory using the association jobCategory. The different functionalities provided by the system.
Figure 15. The navigation model
698
Persistence in the MDA Approach WEb sErVIcEs
One of the advantages of using the MDA approach A new approach that is currently gaining strong
for developing complex enterprise systems is support in the software industry is based on
that the persistence of data is done automatically. considering enterprise solutions as federations
As previously mentioned, MDA tools interpret of services connected via well-defined contracts
the UML model and automatically generate the that define their service interfaces. The resulting
database schema for the corresponding model. system designs are often referred to as Service
Figure 16 shows the database schema generated Oriented Architectures (SOAs) (Brown, Conal-
by Virtual Enterprise tool (Visual Enterprise, 2009) len & Tropeano, 2005, pp. 403-432; Johnston &
for the UML model presented in Figures 11, 12 Brown, pp. 624-636).
and 13. The user has only to define the name of The technology used for developing web-
the database and some other parameters and the services is not relevant to the service a system
database and the functionalities for storing the provides to its users/clients. The view of the
data onto the database are generated by the MDA service should be totally independent from its
tool. Note that the MDA tool does not normalize implementation in some underlying technolo-
the model; therefore it is the user’s responsibil- gies. As previously mentioned in this paper, the
ity to design a UML model that does not require main characteristics of the MDA approach is
normalization. the independence of the model describing the
Figure 16. Database schema generated from the UML model
699
business from its implementation in underlying the ERP solution and therefore large amounts of
technologies. Therefore, using the MDA approach money must be invested in consultancy for ERP
to design and develop web-services will provide providers.
the same benefits; independence, flexibility and A solution to these problems is the use of
speed in software development. the MDA approach to configure functionalities
MDA provides an environment to design provided by an ERP system. MDA provides
Web services at a more abstract level than that enterpriser managers with a high level model,
of technology-specific implementations. The totally independent of ERP’s target platform, at
technologies implementing Web services depend the level of business processes and expressed
on Port 80. The fact that developers of Web using visual tools. The UML extension proposed
services program directly in these technologies by (Ericsson & Penker, 2000) closely matches
makes software development vulnerable to rapid business modeling standards and complements
obsolescence and is also far too labor-intensive. the UML Business Modeling Profile published
Systems built using MDA exhibit more flex- by IBM (Johnston, 2004).
ibility and agility in the face of technological One of the most relevant efforts to use the MDA
change—as well as a higher level of quality and approach in a context of an ERP implementation
robustness, due to the more formal and accurate is presented by (Dugerdil & Gaillard, 2006). They
specification of requirements and design. have used Eriksson-Penker’s UML profile (Erics-
son & Penker, 2000) and implemented it using
MDA and ErPs IBM’s XDE modeling tool to create a business
modeling environment in the context of an ERP
Enterprise Resource Planning (ERP) systems are implementation.
complex solutions for managing the multi-facet ERPs provide a large amount of functionalities
aspect of modern enterprises. Initially, ERPs were and their use depends on the particular needs of
used only in the context of large corporations. the business model of a corporation. In order to
Currently, there is a stronger effort to use ERPs activate the functionalities of an ERP system,
by small and middle size companies to develop some configuration tables need to be created that
their core information technology systems (van will make available the necessary components/
Everdingen Y., van Hillergersberg & Waarts, modules. As previously mentioned producing
2000). There is a wide spread opinion in the indus- configuration tables is difficult and requires a
try that ERPs can be seen as a viable alternative to detailed knowledge of the ERP system and a
custom application development for the standard considerable amount of manual labor. As the
management needs (Dugerdil & Gaillard, 2006). ERP’s functionalities to be activated depend on
Besides numerous advantages of using ERPs as the business model, then it is natural to develop a
a basis for the IT needs, there are several major business model at a high level of abstraction using
disadvantages that are a serious obstacle for using UML-based visual tools. Constructing a high level
ERPs in a large scale by small and medium enter- business model is also one of the best practices in
prises. One major issue is the inherent complexity ERP implementation (Thomas, 2002). The use of
of ERPs; often their implementation requires a the MDA approach when the target system is an
large dependency on the ERP provider that most ERP requires that the transformations leading a
of the managers would like to avoid at any cost. PIM into a PSM, instead of generating code as it is
Other consequence of the inherent complexity of the case in general, they will generate configura-
ERPs is that most of the clients trying to imple- tion tables that would activate the required ERP
ment them do not have a deep understanding of modules (Dugerdil & Gaillard, 2006).
700
MODEL-DrIVEN bUsINEss IT systems can be refined or developed. An IT

INtEGrAtION system that does not totally support the business
model is inadequate. The creation of UML exten-
Business process integration and management sion for business modeling (Ericsson & Penker,
(BPIM) is an important element in the continu- 2000) creates a homogenous environment where
ous efforts enterprises undertake to adjust and the business model and the supporting software
transform their business to respond to many chal- model can be designed and developed using the
lenges of the competition. BPIM solutions should same formalism. Therefore, there is hope this new
be efficient, fast, reusable, robust and low-cost. environment will help eliminating discrepancies
Initially integration solutions were focused that could exist between the business model and
on connecting systems. The approach used was the software model. Any change occurring in
to provide various Enterprise Application Inte- the business model will be propagated down to
gration (EAI) adapters that would establish an the IT system. We believe that the model-driven
ad hoc point-to-point connection (Zhu, Tian, T., approach is the key to solving above-mentioned
Li, Sun & al, 2004). Later, the same problem challenges.
was addressed by creating a hub or a centralized
integration engine hosting the integration logic. Industrial Use
Such hubs are IBM WebSphere Business Integra-
tion (WBI), Microsoft BizTalk Server and BEA MDA is making a constant progress in the soft-
Weblogic Integrator to name a few. This solution ware industry. Away from any hype, MDA gives
facilitates the creation, maintenance, and chang- companies a viable alternative to application
ing of integration logic and provides a better development instead of corporate stagnation or
environment managing change in a more flexible offshore (Pastor & Juan Carlos Molina, 2007). Not
and efficient way. only the number of MDA-compliant vendors is
Although the creation of hubs facilitated the increasing over the time but most importantly, the
overall communication among different enter- number of companies that are using this approach
prise modules, it did not address the heart of the for developing large scale enterprise information
problem: is how to design, develop, maintain, systems. Following are some examples of success
and utilize the integration hub for real business in using the model driven architecture. A complete
cases (Zhu, Tian, T., Li, Sun & al, 2004). Exist- list of companies that have this approach suc-
ing solutions were very much based on trying cessfully can be found in (Object Management
adjusting exiting IT systems to address change Group, 2000).
in enterprise business. Daimler-Chrysler has used the MDA approach
Currently, there is a widely-accepted opinion, for developing its information system. The results
(Frankel, 2003; Johnston & Brown, 2005; Kleppe, of this experience include (Object Management
Warmer & Blast, 2003; Pastor & Juan Carlos Mo- Group, 2000):
lina, 2007; Sewal, 2003; Thomas, 2002) to name
a few, that today’s business integration solutions • 15% increase in development productivity
must be elevated to the design and analysis of in first year,
high-level business strategy and processes be- • ROI (Return Of Investment) in less than 12
fore any implementation by the IT department. months,
Therefore, creating a business model representing • Expected total productivity increase of
the goals and objectives of the enterprise must 30% in second year compared to a non-
come first and must determine the way in which MDA approach
701
Lockheed Martin Aeronautics has used the and developers translate models into a program-
OMG’s MDA to develop the F-16 Modular Mis- ming language. The use of MDA narrows the gap
sion Computer Application Software. Their goal existing between developers and modelers and
was to achieve cross-platform compatibility and creates the need for a modeler/developer of high
increased productivity and quality, all in the level. Modeling is a more abstract activity than
context of the demanding environment of avion- coding and therefore, while the market is full of
ics software development (Object Management developers in all major technologies (Java, .NET,
Group, 2000). etc) there is great deficit in high level modelers
Conquest, Inc., is a premier provider of proficient in UML and its various profiles. Train-
advanced large-scale systems and software ing developers to become modelers is a necessity
technology solutions to federal and commercial considering the important savings provided by the
customers. MDA helped facilitate communication use of the MDA approach (Figure 17). The MDA
by graphically representing vast amounts of data approach offers automation of code generation
into discrete views that could be reviewed and that includes:
understood across different organizational groups
and business areas. Modeling also encouraged col- • Automatically generating complete pro-
laboration between groups, helping them identify grams from models (not just class/methods
redundant and non-mission enhancing activities, skeletons and fragments)
and driving significant cost savings (Object Man- • Automatically verifying models at high
agement Group, 2000). A more detailed study that level of abstraction (for example, by ex-
provides reasons and advantages of using the MDA ecuting them)
approach is presented in (Pastor & Juan Carlos
Molina, 2007). Figure 17, borrowed from (Sewal, History teaches us that automation is by far the
2003), presents a convincing list of savings when most effective technological means for boosting
using the MDA approach. productivity and reliability. It will take time and
Currently, the use of the MDA approach is efforts for the MDA to widely be accepted and
rather limited; there is a relatively small number only the future will decide when the MDA will
of developers that master this technology at indus- become a mainstream technology.
trial level. Historically in the software industry,
there is a well defined gap between developers
and modelers/architects and they have very dif-
ferent tasks in the software development process.
Modelers design the system and its architecture
Figure 17. Savings from using the MDA approach in industry
702
cONcLUsION better environment to implement the same business

model in a newer underlying technology. As an
This paper describes the concept and the feasibility example, the switch from a traditional networking
of using MDA-based tools for designing, devel- environment into a wireless environment comes
oping and implementing Enterprise Information without the efforts and pains caused by the process
Systems using the MDA approach. The center of of rewriting the business model using another
the approach is a conceptual model that expresses computing platform. Important efforts are directed
concepts from the domain problem and their re- to use the MDA approach modeling the business
lationships. The conceptual model is built using model when the target platform is an ERP.
UML, a standard in the software industry. The MDA allows for a better and faster update
with the latest achievements in software engineer-
model is developed visually and the language used
ing techniques, as once new design patterns are
is simple and understandable to programmers and
invented, they will be implemented in the MDA
non-programmers alike and therefore it facilitates
tools by venders and therefore the quality of the
the dialog with stakeholders.
code obtained at the end of the process will be the
This modeling paradigm is specialist-centric,
same independently of the qualifications of team
as it allows for a greater participation of specialist of developers.
in model construction. The model is constructed As MDA is a relatively new modeling para-
conceptually, focusing on the business logic that digm, it is not well-known in the community of
is familiar to the specialists. Thus, the intellectual software developers. There is a substantial lack of
investment spent for the design of the business qualified modelers able to apply at large scale this
model is preserved as it is not affected by the new modeling paradigm.
changes of underlying technologies. Currently, there is a wide variety of commercial
The conceptual model is constructed at a tools available pretending to apply MDA prin-
higher level of abstraction without considering ciples as they are defined by OMG. Some allow
any implementation or computing platform issues. users to choose the implementation environment
Therefore, the model is platform independent; it (Pastor & Juan Carlos Molina, 2007) and some
can be implemented in different programming en-
have a defined implementation technology (Visual
vironments and computing platforms. A platform
Enterprise, 2009). Different tools provide differ-
independent model can be transformed into several
ent level of code generation and vendors of these
platform specific models that take into consideration
tools claim to be MDA-compliant. There is no
different computing platforms.
any formalism for checking the compliance with
Once the business logic of the problem is clari-
fied and expressed in the model using a high level OMG principles. Therefore, there is need to have
of abstraction, then implementation issues can be better collaboration and better standards for what
addressed. Implementation details are applied to would be an MDA-compliant product.
the general model by a set of transformations. The
model obtained considers a particular implementa-
tion environment and therefore is specific to the rEFErENcEs
selected computing platform. A platform specific
model contains all the necessary details so that Booch, G. (1999). The unified modeling language
code can be generated automatically. Code can be user guide. Reading, MA: Addison Wesley.
generated in a number of programming environ- Booch, G., Rumbaugh, J., & Jacobson, I. (1999).
ments such as Java, C#, .NET etc. The unified modeling language user guide. Read-
MDA separates the business model form the
ing, MA: Addison Wesley.
implementation technologies and thus, it creates a
703
Brown, W. A., Conallen, J., & Tropeano, D. (2005). Johnston, S. (2004). Rational UML profile for busi-
Practical insights into MDA: Lessons from the ness modeling. IBM developerworks. Retrieved
design and use of an MDA toolkit . In Beydeda, from http://www.ibm.com/developerworks/ratio-
S., Book, M., & Gryhn, V. (Eds.), Model-driven nal/library/5167.html.
software development (pp. 403–432). New York:
Kennedy Carter Inc. (n.d.). Retrieved 2009, from
Springer. doi:10.1007/3-540-28554-7_18
Kennedy Carter Web site http://www.kc.com
Dugerdil, P., & Gaillard, G. (2006). Model-driven
Kleppe, A., Warmer, J., & Blast, W. (2003). MDA
ERP implementation. In Proceedings of the 2nd
explained the model driven architecture: Practice
international workshop on model-driven enter-
and promise. Reading, MA: Addison Wesley.
prise information systems.
McNeile, A., & Simons, N. (2003). MDA the
Ericsson, E. H., & Penker, M. (2000). Business
visison with the hole. White Paper, Metamaxim
modeling with UML business patterns at work.
Ltd. Retrieved from http://www.metamaxim.com.
Needham, MA: OMGPress.
McNeile, A., & Simons, N. (2004). Methods of
Frankel, D. S. (2003). Model driven architecture
behaviour modelling a comentary on behaviour
applying MDA to enterprise computing. New York:
modelling techniques for MDA. White Paper,
Wiley Publishing Inc. OMG Press.
Metamaxim Ltd. Retrieved from http://www.
Gasevic, D., Djuric, D., & Devedzic, D. (2006). metamaxim.com
Model driven architecture and ontology develop-
Mellor, S. J., & Balcer, M. J. (2002). Executable
ment. Berlin: Springer.
UML, a foundation for model driven architecture.
Guide, M. D. A. V1.0.1. (n.d.). Retrieved 2003, Reading, MA: Addison Wesley.
from Object Management Group Web site http://
Meyer, B. (1988). Object-oriented software con-
www.omg.org
struction. Englewood Cliffs, NJ: Prentice Hall.
How systems will be built. (n.d.). Retrieved 2009,
Meyer, B. (1992). Eiffel: The language. Upper
from Object Management Group Web site http://
Saddle River, NJ: Prentice Hall.
omg.org
(n.d.). Retrieved 2003, from MDA Guide V1.0.1
Johnston, J. K., & Brown, A. W. (2005). A model
Web site: http://www.omg.org
driven development approach to creating service-
oriented solutions . In Model-driven software Object Management Group. (n.d.). Retrieved
development. New York: Springer. 2009, from Object Management Group Web site:
http://omg.org
Johnston, J. K., & Brown, A. W. (2005). A model
driven development approach to creating service- Object management group [OMG]. (2009). Re-
oriented solutions . In Model-driven software trieved from http://omg.org
development (pp. 624–636). New York: Springer.
OMG Model driven architecture: How systems will
Johnston, J. K., & Brown, A. W. (2005). Model- be built. (2009). Retrieved from http://omg.org
driven development approach to creating service-
OMG/MDA. (2009). How systems will be built.
oriented solutions . In Beydeda, S., Book, M., &
Retrieved 2009, from Object Management Group
Gyhn, V. (Eds.), Model-driven software develop-
Web site http://omg.org
ment. New York: Springer.
704
Oscar, P., & Juan Carlos Molina. (2007). Model- Sewal, S. J. (2003). Executive justification for
driven architecture in practice. Germany: adopting model driven architecture. Retrieved
from http://omg.org/mda/presentations.html
Papajorgji, P., Clark, R., & Jallas, E. (2009). The
Model driven architecture approach: A framework Thomas, L. J. (2002). ERP et progiciel de gestion
for developing complex agricultural Systems. In integres. Paris: Dunod.
P. Papajorgji & P. M. Pardalos (Eds.), Advances
van Everdingen, Y., van Hillergersberg, J., &
in modeling agricultural systems. New York:
Waarts, J. (2000). ERP adoption bu european
Springer. (Springer Optimization and Its Ap-
midsize companies. Communications of the ACM,
plications).
43(4), 27–31. doi:10.1145/332051.332064
Papajorgji, P., & Pardalos, P. M. (2006). Software
Visual Enterprise. (2009). Retrieved 2009, from
engineering techniques applied to agricultural
http://intelliun.com
systems an object-oriented and UML approach.
New York: Springer. Warmer, J., & Kleppe, A. (1999). The object
constraint language precise modeling with UML.
Papajorgji, P., & Shatar, T. (2004). Using the
Reading, MA: Addison Wesley.
unified modeling language to develop soil water-
balance and irrigation-scheduling models. Envi- Zhu, J., Tian, Z., T., Li, Sun, W., & al, e. (2004).
ronmental Modelling & Software, 19, 451–459. Model driven business process integration and
doi:10.1016/S1364-8152(03)00160-9 management. A case study with bank sinoPac
regional platform. IBM Journal of Research and
Pastor, O., & Molina, J. C. (2007). Model-driven
development.
architecture in practice. Germany: Springer
Verlag.
705
706
Chapter 3.10
Impact of Portal
Technologies on Executive
Information Systems
Udo Averweg
Information Services, eThekwini Municipality & University of KwaZulu-Natal, South Africa
Geoff Erwin
Cape Peninsula University of Technology, South Africa
Don Petkov
Eastern Connecticut State University, USA
INtrODUctION technologies integrate information, content, and

enterprise applications. However, the term portal
Internet portals may be seen as Web sites which has been applied to systems that differ widely
provide the gateway to corporate information in capabilities and complexity (Smith, 2004). A
from a single point of access. Leveraging knowl- portal aims to establish a community of users with
edge—both internal and external—is the key to a common interest or need.
using a portal as a centralised database of best Portals include horizontal applications such
practices that can be applied across all departments as search, classification, content management,
and all lines of business within an organisation business intelligence (BI), executive information
(Zimmerman, 2003). The potential of the Web systems (EIS), and a myriad of other technolo-
portal market and its technology has inspired gies. Portals not only pull these together but are
the mutation of search engines (for example, Ya- also absorbing much of the functionality from
hoo®) and the establishment of new vendors in these complementary technologies (Drakos,
that area (for example, Hummingbird® and Brio 2003). When paired with other technologies,
Technology®). A portal is simply a single, distilled such as content management, collaboration, and
view of information from various sources. Portal BI, portals can improve business processes and
boost efficiency within and across organisations
DOI: 10.4018/978-1-59140-993-9.ch030 (Zimmerman, 2003). This chapter investigates the
Impact of Portal Technologies on Executive Information Systems
level of impact (if any) of portal technologies on function with Web servers and browsers (Price-
EIS. It proceeds with an overview of these tech- waterhouseCoopers, 2002). With EIS established
nologies, analysis of a survey on the impact of in organisations and the presence of portal tech-
Web-based technologies on EIS implementation, nologies, there is thus a need to investigate the
and conclusions on future trends related to them. link (if any) between EIS and portal technologies.
Web-based technologies are causing a reex-
amination of existing information technology (IT)
bAcKGrOUND ON POrtAL implementation models, including EIS (Averweg,
tEcHNOLOGIEs AND EIs 2003). Web-based tools “are very much suited” to
executives key activities of communicating and
Gartner defines a portal as “access to and interac- informing (Pijpers, 2001). With the emergence of
tion with relevant information assets (information/ global IT, existing paradigms are being altered,
content, applications and business processes), which is spawning new considerations for suc-
knowledge assets and human assets, by select cessful IT implementation. Challenges exist in
target audiences, delivered in a highly personalized building enterprise portals as a new principle of
manner” (Drakos, 2003). Drakos (2003) suggests software engineering (Hazra, 2002). Yahoo® is an
that a significant convergence is occurring with example of a general portal. Yahoo® enables the
portals in the centre. Most organisations are be- user to maintain a measure of mastery over a vast
ing forced to revisit their enterprise-wide Web amount of information (PricewaterhouseCoopers,
integration strategies (Hazra, 2002). A single view 2001). Portals are an evolutionary offshoot of the
of enterprise-wide information is respected and Web (Norwood-Young, 2003). The Web is “a
treasured (Norwood-Young, 2003). Enterprise perfect medium” for deploying decision support
Information Portals are becoming the primary way and EIS capabilities on a global basis (Turban et
in which organisations organise and disseminate al., 1999).
knowledge (PricewaterhouseCoopers, 2001). As the usage of IT increases, Web-based tech-
EIS grew out of the development of informa- nologies can provide the means for greater access
tion systems (IS) to be used directly by executives to information from disparate computer applica-
and used to augment the supply of information by tions and other information resources (Eder, 2000).
subordinates (Srivihok, 1998). For the purposes of Some Web-based technologies include: intranet,
this article, an Executive Information System is Internet, extranet, e-commerce business-to-busi-
defined as “a computerized system that provides ness (B2B), e-commerce business-to-consumer
executives with easy access to internal and external (B2C), wireless application protocol (WAP), and
information that is relevant to their critical success other mobile and portal technologies. The portal
factors” (Watson et al., 1997). EIS are an impor- has become the most-desired user interface in
tant element of the information architecture of an Global 2000 enterprises (Drakos, 2003).
organisation. Different EIS software tools and/or
enterprise resource planning (ERP) software with
EIS features exist. sUrVEy OF WEb-bAsED
EIS is a technology that is emerging in response tEcHNOLOGIEs’ IMPAct ON EIs
to managers’ specific decision-making needs
(Turban et al., 1999). Turban (2001) suggests that The technology for EIS is evolving rapidly and
EIS capabilities are being “embedded in BI.” All future systems are likely to be different (Sprague
major EIS and information product vendors now & Watson, 1996). EIS is now clearly in a state of
offer Web versions of their tools, designed to flux. As Turban (2001) notes, “EIS is going through
707
a major change.” There is therefore both scope sample selected included organisations with actual
and need for research in the particular area of EIS EIS experience with representatives from the fol-
being impacted by portal technologies as execu- lowing three constituencies: (1) EIS executives/
tives need systems that provide access to diverse business end-users; (2) EIS providers; and (3) EIS
types of information. Emerging (Web-based) tech- vendors or consultants. These three constituen-
nologies can redefine the utility, desirability, and cies were identified and used in EIS research by
economic viability of EIS technology (Volonino, Rainer and Watson (1995). A formal extensive
et al., 1995). There exists a high degree of similar- interview schedule was compiled and used for the
ity between the characteristics of a “good EIS” semi-structured interviews. Those were conducted
and Web-based technologies (Tang et al., 1997). during May-June 2002 at organisations in the large
With the absence of research efforts on the impact eThekwini Municipal Area (EMA) in the eastern
of portal technologies on EIS implementations, part of South Africa, including Durban, which is
this research begins to fill the gap with a study of the most populous municipality in the country,
thirty-one selected organisations in South Africa with a geographic area size of 2,300 km2 and
which have implemented EIS. a population of 3.09 million citizens (Statistics
A validated survey instrument was developed South Africa, 2001).
and contained seven-point Likert scale statements The number of surveyed interviewees and
(anchored with (1) Not at all and (7) Extensively) associated percentages per constituency for the
dealing with how an interviewee perceives specific three EIS constituencies are reflected in Table 1.
Web-based technologies impacted his organisa- The respondents in the organisations surveyed
tion’s EIS implementation. The selected Web- reported a wide range of available commercially
based technologies were: (1) intranet; (2) Internet; purchased EIS software tools and/or ERP software
(3) extranet; (4) e-commerce: business-to-business with EIS features. These included Cognos®,
(B2B); (5) e-commerce: business-to-consumer JDEdwards BI®, Oracle®, Hyperion®, Lotus
(B2C); (6) wireless application protocol (WAP) Notes®, Business Objects®, and Pilot®. Cognos®
and other mobile technologies; and (7) any other was the most popular EIS software tool, compris-
Web-based technologies (for example, portal ing 60% of the sample surveyed.
technologies). The questionnaire was adminis- From the data gathered through the authors’
tered during a semi-structured interview process. survey instrument, a tally and associated percent-
A similar approach was adopted by Roldán and age of the perceived degree to which specific Web-
Leal (2003) in their EIS survey in Spain. based technologies impacted the respondent’s
The sample was selected using the unbiased current EIS implementation in the organisations
“snowball” sampling technique. This technique surveyed is reflected in Table 2. Table 2 shows that
was also used by Roldán and Leal (2003). The only seven (22.5%) organisations surveyed report
Table 1. EIS constituencies and number of interviewees surveyed per constituency
Number of interviewees surveyed and

Stakeholder groups (constituencies)
associated percentage of total sample
EIS executives/business end-users 20 (64.5%)
EIS providers 7 (22.6%)
EIS vendors or consultants 4 (12.9%)
SAMPLE SIZE 31 (100%)
708
that the Intranet significantly impacted their EIS surveyed. Only one respondent (3.2%) reported
implementation. Intranets are usually combined that WAP and other technologies had extensively
with and accessed via a corporate portal (Turban, impacted the EIS implementation in her organisa-
et al., 2005). The level of impact by the Internet tion. A possible explanation for this result is that
on EIS implementation is slightly lower with six the EIS consultant was technically proficient
(19.4%) organisations surveyed reporting that in WAP technologies. The potential benefits of
the Internet has significantly impacted their EIS mobile access to portals are numerous and self-
implementation. While 24 (77.4%) organisations evident. PricewaterhouseCoopers (2002) notes
surveyed report that the Extranet had no impact that organisations must first establish the benefits
on their organisation’s EIS implementation, the of mobile access to its portal and assess the value
balance of the data sample (22.6%) reports dif- of providing those benefits via mobile access to
ferent degrees of impact. the organisation.
The results in Table 2 show that the vast ma- According to Table 2, three interviewees re-
jority (90.4%) of respondents reports that e- ported that their organisation’s EIS implementa-
commerce: (B2B) has not impacted EIS imple- tions were significantly impacted (“Very much”
mentation in organisations surveyed. A slightly and “Extensively”) by portal technologies. At
lower result (83.9%) was reported for e-commerce: first this may appear to be noteworthy as the
(B2C). One possible explanation for the e-com- portal technology impact on EIS implementations
merce (B2B) and (B2C) low impact levels is that (9.7%) is higher than that on the extranet (6.5%),
the software development tools are still evolving e-commerce: (B2B) (6.4%), e-commerce: (B2C)
and changing rapidly. (6.4%), and WAP and other technologies (3.2%)
WAP and other mobile technologies have impacts. However, it should be noted that the
no (93.6%) or very little (3.2%) impact on EIS impact levels of all the Web-based technologies
implementations. Of the seven Web-based tech- assessed are fairly low. This still means that after
nologies given in Table 2, WAP and other mobile the Intranet and Internet, portal technologies have
technologies have the least impact (combining the third highest impact on EIS implementa-
“Somewhat much,” “Very much,” and “Exten- tions in organisations surveyed. Combining the
sively”) on EIS implementation in organisations results (“Somewhat much,” “Very much,” and
Table 2. Tally and associated percentage of the degree to which specific Web-based technologies impacted
respondent’s current EIS implementation
The degree to which Web-based technologies impacted respondent’s

EIS implementation (N=31)
Very Somewhat Somewhat Very
Web-based technology Not at all Uncertain Extensively
little little much much
Intranet 17 (54.8%) 2 (6.5%) 2 (6.5%) 0 (0.0%) 3 (9.7%) 4 (12.9%) 3 (9.6%)
Internet 21(67.7%) 1 (3.2%) 1 (3.2%) 0 (0.0%) 2 (6.5%) 3 (9.7%) 3 (9.7%)
Extranet 24 (77.4%) 1 (3.2%) 2 (6.5%) 1 (3.2%) 1 (3.2%) 2 (6.5%) 0 (0.0%)
E-Commerce: (B2B) 28 (90.4%) 1 (3.2%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 1 (3.2%) 1 (3.2%)
E-Commerce: (B2C) 26 (83.9%) 1 (3.2%) 1 (3.2%) 0 (0.0%) 2 (6.5%) 0 (0.0%) 1 (3.2%)
WAP and other mobile technologies 29 (93.6%) 1 (3.2%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 1 (3.2%)
Portal technologies 26 (83.8%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 2 (6.5%) 2 (6.5%) 1 (3.2%)
709
Table 3. Descending rank order of impact levels of Web-based technologies on current EIS implementation
Rank Web-based technology Tally and level of impact on EIS implementations (N=31)
1 Intranet 10 (32.2%)
2 Internet 8 (25.9%)
3 Portal technologies 5 (16.2%)
4 Extranet 3 (9.7%)
4 E-Commerce: (B2C) 3 (9.7%)
6 E-Commerce: (B2B) 2 (6.4%)
7 WAP and other mobile technologies 1 (3.2%)
“Extensively”) for each of the seven Web-based their future EIS implementations, eight (25.8%)
technologies, Table 3 gives a descending ranking respondents were unsure of future impact levels
order of the levels of impact of different Web-based by the extranet. Twelve (38.7%) respondents
technologies on EIS implementations. indicated that it is unlikely that e-commerce:
A tally and associated percentage of the per- (B2B) will impact future EIS implementations.
ceived degree to which specific Web-based tech- Almost half (48.4%) of organisations surveyed
nologies will impact a respondent’s future EIS reported that it is unlikely that e-commerce: (B2C)
implementation is given in Table 4. These are will impact future EIS implementations. WAP
obtained from the data gathered using the authors’ and other mobile technologies have similar
survey instrument. (42.0%) unlikely future levels of impact.
Table 4 reflects that only two (6.4%) organisa- It is striking to note that 21 (67.7%) respon-
tions surveyed reported that it is unlikely that the dents indicated that it is Extremely unlikely that
Intranet will impact future EIS implementations. portal technologies will impact future EIS imple-
The unlikeliness of impact by the Internet on mentations. This result (when combined with the
future EIS implementations is somewhat higher “Slightly unlikely” and “Quite unlikely” degrees)
(16.1%). While seven (22.6%) respondents indi- rises to 24 (75.2%) organisations surveyed. This
cated that is unlikely that the Extranet will impact finding is somewhat surprising considering that
Table 4. Tally and associated percentage of the expected degree to which specific Web-based technolo-
gies will impact respondent’s future EIS implementations
The expected degree to which Web-based technologies will impact respondent’s

future EIS implementations (N=31)
Extremely Quite Slightly Slightly Quite Extremely
Web-based technology Uncertain
likely likely likely unlikely unlikely unlikely
Intranet 17 (54.8%) 7 (22.6%) 3 (9.7%) 2 (6.5%) 0 (0.0%) 1 (3.2%) 1 (3.2%)
Internet 12 (38.8%) 6 (19.3%) 5 (16.1%) 3 (9.7%) 1 (3.2%) 1 (3.2%) 3 (9.7%)
Extranet 6 (19.3%) 7 (22.6%) 3 (9.7%) 8 (25.8%) 0 (0.0%) 1 (3.2%) 6 (19.4%)
E-Commerce: (B2B) 3 (9.7%) 9 (29.0%) 4 (12.9%) 3 (9.7%) 2 (6.5%) 4 (12.9%) 6 (19.3%)
E-Commerce: (B2C) 2 (6.5%) 9 (29.0%) 4 (12.9%) 1 (3.2%) 2 (6.5%) 4 (12.9%) 9 (29.0%)
WAP and other mobile technologies 1 (3.2%) 8 (25.8%) 5 (16.1%) 4 (12.9%) 0 (0.0%) 3 (9.7%) 10 (32.3%)
Portal technologies 3 (9.7%) 2 (6.5%) 1 (3.2%) 1 (3.2%) 1 (3.2%) 2 (6.5%) 21 (67.7%)
710
portal technologies currently have the third high- levels of Web-based technologies on future EIS
est level of impact on EIS implementations in implementation matches the current rank order
organisations surveyed. An explanation for this levels of impact of Web-based technologies on
finding is that possibly some respondents are not EIS implementations (see Table 3). Second, while
aware of the existence of such technology. Roldán nearly three quarters (75.2%) of respondents
and Leal (2003) report that with the availability of surveyed report that it is unlikely that portal tech-
Web-based technologies “together with the need nologies will impact future EIS implementations
to build something similar to an EIS but focused (see Table 4), seen in the context of the other six
on all members of the organisation has led to the Web-based technologies, portals still appear in the
development of the enterprise information portal top three rankings. This is an important consid-
(EIP) concept, which, to some extent represents eration for IS practitioners when planning future
the latest incarnation of EIS.” According to Trow- EIS implementations. Third, when comparing
bridge (2000), two elements characterise these current and future impact levels of Web-based
systems according to the respondents: EIP “acts technologies on EIS, there is a positive impact
as a single point of access to internal and external trend for all Web-based technologies. The largest
information” and “gives users access to disparate trend increase is the Intranet rising from 32.2%
enterprise information systems.” to 87.1%. As Basu, et al. (2000) report, the use
According to Table 4, combining the positive of Web-based technologies in the distribution of
attitude results (“Extremely likely,” “Quite likely,” information is becoming widespread. These tech-
and “Slightly likely”) for each of the seven Web- nologies will impact future EIS implementations.
based technologies, Table 5 gives a descending The findings of this survey show that while
ranking order of the expected degree to which EIS have a significant role in organisations in
Web-based technologies will impact respondents’ a large South African metropolitan area, their
future EIS implementations. technological base is not affected considerably by
the latest innovations of Web-based technologies,
including portals. A potential limitation of the
FUtUrE trENDs AND research is the localised sample involved in the
cONcLUsION investigation, but given the highly developed IT
infrastructure of most South African companies,
We may notice three significant trends from the our findings can be cautiously generalised for most
data in Table 5. First, this rank order of impact other countries. The role of portals is to integrate
Table 5. Descending rank order of impact levels of Web-based technologies on future EIS implementation
Rank Web-based technology Tally and level of impact on future EIS implementations
1 Intranet 27 (87.1%)
2 Internet 23 (74.2%)
3 Portal technologies 16 (51.6%)
3 Extranet 16 (51.6%)
5 E-Commerce: (B2C) 15 (48.4%)
6 E-Commerce: (B2B) 14 (45.1%)
7 WAP and other mobile technologies 6 (19.4%)
711
potential information to the users. IT developers Pijpers, G. G. M. (2001). Understanding senior

must be aware of emerging trends in the portal executives’ use of information technology and
technology market to create systems that will the Internet. In Murugan Anandarajan & Claire
be able to incorporate the latest technological A. Simmers (Eds.), Managing Web usage in the
developments and new methods of information workplace: A social, ethical and legal perspective.
delivery and presentation for organisations. As the Hershey, PA: Idea Group Publishing.
use of Web-based technologies in the distribution
PricewaterhouseCoopers. (2001). Technology
of information in organisations becomes more
forecast: 2001-2003. Mobile Internet: Unleashing
widespread, it is envisaged that the impact level
the power of wireless. Menlo Park, California.
of portal technologies on future EIS implementa-
tions will increase significantly. PricewaterhouseCoopers. (2002). Technology
forecast: 2002-2004. Volume 1: Navigating the
future of software. Menlo Park, California.
rEFErENcEs
Rainer, R. K. Jr, & Watson, H. J. (1995). The
Averweg, U., Cumming, G., & Petkov, D. (2003, keys to executive information system success.
July 7-10). Development of an executive informa- Journal of Management Information Systems,
tion system in South Africa: Some exploratory 12(2), 83–98.
findings. In Proceedings of a Conference on Roldán, J. L., & Leal, A. (2003). Executive in-
Group Decision and Negotiation (GDN2003) held formation systems in Spain: A study of current
within the 5th EURO/INFORMS Joint International practices and comparative analysis. In Forgionne,
Meeting, Istanbul, Turkey, 7-10 July. Gupta, & Mora (Eds.), Decision making support
Basu, C., Poindexter, S., Drosen, J., & Addo, T. systems: Achievements and challenges for the new
(2000). Diffusion of executive information systems decade, Chapter 18, 287-304. Hershey, PA: Idea
in organizations and the shift to Web technologies. Group Publishing.
Industrial Management & Data Systems, 100(6), Smith, M. A. (2004). Portals: Toward an ap-
271–276. doi:10.1108/02635570010320484 plication framework for interoperability.
Drakos, N. (2003, August 4-6). Portalising your Communications of the ACM, 47(10), 93–97.
enterprise. Gartner Symposium ITXPO2003, doi:10.1145/1022594.1022600
Cape Town, South Africa, 4-6 August. Sprague, R. H., Jr., & Watson, H. J. (1996). Deci-
Eder, L. B. (2000). Managing healthcare infor- sion support for management. Upper Saddle River,
mation systems with Web-enabled technologies. NJ: Prentice-Hall.
Hershey, PA: Idea Group Publishing. Srivihok, A. (1998). Effective management of
Hazra, T. K. (2002, May 19-25). Building enter- executive information systems implementations:
prise portals: Principles to practice. In Proceedings A framework and a model of successful EIS imple-
of the 24th international conference on Software mentation. PhD dissertation. Central University,
Engineering, Orlando. Rockhampton, Australia.
Norwood-Young, J. (2003). The little portal that Statistics South Africa. (2001). Census 2001
could. In Wills (Ed.), Business solutions using digital census atlas. Retrieved July 5, 2006
technology platform, 1(4), 14-15. from,http://gis-data.durban.gov.za/census/index.
html [Accessed on 5 July 2006]
712
Tang, H., Lee, S., & Yen, D. (1997). An investi- KEy tErMs AND DEFINItIONs
gation on developing Web-based EIS. Journal of
CIS, 38(2), 49–54. Executive Information System: A comput-
erised system that provides executives with easy
Trowbridge, D. (2000). EIP—More profitable access to internal and external information that is
for integrators than users? Computer Technology relevant to their critical success factors.
Review, 20(5), 20. Extranet: A private Internet that connects
Turban, E. (2001). California State University, multiple organisations.
Long Beach and City University of Hong Kong, Intranet: A private Internet for an organisation.
USA. Personal Communication, 7 October. Portal: Provides access to and interaction with
relevant information assets (information/content,
Turban, E., McLean, E., & Wetherbe, J. (1999). applications and business processes), knowledge
Information technology for management. New assets, and human assets, by select target audi-
York: John Wiley & Sons. ences, delivered in a highly personalised manner.
Turban, E., Rainer, R. K., & Potter, R. E. (2005). Web-based Technologies: Technologies
Introduction to information technology (3rd Ed.). which are core to the functioning of the World
New York: John Wiley & Sons. Wide Web.
Wireless Application Protocol (WAP): A
Volonino, L., Watson, H. J., & Robinson, S. (1995). collection of standards for accessing online in-
Using EIS to respond to dynamic business condi- formation and applications from wireless devices
tions. Decision Support Systems, 14(2), 105–116. such as mobile phones, two-way radios, pagers,
doi:10.1016/0167-9236(94)00005-D and personal digital assistants.
Watson, H. J., Houdeshel, G., & Rainer, R. K., Jr. World Wide Web: The universe of network-
(1997). Building executive information systems accessible information, supported by a body of
and other decision support applications. New software, and a set of protocols and conventions
York: John Wiley & Sons. (http://www.w3.org/WWW).
Zimmerman, K. A. (2003). Portals: Not just

a one-way street. KMWorld, Creating and
Managing the Knowledge-Based Enterprise,
12(8), September. Retrieved 27 July, 2007 from
http://www.kmworld.com/Articles/PrintArticle.
aspx?ArticleID-9496
This work was previously published in Encyclopedia of Internet Technologies and Applications, edited by Mario Freire and
Manuela Pereira, pp. 215-221, copyright 2008 by Information Science Reference (an imprint of IGI Global).
713
714
Chapter 3.11
A Voice-Enabled Pervasive Web
System with Self-Optimization
Capability for Supporting
Enterprise Applications
Shuchih Ernest Chang
National Chung Hsing University, Taiwan
AbstrAct shows that multimodal user-interface pages can

be generated by using XSLT stylesheet which
Other than providing Web services through popular transforms XML documents into various formats
Web browser interfaces, pervasive computing may including XHTML, WML, and VoiceXML. It also
offer new ways of accessing Internet applications describes how VWS was designed to provide an
by utilizing various modes of interfaces to interact adaptive voice interface using an Apache Web
with their end-users, and its technology could server, a voice server, a Java servlet engine, and
involve new ways of interfacing with various a genetic algorithm based voice Web restructur-
types of gateways to back-end servers from any ing mechanism.
device, anytime, and anywhere. In this chapter,
mobile phone was used as the pervasive device
for accessing an Internet application prototype, INtrODUctION
a voice-enabled Web system (VWS), through
voice user interface technology. Today’s Web Mobile phone and Internet brought us to a new
sites are intricate but not intelligent, so finding era by offering a new way for person to person
an efficient method to assist user searching is par- communication and facilitating companies and
ticularly important. One of these efficient methods their customers in conducting business through
is to construct an adaptive Web site. This chapter electronic commerce (Gulliver, Serif & Ghinea,
2004; Toye, Sharp, Madhavapeddy & Scott, 2005;
Roussos, Marsh & Maglavera, 2005). Because
DOI: 10.4018/978-1-60566-146-9.ch008
A Voice-Enabled Pervasive Web System with Self-Optimization Capability
of the pervasive nature of empowering people According to a study from Telecom Trends
to use it anywhere and anytime, mobile phone is International, the number of mobile commerce
becoming one of the most pervasive devices in the users worldwide will grow from 94.9 million
world (Chang & Chen, 2005; Ballagas, Borchers, in 2003 to 1.67 billion in 2008, and the global
Rohs & Sheridan, 2006). With the rapid spread of revenues generated from mobile commerce are
mobile phone devices and the convergence of the expected to expand from $6.86 billion in 2003
phone and the personal digital assistant (PDA), to $554.37 billion in 2008 (de Grimaldo, 2004).
there is an increasing demand for a multimodal A report from ZDNetAsia states that more than
platform that combines the modalities of various half of 3G traffic would be voice and voice is
interface devices to reach a greater population still the platform on which our business is run
of users. While there is a growing demand for (Tan, 2005). A study reported by the Kelsey
technologies that will allow users to connect to Group claims that expenditures for speech-related
the Internet from anywhere through devices that services worldwide are expected to reach $41
are not suitable for the use of traditional key- billion by 2005 (The Kelsey Group, 2001). This
board, mouse, and monitor (Zhai, Kristensson & report also estimates a 60-65% average annual
Smith, 2005), the constraints of a typical mobile growth rate for voice services globally by 2005,
device, such as small screen size, slow speed, and with the U.S. market expected to be 20-25% of
inconvenient keyboard, make it cumbersome to this total. A recent example to the continuation
access lengthy textual information (Anerousis & of this trend can be illustrated by an outstanding
Panagos, 2002). In Taiwan, the penetration rate of growth (350 percent increase in quarterly revenue)
mobile phone (104.6%)1 is much higher than the of speech self-service marketplace reported by
penetration rates of other major telecom services, Voxify, Inc. (Market Wire, 2006). It is believed
including local telephone: 58.2%, Internet: 71.3%, that the demand for mobile commerce has created
and broadband Internet: 68.7% (Institute for a market for voice-enabled applications accessible
Information Industry, 2007). However, the same by mobile phone.
survey also shows that the utilization rate of ac- Traditionally, Interactive Voice Response
cessing Internet from wireless devices is relatively (IVR) systems are based on proprietary hardware
low, with a penetration rate slightly lower than and software technology, with development and
50%, mainly because the text-based interaction deployment tightly integrated on the same hard-
between mobile devices and Web sites is very ware platform (Turner, 2004). This has resulted in
limited. However, voice interface does not have high development costs. Non-portable proprietary
these limitations, because voice interaction could software cannot be deployed on different platforms
escape the physical limitations on keypads and and it is also inherently difficult to upgrade or
displays as mobile devices become ever smaller modify (Dettmer, 2003). A multi-modal language
and it is much easier to say a few words than it is needed to support human-computer dialogs via
is to thumb them in on a keypad where multiple spoken input and audio output. As an optimum
key presses may be needed for entering each letter solution, VoiceXML (Voice eXtensible Markup
or character (Rebman, Aiken & Cegielski, 2003). Language), a markup language for creating voice-
Using voice as a medium to operate mobile devices user interfaces, bridges the gap between the Web
also enables user’s hands to engage in some other and the speech world by utilizing speech and
activities without losing the ability to browse the telephone touchtone recognition for input and
Internet through voice commands (Feng, Sears & prerecorded audio and text-to-speech synthesis
Karat, 2006). (TTS) for output (Larson, 2003). It is based on
the World Wide Web Consortium’s (W3C’s) eX-
715
tensible Markup Language (XML) and leverages concern for users. A voice Web site may have
the Web paradigm for application development hundreds of pages. When users visit the pages by
and deployment. By having a common language, a phone, they only depend on the voice to navigate
application developers, platform vendors, and tool the site. Users must wait for the voice menus by
providers can all benefit from code portability and sequences, so they spend relatively longer time
reuse. Furthermore, to reduce the cost of build- using the voice channel than a Web browser in
ing and delivery of new capabilities to telephone waiting for and/or selecting their desired choices.
customers, providing voice access to Web-based Unduly designed VUIs will be inefficient and fail
applications is an attractive option. VoiceXML to serve users’ needs for promptness. It not only
makes it possible for companies to write shared causes users’ dissatisfaction, but also introduces
business logic once and focus their resources on unnecessary inefficiency and redundancy to the
developing only the specific user interface for system (especially for the voice recognition and
each device they support. synthesis components). Thus, it is necessary to
Due to the above mentioned facts and analyses, design a voice user interface optimization mecha-
a voice-enabled web system (VWS), utilizing the nism in the voice-enabled Web system (VWS)
voice user interface technology, was designed application. The proposed VWS system would
and implemented through our project conducted apply genetic algorithms (GAs) to find a reason-
in Taiwan. Voice mobile phone was chosen as ably good arrangement of the site map in a voice
the pervasive device for accessing our Internet Web site, so the restructured site map will make
application prototype, a VWS-based service, users’voice browsing experience more responsive.
for two reasons. Firstly, as mentioned earlier the This GA-based VUI optimization mechanism was
penetration rate of mobile phone is much higher implemented as a software application so that it
than the rates of other major telecom services, and can be integrated with the VWS system.
mobile phone is associated tightly with people’s The subsequent sections of this chapter are
daily life. Secondly, the use of speech for input and organized as follows. Section 2 provides the
output is inherent in the minds of mobile phone backgrounds on voice-enabled Web system,
users. The system implemented in this research adaptive Web site, and genetic algorithm. Section
has several advantages over systems using other 3 describes the architecture of the VWS system
mobile devices such as Palm PDA, BlackBerry, and how to generate multimodal user-interface
and Pocket PC. For example, VWS users can obtain pages by using XSLT stylesheet. To illustrate how
information through voice instead of looking at the GA-based VUI optimization process works,
the monitor, and VWS eliminates the requirement the experiment method together with two simple
of keyboard or mouse through the use of voice examples was presented in Section 4. Section 5
user interface. covers more comprehensive experiment results,
Voice channels which differ from the Web and Section 6 concludes this chapter after the
have some limitations. For instance, the greatest discussions.
limitation is that voice channels only support one
type of access, acoustic, from a phone. Tomorrow’s
voice Web sites will serve up voice user interfaces LItErAtUrE rEVIEW
(VUIs) alongside the graphical user interfaces
(GUIs) they serve up today (Teo & Pok, 2003). Voice-Enabled Web system
The primary objective for VUIs must be creating
a positive experience for the users. In addition, A voice-enabled Web system is a system which
getting responses promptly is also an important provides users a voice channel, such as telephone,
716
to access Web applications. With voice-enabled its applications related to the users’ activities of
Web systems, firms can provide desirable voice- navigating voice portals with voice commands.
based Internet services, such as online customer V-commerce examples include the use of speech
service, online transaction service, and self-served technology over the telephone in commercial ap-
service, through both the conventional browser plications such as buying cinema/airline tickets,
interface and the new voice interface. Our voice- banking, account transferring, stock trading, and
enabled Web system combines XML based purchasing from mail-order companies (Goose,
mark-up languages, automatic speech recogni- Newman, Schmidt & Hue, 2000;Yamazaki,
tion (ASR), text to speech (TTS), and Web tech- Iwamida & Watanabe, 2004).
nologies. We use the emerging standard markup
language, VoiceXML, which defines a common Adaptive Web site
format to allow people to access Web content via
any phone (Larson, 2003). The VoiceXML uses Web users often get lost on the Internet due to its
XML tags to represent call flows and dialogs. The complicated structure and the information over-
development of the VoiceXML standard by AT&T, load problem. One of the most important func-
IBM, Lucent Technologies, and Motorola has led tions of a Web site is to assist users in searching
to a proliferation in recent years of voice-enabled information by using various Web intelligence
Web systems. By using this standard Web-based methods (Li & Zhong, 2004). One of those ef-
language, data then can be easily exchanged in ficient methods is to construct an adaptive Web
voice-enabled Web systems. site. Adaptive Web sites are sites that automati-
Voice-enabled Web technology is being de- cally improve their organization and presentation
ployed in a broad range of industries such as by learning from visitor access patterns (Kohrs
banking and retailing. With the launch of the & Merialdo, 2001).
first “voice portal”, which provides telephone Joachims, Freitag, and Mitchell (1997) initi-
users with speech-enabled access (via the nature ated an adaptive Web project called WebWatcher,
language interface) to Web-based information and which is a “tour guide” agent for the World Wide
applications, the voice-enabled Web technology Web, and its strategy for giving advice is learned
caught people’s attention. It is speculated that from feedback from earlier tours. WebWatcher
various industries will soon adopt it to develop uses the paths of people who indicated success
suitable Web system to serve their own business as examples of successful navigations. It groups
purposes. Internet portal companies such as AOL people based on their stated interests rather than
and Yahoo and other companies like Tellme Net- customizing to each individual. Perkowitz and
works, Hey Anita, and Internet Speech have been Etzioni (2000) focused on the problem of index
developing voice portals for providing several page synthesis. An index page is a page consisting
services. Generally speaking, a voice portal, like of links to a set of pages that cover a particular
an Internet portal, is a single place where content topic at a site. Their goal is to transform the Web
from a number of sources is aggregated. For ex- site into a better one – a new index page. They
ample, information services such as traffic reports, assume these groups of pages represent coherent
weather reports, stock quotes, bill inquiry/pay, topics in users’ minds, and analyze the Website’s
restaurant/hotel recommendations, department access logs to find groups of pages that often oc-
store promotion information, cinema reviews, cur together in user visits. Su, Yang, Zhang, Xu,
news, and e-mail can be accessed via voice portal. Hu, and Ma (2002) designed an adaptive Web
Recently, an emerging term called “v-commerce” interface based on Web log analysis and Web
has been used to describe the technology and page clustering. They also tried to improve users’
717
performance by introducing index pages that mini- (i.e., the Web site structure). The process of our
mize overall user browsing costs. Smith and Ng voice user interface optimization is illustrated
(2003) presented LOGSOM, a prototype system in Figure 1. According to the needs specified in
that organizes Web pages on a self-organizing a time-based or event-driven configuration file,
map (SOM) according to user navigation patterns. VWS can extract its site map and invoke the VUI
They clustered the Web pages according to the optimization mechanism to derive a restructured
users’ navigation behaviors, rather than accord- site map, which is used to substitute the original
ing to the Web content. Instead of organizing the one for providing users a better browsing expe-
Web pages according to the words contained in rience with a more efficient and more effective
the Web pages, they kept track of the interest voice navigation paths.
of the Web users, and organized the Web pages
according to their interest. In this way, the SOM Genetic Algorithm: basic concepts
provided by LOGSOM can be updated regularly
to reflect the current interest of the Web users. Genetic algorithms (GAs), which use randomized
In addition, for the purpose of personalization or search and optimization techniques, are designed
recommendation, many different kinds of adaptive to simulate processes in natural system necessary
Web sites have been explored recently. A simple for evolution, that is, the solutions to a problem
but common example is that some Web sites al- solved by GAs are derived through an evolution-
low users to personalize the sites for themselves ary process, which is based on a mechanism of
such as customizing the lists of favorite links. natural selection to search for an improved solution
Some other more complicated approaches may that optimizes a given fitness function. As shown
use various data mining, Web mining, content- in Figure 2, GA begins with a set of randomly
based filtering, collaborate filtering, and other created solutions called population. Pairs of solu-
techniques to offer users personalized or adapted tions are taken and used to produce offspring of
information and services (Chang, Changchien next generation, that is, a new population. This is
& Huang, 2006; Changchien, Lee & Hsu, 2004; motivated by a hope, that the new population will
Wang & Shao, 2004). be better than the old one. In the selection stage,
As mentioned in the previous subsection, a parent solutions (which are selected to produce
voice-enabled Web system provides landline offspring solutions) are selected according to
and mobile telephone users a new channel for their fitness – the more suitable they are the more
voice-based Web browsing, which allows users
to navigate and traverse the structure of the Web
site entirely by voice. In terms of voice browsing Figure 1. The process of voice user interface
or navigating the VUI structure of voice-enabled optimization
Web sites, it would be more desirable to reduce
users’ navigation time, mainly because of the
one dimensional nature of the voice channel
which causes the users to spend much time in
sequentially listening to various choices. In our
research, a simple genetic algorithm (SGA) based
voice user interface optimization mechanism,
which will be described later in this chapter, was
designed to realize an adaptive voice-enabled
Web site by automatically adapting the site map
718
chances they get to reproduce. Crossover operates leads to a more effective algorithm while compar-
on selected parent solutions to create new (off- ing it with an SGA. HGAs may allow to escape
spring) solutions. The simplest way to do that is to from a local optimization and to find better solu-
choose randomly some crossover point and copy tions. SGA is currently used in our initial imple-
everything before this point from the first parent mentation of the VUI optimization mechanism,
and then copy everything after the crossover point and for future research we plan to experiment on
from the other parent. Specific crossover made for various HGA improving approaches, such as ruin
a specific problem can improve performance of and recreate (Misevicius, 2003), tabu search
the genetic algorithm. Mutation, which randomly (Moscato & Cotta, 2003), branch and bound (Al-
modifies the genetic structures of some members Khayyal & Sherali, 2000), and simulated anneal-
of each new generation, is intended to prevent ing (Kirkpatrick, Gelatt & Vecchi, 1983).
falling of all solutions in the population into a
local optimum of the solved problem. Culling,
which takes place each time before going to the systEM ArcHItEctUrE
next generation, is for updating the population.
This iteration/revolution process is repeated until Our VWS prototype was implemented using open
an acceptable or optimum solution is found or technologies including eXtensible HyperText
until some fixed time limit. Markup Language (XHTML), XML, eXtensible
In contrast to the above described simple ge- Stylesheet Language for Transformations (XSLT),
netic algorithms (SGAs) that rely on the concept VoiceXML, MySQL database, Apache Web server,
of biological evolution, the hybrid genetic algo- Apache Tomcat application server, and various
rithms (HGAs), which are based on ideas evolu- Java APIs, such as: Java Servlet, Java Server
tion (Misevicius, 2003; Moscato & Cotta, 2003), Page (JSP), Java Database Connection (JDBC),
apply an improvement procedure to each offspring Java Cryptography Extension (JCE), and others.
or mutated individual. Instead of random solutions, Not only is Java suggested as the “write once,
HGAs operate with the improved solutions. This run everywhere” computer language in writing
application for various smart phones (Chang
& Chen, 2005), but Java’s modular nature al-
Figure 2. The procedure of a genetic algorithm
lows it to expand and develop solutions for new
computational problems. It has been evolving
from a popular client applet language to a cross
platform GUI builder and an application server
platform. This same modular nature now allows
Java to drive wireless and multimodal applica-
tions. Java 2 Platform, Micro Edition (J2ME) is
designed for nonbrowser-based devices and it is
not exactly a subset of Java 2 Platform, Standard
Edition (J2SE) (Sun Microsystems, 2004). J2ME
keeps some of the J2SE core library application
programming interfaces (APIs), but substitutes
others with lightweight components through the
javax.microedition package. As shown in Figure
3, a multimodal application architecture, which
offers new ways of accessing web applications
719
Figure 3. The multimodal approach for supporting voice-enabled web applications
from any device at any location, was adopted in this ables the exchange of self-describing information
study by utilizing various modes of interfaces to elements between computers. In addition to com-
interact with end users (Chang and Minkin, 2006). bining XML based mark-up languages, automatic
In addition to the conventional browser inter- speech recognition (ASR), text to speech (TTS),
face and the targeted voice interface, our multi- and Web technologies, our VWS also uses the
modal web system approach also provides the emerging standard markup language, VoiceXML,
ability to access web-based information and ap- which defines a common format to allow people
plications from multiple methods or channels such to access Web content via any phone (Chang &
as a PDA, smart phone, Pocket PC, or Black- Minkin, 2006). There were two options considered
Berry. This multimodal approach facilitates the in our study for enabling telephony hardware to
sharing of the business logic and back-end pro- integrate with the voice server: Intel Dialogic-
cesses in a multiple-tiered application environ- based voice server system and Cisco telephony
ment, and thus freeing up the time and resources platform. The voice server on the Dialogic platform
for concentrating on the design and implementa- utilizes a specialized telephony card manufactured
tion specifics of user interface for each device. by Dialogic, which is connected directly to the
Multimodal applications may use both wireless telephony interface. Calls are then managed by
and voice devices. It is obvious from the name of the Dialogic platform to pass incoming calls to
the platform that J2ME supports wireless tech- the Voice Server application. The other one uses
nologies such as PDA or smart phone. Many the Cisco telephony platform. The voice server
J2ME-enabled devices will support a voice chan- facilitates the deployment of voice applications
nel, and as such may also be used to interact with by interfacing with various voice standards
VoiceXML-based services over the phone voice (Rodriguez et al., 2002). The voice server for
connection. Cisco utilizes the Voice over IP (VoIP) protocol.
To build the VWS prototype, a voice server Normally the voice server would be configured to
was used as the platform that enabled the creation work with a Cisco voice router that has a telephony
of voice applications through industry standards, interface connection. When a phone call is made,
including XML, VoiceXML and Java (Burke, the voice router will convert the call to VoIP and
2001; Rodriguez, Ho, Kempny, Pedreschi, & then redirect the voice packets to the voice server.
Richards, 2002). XML facilitates the concept of The system architecture is illustrated in Figure 4.
application integration and data sharing, and en-
720
Figure 4. The architecture of the VWS system
When Voice Server starts, VoiceXML brows- text to speech (TTS) engine, which is a key com-
ers start up and wait for calls. Each VoiceXML ponent of the voice server, converts VoiceXML
browser works for one telephone call. When a scripts into speech and delivers the voice re-
user places a call to a designated phone number, sponses to the user via telephone channel. The
a computer on the voice site (i.e. the voice server) process will continue, simulating a natural lan-
answers the call and retrieves the initial VoiceX- guage conversation between the caller and the
ML script from a VoiceXML content server, which voice server.
can be a Web server located anywhere on the Web. We decided to use Java servlets to access the
An interpreter on the voice site parses and executes database, and used Java Database Connectiv-
the script by playing prompts, capturing respons- ity (JDBC) to connect to a relational database,
es, and passing the responses to a speech recogni- MySQL. The driver that we used was mysql-con-
tion engine on the voice system. Just as a Web nector-java-3.1.1-alpha-bin.jar, which is available
browser renders HyperText Markup Language for free download at http://dev.mysql.com. Java
(HTML) documents visually, a VoiceXML inter- servlets were used for validating login, construct-
preter on the voice site renders VoiceXML docu- ing user request, processing the request from end
ments audibly and allows telephone users to access user, and generating output results in the form of
services that are typically available to Web users. an XML document (Burke, 2001; Rodriguez et
Once the voice system gets all the necessary in- al., 2002). Afterwards, XSLT was used to convert
formation from the caller, the interpreter translates XML documents (generated by Java servlets) into
them into a request to the VoiceXML content XHTML documents, VoiceXML documents, and
server, i.e. the web server. When the web server WML (Wireless Markup Language) decks to suit
receives the request, it returns a VoiceXML page different devices. XSLT can be used to perform
with either a canned response or dynamically additional tasks within an application that uses
generated VoiceXML scripts, containing the in- XML as its main data representation model (Burke,
formation requested by the caller. Responses are 2001). The voice server, which contains the voice
passed from the Web server to the voice site via recognition and the synthesis engines used to au-
HyperText Transfer Protocol (HTTP). Finally, the tomate the conversation between the site and the
721
caller, is set up between the phone and the web in Asia please say TWO, for European countries
server to interpret the VoiceXML documents and please say THREE, ...,” and Joe may opt to say
act as a middleware processor. Any web site can TWO and traverse to N28 of the VUI tree shown
be a VoiceXML content server. Services provided in Figure 5. The interactions between Joe and the
by this VWS system can give subscribers access VWS service continue until Joe navigates to a leaf
to contents offered by different sources of Internet node of the VUI tree. At this point, Joe can finally
applications and services through PSTN (Public listen to the weather information of his interest.
Switched Telephone Network) telephones, wired Instead of formally describing the VUI optimiza-
or wireless. tion mechanism, we would like to show it first by
a simple example, which can easily illustrate not
only the principle and the potential improvement
EXPErIMENt MEtHOD of the VUI optimization process but also the model
FOr VUI OPtIMIZAtION used in the optimization experiments.
One of our research objectives is to find a rea- An Example of VUI Optimization

sonably good arrangement of the site map in a
voice Web site. To serve this need, a simulation In this simulation example, root node is at level 0,
approach was designed to experiment with vari- and Ni,j is the jth node at level i. Node access time
ous hierarchically structured site maps (i.e., VUI is the time length of audio heard by users during
structures or VUI trees), which were modeled by each visit to the node. Node access count is the
a tree structure as illustrated in Figure 5. When frequency users navigate through and access to
a user, Joe, uses his mobile phone and makes a the node in a specific term. Leaf node access time
phone call to the VWS site, he will reach the root is the total access time users navigate from Root
node of the VUI tree and hear a greeting message, to the leaf node. Total time is the summations of
which may look like “Welcome to XYZ online every leaf node access time multiply leaf node
service portal. If you would like to get a stock access count.
quote please say ONE, for restaurant reservation In a voice site, users’ destination must end at
please say TWO, for weather report please say leaf nodes; therefore, we focus on the leaf nodes
THREE, …” Joe may say THREE and traverse in our model. It means only leaf nodes have the
the VUI tree to N13, the node at the next level property of “access counts.” The properties of
for weather report. For the moment, Joe will be the nodes used in this example are described in
answered with more choices such as: for domestic Table 1. In this case, the values of node access
weather report please say ONE, for other countries time are randomly created in the range between
Figure 5. A VUI tree for modeling the hierarchically structured site map
722
Table 1. The properties of the nodes in the VUI tree shown in Figure 5.
Node Root N1,1 N1,2 N1,3 N2,1 N2,2 N2,3 N2,4 N2,5 N2,6 N2,7 N2,8 N2,9 N2,10
Node access time 5 6 5 4 5 3 6 8 5 4 7 4 8 3
Leaf node access time n/a n/a n/a n/a 16 19 25 24 29 33 27 31 39 42
Node access count n/a n/a n/a n/a 70 100 40 40 60 100 70 100 50 60
In this case, the total time
= 16*70 + 19*100 + 25*40 + 24*40 + 29*60 + 33*100 + 27*70 + 31*100 + 39*50 + 42*60
= 1,120 + 1,900 + 1,000 + 960 + 1,740 + 3,300 + 1,890 + 3,100 + 1,950 + 2,520
= 19,480 (seconds) -------- (1)
3 and 8 seconds, the values of node access count Using the VUI optimization mechanism, we
are only applicable to leaf nodes and randomly can restructure the VUI tree and calculate the new
generated between 40 and 100, and each value of total time. The new tree structure is shown in
leaf node access time is derived from the values Figure 6, in which the nodes annotated with star
of node access time of all nodes on the path from marks [*] were restructured, and the properties
the root navigating to that particular leaf node. of the nodes of the restructured tree are described
For example, the leaf node access time of N2,1 is in Table 2. Note that while the values of node
the summation of node access time values of the access time and node access count on all nodes
root, N1,1, and N2,1 (i.e., 5 + 6 + 5 = 16), and the are unchanged, the values of leaf node access time
leaf node access time of N2,6 is the summation on some leaf nodes are changed because the VUI
of node access time values of the root, N1,1, N1,2, tree is restructured.
N2,4, N2,5, and N2,6 (i.e., 5 + 6 + 5 +8 + 5 + 4 = 33).
Figure 6. A restructured VUI Tree [Nodes with star mark (*) were restructured.]
Table 2. The properties of the nodes in the restructured voice Web site tree
Node Root N1,1 N1,2 N1,3 *N2,2 *N2,1 *N2,3 *N2,6 N2,5 *N2,4 *N2,8 *N2,10 N2,9 *N2,7
Node access time 5 6 5 4 3 5 6 4 5 8 4 3 8 7
Leaf node access time n/a n/a n/a n/a 14 19 25 20 25 33 24 27 35 42
Node access count n/a n/a n/a n/a 100 70 40 100 60 40 100 60 50 70
In the new condition, the total time
= 14*100 +19*70 + 25*40 + 20*100 + 25*60 + 33*40 + 24*100 + 27*60 + 35*50 + 42*70
= 1,400 + 1,330 + 1,000 + 2,000 + 1,500 + 1,320 + 2,400 + 1,620 + 1,750 + 2,940
= 17,260 (seconds) -------- (2)
723
In this sample case of using our optimization services. While there are many ways to restruc-
model, the improvement (time decreases) is about ture the VUI tree, in our initial design of the VUI
11.40% ([(1) – (2)] / (1) 0.1139), even though this optimization experiment we only consider the
is just a simple illustration. As voice pages increase most straightforward approach of rearranging the
numerically and calculations are adopted in com- sequence of choices under every nonleaf node.
puterized computations, it is expected that the Other approaches, such as the operation of node
improvements, achieved by GA-based optimiza- promotion/demotion (by moving some nodes
tion approaches for the proposed voice-enabled to higher/lower levels of the VUI tree), and the
Web system, will be very attractive and better creation of extra links (for providing short-cuts to
than the result of this illustration. VUI tree navigation), may be considered in other
experiments conducted in the future. Changing
Genetic Algorithm for the order of child nodes for every nonleaf node
VUI Optimization can create many different restructured VUI trees,
and every VUI tree represents a possible solution.
The VUI optimization process of restructuring a Thus, for a VUI tree with hundreds of nodes, the
voice-enable Web site can be modeled by a tree potential number of restructured VUI trees can
structure (i.e., the VUI tree) as shown in Figure be very high. For example, let’s consider a VUI
5. For each VUI optimization experiment, our tree with 200 nonleaf nodes and assume each
simulation program will create a VUI tree, with nonleaf node has six child nodes (i.e. there are
the values of node access time of all nodes and 6 choices available on each non-leaf node), then
the values of node access count of all leaf nodes there could be potentially up to (6!)200 = 2.9275 *
randomly generated. These values will remain 10571 restructured VUI trees, or possible solutions
unchanged throughout the entire GA evolution to be considered in our problem of VUI optimi-
process. However, since each value of leaf node zation. This permutation procedure is randomly
access time of a leaf node is derived from the applied to all non-leaf nodes of the initial VUI
values of node access time of all nodes on the tree to generate nine more trees, and all these ten
path from the root navigating to that particular trees are put into the population for subsequent
leaf node, it may change when the VUI structure evolutions.
changes. As illustrated in the previous example, In our genetic algorithm, every nonleaf node
each possible solution of this problem can be is encoded by the sequence of choices available
represented by a VUI tree, and the fitness function in that node. For a non-leaf node N randomly
of this GA-based optimization is the total time, selected from tree A, its sequence of choices (or
which is the summations of every leaf node ac- order of child nodes) can be represented as N(A)
cess time multiply leaf node access count. Thus, = 1 2 3 4 5 6 7 8 9, and at the same time, the same
the objective of our optimization process is to non-leaf node N in tree B may have a different
minimize the fitness function: sequence of choices represented as N(B) = 4 5 6
9 1 2 7 3 8. The representations of GA variables
n
F (X ) = ∑ LeafNodeAccessTime (i ) × LeafNodeAccessCount (i ) can also be called chromosomes. Assume that
 
i =k
our algorithm randomly selects tree A and tree
B from the population, identifies node N for the
crossover operation, and then selects two posi-
The VUI tree of a voice-enabled Web site
tions, for example, the 3rd and the 5th positions,
can be restructured to represent alternative Web
to define how to crossover chromosomes for
site maps offering exactly the same contents and
producing new chromosomes, that is:N(A) = 1 2
724
3 4 5 6 7 8 9 → crossover → N(A) = 1 2 6 9 1 Windows XP operating system. The simulation

6 7 8 9N(B) = 4 5 6 9 1 2 7 3 8 → crossover → results obtained from 22 experiments are sum-
N(B) = 4 5 3 4 5 2 7 3 8 marized in Table 3, and the simulation results
The mapping relationship of this crossover derived from 15 additional experiments are sum-
operation can be identified as:3 ↔ 6 4 ↔ 9 5 ↔ 1 marized in Table 4. In these experiments, a total
Then, the mapping relationships are applied of 37 VUI trees were generated with different
to the unexchanged part of the chromosomes parameter settings (including the level of VUI
to create the following chromosomes for their tree, the number of children for each nonleaf node,
offspring.N’(A) = 5 2 6 9 1 3 7 8 4N’(B) = 9 1 the number of simulation cycles, access time,
3452768 access count, and so on). Furthermore, the initial
Mutation may also be applied randomly, at a total time and the optimized better total time
pre-defined probability rate, to the chromosomes. derived in each experiment were used to calculate
For example, the chromosome of N’(B) = 9 1 3 the improvement by the following straightforward
4 5 2 7 6 8 can further be encoded into a binary formula:Improvement (%) = (Initial Total Time
string:N’(B) = 1001 0001 0011 0100 0101 0010 – Better Total Time) / Initial Total Time
0111 0110 1000 Both Table 3 and Table 4 show the experiment
The algorithm can randomly modify some bits results obtained from the aforementioned SGA
to simulate the mutation operation, such as:N’(B) simulations. As you can see, most cases in these
= 1001 0001 0111 0100 01000010 0111 0110 two tables have improvements over 15%, and it is
1000 = 9 1 7 4 4 2 7 6 8 noted that the improvements of some cases even
The mapping relationship of this mutation reach nearly 50%. Besides, we noticed that the
operation can be identified as:3 ↔ 7 5 ↔ 4 results of large tree structures are not significant
Again, the mapping relationships are applied (shown as gray-highlighted items in Table 3 and
to the unexchanged part of the chromosome to Table 4), so we tried to increase searching itera-
create the following new chromosome:N”(B) = tions and recalculate them.
917542368 As shown in Table 5, when we changed
From the above mentioned population cre- searching iterations from 20,000 times to 500,000
ation, selection, crossover, and mutation opera- times, we found the improvement got a dramatic
tions, new chromosomes can be created and used increase. It means the more searching iterations
to generate new VUI trees (solutions). Each newly we give, the more improvement we get. In other
generated VUI tree is evaluated by the fitness words, for large tree structures, we can eventually
function to decide whether it is a better solution, get satisfied improvements, only if we increase
and this evaluated value is used by the culling searching iterations.
operation to decide whether the newly generated
VUI tree should be placed into the population for
replacing other less qualified VUI tree. DIscUssION AND cONcLUsION
While pervasive computing continues to affect

EXPErIMENt rEsULt more and more people in the world, there will be
inevitably plenty of opportunity and revolution-
Our system was implemented on a 1.73 GHz ary benefits for everyone who participates. The
Pentium M 740 laptop PC with 1 GB RAM, and most significant pervasive computing applica-
on another 2.8 GHz Pentium 4 desktop PC with tions have been in the enterprise market rather
2 GB RAM. Both machines are running Microsoft than the consumer sector; however, the future of
725
Table 3. Results of SGA simulations (number of children = fixed; access time = 2 ~ 15; access counts
= 1 ~10,000)
Level Number of Children Iterations Initial Total Time (s) Better Total Time (s) Improvement (%)
2 3 20,000 1,886,608 960,008 49.1
2 4 20,000 5,008,968 2,705,967 46.0
2 5 20,000 9,297,450 5,156,684 44.5
2 6 20,000 14,627,224 9,079,845 37.9
3 3 20,000 7,160,233 3,804,627 46.9
3 4 20,000 25,905,538 16,690,080 35.6
3 5 20,000 65,340,923 45,643,277 30.1
3 6 20,000 155,856,635 109,391,621 29.8
4 3 20,000 33,264,675 23,520,746 29.3
4 4 20,000 139,935,642 107,410,143 23.2
4 5 20,000 431,157,562 331,317,272 23.2
4 6 20,000 1,132,148,584 879,033,227 22.4
5 3 20,000 124,114,791 92,712,893 25.3
5 4 20,000 652,978,855 535,596,066 18.0
5 5 20,000 2,700,049,188 2,450,119,893 9.3
5 6 20,000 8,416,055,793 7,975,813,132 5.2
6 3 20,000 419,713,448 339,456,804 19.1
6 4 20,000 3,194,265,541 2,680,976,660 16.1
6 5 20,000 16,141,177,101 15,988,455,697 1.0
6 6 20,000 60,592,359,070 60,590,742,389 0.0
Table 4. Results of SGA simulations (number of children = varied; access time = 2 ~ 15; access counts
= 1 ~10,000)
2 4 to 5 20,000 6,671,516 4,049,286 39.3
2 3 to 6 20,000 5,841,651 3,263,079 44.1
2 2 to 7 20,000 3,803,479 2,071,459 45.5
3 4 to 5 20,000 49,441,465 35,029,192 29.2
3 3 to 6 20,000 34,589,124 24,112,238 30.3
3 2 to 7 20,000 26,332,244 20,114,552 23.6
4 4 to 5 20,000 310,882,476 249067717 19.9
4 3 to 6 20,000 237,910,059 195,086,909 18.0
4 2 to 7 20,000 167,725,927 134,812,794 19.6
5 4 to 5 20,000 1,637,790,586 1,249,947,619 23.7
5 3 to 6 20,000 1,434,332,917 1,258,053,217 12.3
5 2 to 7 20,000 931,088,499 816,875,439 12.3
6 4 to 5 20,000 8,997,595,006 8,892,944,474 1.2
6 3 to 6 20,000 7,832,316,476 7,731,169,501 1.3
6 2 to 7 20,000 4,932,297,039 4,744,909,127 3.8
726
Table 5. Results of SGA simulations with more iterations (access time = 2 ~ 15; access counts = 1 ~10,000)
5 6 500,000 8,416,055,793 7,087,376,995 15.8
6 6 500,000 60,592,359,070 53,334,886,394 12.0
6 4 to 5 500,000 8,997,595,006 7,338,143,520 18.4
6 3 to 6 500,000 7,832,316,476 6,768,258,648 13.6
pervasive computing will be supplemented by users to connect to the Internet from anywhere
applications used by a wider variety of profession- through devices that are not suitable for the use of
als and by more horizontal applications. Eventu- traditional keyboard, mouse, and monitor. In the
ally the access to the conventional desktop and near future, human-computer voice interfaces will
Internet applications through pervasive devices become important tools for solving the accessibil-
will become very attractive and could lead to ity limitations of conventional human-computer
pervasive computing being used as much in the interfaces. Based on the multimodal architecture,
consumer sector as it is in the enterprise world. this chapter describes how a voice-enabled web
To support various types of pervasive devices system (VWS) prototype could be implemented
in a conventional way, multiple applications to provide an interactive voice channel using an
have to be independently developed with each Apache web server, a voice server, and a Java
to satisfy one type of devices. This practice will servlet engine. We also showed through our
exponentially increase the cost, complexity, and project that multimodal user interface pages could
manageability of a system when new devices or be generated by using technologies including:
changes are introduced. To resolve this issue, our eXtensible Markup Language (XML), eXtensible
project researched on both theoretical concepts Stylesheet Language for Transformations (XSLT)
of the technologies and practical applications of (Burke, 2001), VoiceXML (Larson, 2003), and
the concepts by adopting a new software applica- Java technology (Sun Microsystems, 2004). As
tion architecture (see Figure 3) that enables one a matter of fact, it is also reconfirmed from our
single application simultaneously interfacing with project that voice interfaces may not only help
various types of distributed devices such as PC’s, solve the accessibility limitations of conventional
handheld computers, PDA’s, WAP-enabled wire- human-computer interfaces, but enable mobile
less devices, phones, and others. This multimodal device users’ hands to engage in some other ac-
application architecture overcomes the difficulties tivities without losing the ability to browse the
by singularizing the business and application logic Internet through voice commands.
while expanding device interfaces. Since common In terms of enhancing users’ experience and
business and application logic is centralized, the improving the overall system performance, a
maintenance and enhancement become much GA-based dynamic structure approach, which
easier. Our multimodal web system, VWS, was can restructure the site map according to users’
designed and implemented based on this archi- demand or the overall performance needs of a
tecture to serve as a “proof of concept” example system, was applied to our VWS system. Our
of this new e-commerce application paradigm. experiment results showed that this optimization
Nowadays, mobile and wireless technologies approach may be adopted by an adaptive VWS
are becoming increasingly prevalent, and there is Systems for supporting large-scale enterprise ap-
a growing demand for technology that will allow plications. To ameliorate the rate of convergence
727
of the optimization approach used by the adaptive Council, Taiwan, for financially supporting this
VWS system, we plan to add some heuristic rules, work under contract number NSC-96-2221-E-
such as ruin and recreate (Misevicius, 2003) and 005-088-MY2.
tabu search (Moscato & Cotta, 2003), to improve
the simple genetic algorithm used in the VWS
system, which will eventually have the ability rEFErENcEs
of self-learning to optimize itself automatically,
dynamically, and effectively. Al-Khayyal, F. A., & Sherali, H. D. (2000). On
In their influential paper on the challenges as- finitely terminating branch-and-bound algorithms
sociated with nomadic or pervasive computing, for some global optimization problems. SIAM
Lyytinen and Yoo (2002) outline eight research Journal on Optimization, 10(4), 1049–1057.
themes and twenty research questions, covering a doi:10.1137/S105262349935178X
wide range of topics in the heart land of information Anerousis, N., & Panagos, E. (2002). Making voice
systems research. If we are to choose a research knowledge pervasive. IEEE Pervasive Comput-
question posed by them that comes nearest to ing / IEEE Computer Society [and] IEEE Com-
what we do here, it is their research question 1.1, munications Society, 1(2), 42–48. doi:10.1109/
namely: “How do we design and integrate sets of MPRV.2002.1012336
personalized mobile services that support users’
task execution in multiple social and physical Ballagas, R., Borchers, J., Rohs, M., & Sheridan,
contexts?” (Lyytinen & Yoo, 2002, p. 380). Our J. G. (2006). The smart phone: A ubiquitous in-
contribution lies in showing how VWS can be put device. IEEE Pervasive Computing / IEEE
designed to provide an interactive voice channel Computer Society [and] IEEE Communications
using readily available information technology Society, 5(1), 70–77. doi:10.1109/MPRV.2006.18
products, such as Apache web server, the voice Burke, E. (2001). JAVA & XSLT. California:
server, and the servlet engine. Furthermore, we O’Reilly.
describe how multimodal user-interface pages for
supporting various wireless devices have been Chang, S. E., Changchien, S. W., & Huang, R.-H.
implemented by using technologies including (2006). Assessing users’ product-specific knowl-
eXtensible Markup Language (XML), eXtensible edge for personalization in electronic commerce.
Stylesheet Language for Transformations (XSLT), Expert Systems with Applications, 30(4), 682–693.
and Java technologies. The last but never the least, doi:10.1016/j.eswa.2005.07.021
compared with the Web browser based interface
Chang, S. E., & Minkin, B. (2006). The imple-
the voice channel is slower, and therefore, there
mentation of a secure and pervasive multimodal
is a need to apply the optimization techniques,
Web system architecture. Information and Soft-
such as the GA-based algorithm described in this
ware Technology, 48(6), 424–432. doi:10.1016/j.
chapter, to enhance the responsiveness of the VWS
infsof.2005.12.012
based services and applications.
Chang, Y.-F., & Chen, C. S. (2005). Smart phone
- the choice of client platform for mobile com-
AcKNOWLEDGMENt merce. Computer Standards & Interfaces, 27(4),
329–336. doi:10.1016/j.csi.2004.10.001
The editorial efforts and the invaluable comments
from the editor are highly appreciated. The author
would also like to thank the National Science
728
Changchien, S. W., Lee, C. F., & Hsu, Y. J. (2004). Kirkpatrick, S., Gelatt, C. D. Jr, & Vecchi, M.
Online personalized sales promotion in electronic P. (1983). Optimization by simulated annealing.
commerce. Expert Systems with Applications, Science, 220(4598), 671–680. doi:10.1126/sci-
27(1), 35–52. doi:10.1016/j.eswa.2003.12.017 ence.220.4598.671
de Grimaldo, S. W. (2004). Mobile Commerce Kohrs, A., & Merialdo, B. (2001). Creating user-
Takes off. Telecom Trends International, Inc., adapted Web sites by the use of collaborative fil-
Virginia. Retrieved November 15, 2007, from tering. Interacting with Computers, 13, 695–716.
http://www.telecomtrends.net/reports.htm doi:10.1016/S0953-5438(01)00038-8
Dettmer, R. (2003). It’s good to talk (speech tech- Larson, J. A. (2003). VoiceXML and the W3C
nology for online services access). IEE Review, speech interface framework. IEEE MultiMedia,
49, 30–33. 10, 91–93. doi:10.1109/MMUL.2003.1237554
Feng, J., Sears, A., & Karat, C.-M. (2006). A lon- Li, Y., & Zhong, N. (2004). Web mining model
gitudinal evaluation of hands-free speech-based and its applications for information gathering.
navigation during dictation. International Journal Knowledge-Based Systems, 17(3), 207–217.
of Human-Computer Studies, 64(6), 553–569. doi:10.1016/j.knosys.2004.05.002
doi:10.1016/j.ijhcs.2005.12.001
Lyytinen, K., & Yoo, Y. (2002). The next
Goose, S., Newman, M., Schmidt, C., & Hue, wave of nomadic computing. Information Sys-
L. (2000). Enhancing Web accessibility via the tems Research, 13(4), 377–388. doi:10.1287/
Vox Portal and a Web-hosted dynamic HTML isre.13.4.377.75
<-> VoxML converter. Computer Networks, 33,
Market Wire. (2006). Voxify Reports Outstanding
583–592. doi:10.1016/S1389-1286(00)00036-0
Growth, Increased Momentum in the Speech Self-
Gulliver, S. R., Serif, T., & Ghinea, G. (2004). Per- Service Marketplace. Retrieved August 10, 2006,
vasive and standalone computing: the perceptual from http://www.findarticles.com/p/articles/
effects of variable multimedia quality. Internation- mi_pwwi/is_200605/ai_n16136434
al Journal of Human-Computer Studies, 60(5/6),
Misevicius, A. (2003). Genetic algorithm hybrid-
640–665. doi:10.1016/j.ijhcs.2003.11.002
ized with ruin and recreate procedure: Application
Institute for Information Industry. (2007). Survey to the quadratic assignment problem. Knowledge-
on the mobile Internet in Taiwan for Q3 2007. ACI- Based Systems, 16(5-6), 261–268. doi:10.1016/
FIND, focus on Internet news and data. Retrieved S0950-7051(03)00027-3
January 2, 2008, from http://www.find.org.tw/
Moscato, P., & Cotta, C. (2003). Gentle introduc-
find/home.aspx?page=many&id=184
tion to memetic algorithms. In F. Glover & G.
Joachims, T., Freitag, D., & Mitchell, T. (1997). Kochenberger (Eds.), Handbook of metaheuris-
WebWatcher: A tour guide for the World Wide tics (pp. 105-144). Boston: Kluwer Academic
Web. In Proceedings of IJCAI-97, Fifteenth Joint Publishers.
Conference on Artificial Intelligence, Nagoya,
Perkowitz, M., & Etzioni, O. (2000). Towards
Japan (pp. 770-775).
adaptive Web sites: Conceptual framework and
case study. Artificial Intelligence, 118, 245–275.
doi:10.1016/S0004-3702(99)00098-3
729
Rebman, C. M. Jr, Aiken, M. W., & Cegielski, C. G. Teo, T., & Pok, S. (2003). Adoption of WAP-
(2003). Speech recognition in the human-computer enabled mobile phones among Internet users.
interface. Information & Management, 40(6), Omega: The International Journal of Man-
509–519. doi:10.1016/S0378-7206(02)00067-8 agement Science, 31, 483–498. doi:10.1016/j.
omega.2003.08.005
Rodriguez, A., Ho, W.-K., Kempny, G., Pedre-
schi, M., & Richards, N. (2002). IBM WebSphere The Kelsey Group (2001, March). The global voice
Voice Server 2.0 Implementation Guide. IBM ecosystem (Analyst Report), The Kelsey Group.
Redbooks, IBM.
Toye, E., Sharp, R., Madhavapeddy, A., & Scott,
Roussos, G., Marsh, A. J., & Maglavera, S. (2005). D. (2005). Using smart phones to access site-spe-
Enabling pervasive computing with smart phones. cific services. IEEE Pervasive Computing / IEEE
IEEE Pervasive Computing / IEEE Computer Computer Society [and] IEEE Communications
Society [and] IEEE Communications Society, Society, 4(2), 60–66. doi:10.1109/MPRV.2005.44
4(2), 20–27. doi:10.1109/MPRV.2005.30
Turner, K. (2004). Analysing interactive voice
Smith, K. A., & Ng, A. (2003). Web page cluster- services. Computer Networks, 45(5), 665–685.
ing using a self-organizing map of user navigation doi:10.1016/j.comnet.2004.03.005
patterns. Decision Support Systems, 35, 245–256.
Wang, F. H., & Shao, H. M. (2004). Effective
doi:10.1016/S0167-9236(02)00109-4
personalized recommendation based on time-
Su, Z., Yang, Q., Zhang, H., Xu, X., Hu, Y.-H., & framed navigation clustering and association
Ma, S. (2002). Corellation-based Web document mining. Expert Systems with Applications, 27(3),
clustering for adaptive Web interface design. 365–377. doi:10.1016/j.eswa.2004.05.005
Knowledge and Information Systems, 4, 151–167.
Yamazaki, Y., Iwamida, H., & Watanabe, K.
doi:10.1007/s101150200002
(2004). Technologies for voice portal platform.
Sun Microsystems. (2004). Information on J2ME Fujitsu Scientific and Technical Journal, 40(1),
and J2SE. Retrieved June 2, 2007, from http:// 179–186.
java.sun.com/j2me/ and http://java.sun.com/j2se/
Zhai, S., Kristensson, P.-O., & Smith, B. A.
Tan, A. (May 2005). Voice to dominate 3G traf- (2005). In search of effective text input inter-
fic, says expert. ZDFNetAsia. Retrieved June faces for off the desktop computing. Interacting
2, 2007, from http://www.zdnetasia.com/news/ with Computers, 17(3), 229–250. doi:10.1016/j.
communications/0,39044192,39231956,00.htm intcom.2003.12.007
This work was previously published in Global Implications of Modern Enterprise Information Systems: Technologies and
Applications, edited by Angappa Gunasekaran, pp. 137-155, copyright 2009 by Information Science Reference (an imprint of
IGI Global).
730
731
Chapter 3.12
Achieving System and
Business Interoperability by
Semantic Web Services
John Krogstie
Norwegian University of Science and Technology (NTNU), Norway, & SINTEF ICT, Norway
Csaba Veres
Norwegian University of Science and Technology (NTNU), Norway
Guttorm Sindre
Øyvind Skytøen
AbstrAct INtrODUctION
Much of the early focus in the area of Semantic Information systems interoperability has become
Web has been on the development of representa- a critical success factor for process and quality
tion languages for static conceptual information; improvement both in private enterprises and the
while there has been less emphasis on how to make public sector (Linthicum, 2003), and recent tech-
Semantic Web applications practically useful in the nological advances to achieve this include web ser-
context of knowledge work. To achieve this, a bet- vices and semantics encoded in ontologies. “The
ter coupling is needed between ontology, service Semantic Web” (Berners-Lee, Hendler & Lassila,
descriptions, and workflow modeling, including 2001) is seen as the next generation of web sys-
both traditional production workflow and interac- tems, providing better information retrieval, better
tive workflow techniques. This chapter reviews services, and enhanced interoperability between
the basic technologies involved in this area to different information systems. The Semantic Web
provide system and business interoperability, and initiative is currently overseen in the semantic
outlines what can be achieved by merging them in web activity of the W3C, and includes a number
the context of real-world workflow descriptions. of core technologies. Some core technologies that
will be relevant to this overview are XML, RDF,
DOI: 10.4018/978-1-60566-146-9.ch010 RDF/S, OWL, and Web Services (SOAP, WSDL,
Achieving System and Business Interoperability by Semantic Web Services
UDDI). Also newer initiatives such as OWL-S The rest of this chapter is structured as follows:
and WSMO are relevant to our work, and will be The next three sections survey ontologies, service
described in more detail in the article. While these models, and workflow models, respectively. Then
technologies are promising, it can still be argued an integrated approach to enterprise and IS devel-
that alone, they are not sufficient to achieve in- opment is presented, where interoperability among
teroperability in the business domain, allowing for the various systems (and enterprises) would be
a smooth integration between different information a major focus. Finally, the last section provides
systems within and between organizations. For this some concluding remarks.
to be accomplished, it is not enough to describe
ontological metadata about the information and
services available – one also needs to know the bAsE tEcHNOLOGIEs
work context in which the different types of in- AND ONtOLOGy
formation and services are requested. As observed
in (Bubenko, 2007) this is often a challenge, as We here briefly describe core technologies within
many ontologists focus on domain ontologies as the area, including XML, RDF, RDF Schema,
such, more than their potential usage in applica- and ontologies including an overview of OWL.
tions, as well as having limited knowledge of
advances in other areas of conceptual modeling XML
during the last decades. Hence there is a need to
integrate ontologies and service descriptions with XML will receive the least coverage in this re-
models of workflows and business processes. view. It is the most general and widespread of the
Most of the work within these areas focuses on technologies we consider, and is therefore likely
automating routine tasks. While computerization to be familiar to the majority of readers. Basically,
automates routine procedures, knowledge-based XML defines a set of syntax rules that can be used
cooperation remains a challenge, where we see a to create semantically rich markup languages for
role for interactive process models. To the extent particular domains. Once a markup language is
that different enterprises use different modeling defined and the semantics of the tags known, the
languages, the interoperability between various document content can be annotated. The XML
models would also emerge as a challenge in its language thus defined can include specification
own respect, in which case some unification effort of formatting, semantics, document meta-data
might be needed (Opdahl & Sindre, 2007), one (author, title, etc.), and so on. XML allows for the
effort in this direction is the Unified Enterprise creation of elements which are XML containers
Modeling Language (UEML)1, not to be confused consisting of a start tag, content, and an end tag.
with the UML. Because of the flexibility of XML in defining
The purpose of this chapter is as follows: domain specific, meaningful markups, it has been
widely adapted as a standard for application inde-
a) To provide an overview of the relevant pendent data exchange. These properties combine
technologies (ontology, service models, to make XML the foundational technology for the
workflow models, including those being semantic web, providing a common syntax for
based on interactive models). authoring web content. XML provides means for
b) To show how these technologies fit together, syntactic interoperability, as well as ways to ensure
both in theory (presented as “The interoper- the validity of a document, and most importantly
ability pyramid”) and in practice. the necessary syntax to define the meaning of
elements in a domain specific application. On
732
the other hand providing the syntax for defining 2. RDF distinguishes between resources and
meaning is only a necessary, but not sufficient properties that are globally qualified, i.e.,
condition for the specification of semantics that are associated with a URI, and those that
allows interoperability are locally qualified. The advantage of a
Building on the XML specification also be- globally qualified resource or property is it
comes necessary because the hierarchical structure can be distinguished from other resources
of XML documents makes them difficult to use for or properties in different vocabularies that
extensible, distributed data definitions. Much of share the same fragment name, in a fashion
the information about relationships in the data is that is analogous to XML namespaces.
implicit in the structure of the document, making 3. As a result of the first two properties, RDF
it difficult to use and update this information in can be used to make statements about Web
a flexible and application independent way. This resources, by relating one URI to another.
is where RDF comes into the picture. 4. It is easy to encode graphs using RDF as it
is based on triples, whereas XML documents
rDF are trees so encoding graphs is more com-
plicated and can be done in several different
The first level at which a concrete data model ways.
is defined on XML is the Resource Description 5. RDF has an explicit interpretation or model
Framework (RDF). Actually, RDF as a data model theory; there is an explicit formal, application
is independent of XML, but we consider it as a independent interpretation of an RDF model
layer extending the XML because of the widely (Hayes, 2004). XML documents also have
practiced XML serialization of RDF in semantic interpretations but they are often implicit in
web applications (RDF/XML2). the processor or parser associated with that
The basic structure of RDF is a triple consisting particular type of XML document.
of two nodes and a connecting edge. These basic
elements are all kinds of RDF resources, and can But in spite of the apparent usefulness of RDF,
be variously described as <things> <properties> there is relatively slow adoption of RDF compared
<values> (Manola & Miller, 2004), <object> with XML (Batzarov, 2004).
<attribute> <value> (Broekstra, Kampman, & There are many possible reasons for this slow
van Harmelen, 2003), or <subject> <predicate> adoption. (Daconta, Obrst, & Smith, 2003) take an
<object> (Powers, 2003). There are alternative optimistic position and attribute the long lead-in
serializations of RDF, including N33, N-Triples4, time to poor tutorials, minimal tool support, and
and Turtle5. Each of these professes some advan- poor demonstration applications, arguing that
tages, for example human readability, but RDF/ once the practical limitations have been overcome,
XML is the normative syntax for writing RDF. adoption will grow rapidly. However, we must not
This relatively simple basic model has several ignore the presence of dissatisfaction with RDF
features that make it a powerful data model for in- in both practitioner and research communities.
tegrating data in dispersed locations (Butler 2002). Some of the challenges for RDF in light of this
dissatisfaction are as follows:
1. RDF is based on triples, in contrast to simple
attribute-value pairs. The advantage of using 1. RDF / XML (or XHTML) integration needs
triples is that this makes the subject of the improvement. The W3C RDF Working
attribute value pair explicit. Group is working on solutions for success-
fully embedding RDF within XHTML (RDF/
733
A6), and tools such as SMORE7 purport to non-RDF based, and syntactically much
making HTML markup easier. But so far simpler RSS 2.010.
there are no high profile, compelling appli-
cations to showcase the advantages of RDF. Clearly there is a great deal of work to be done
For example microformats8, which can be in establishing RDF as a core technology that adds
seen as a very simple version of RDF/A but value to the widely adopted XML syntax alone.
are “designed for humans first and machines There are some fledgling ventures launched in
second”, have enjoyed a rapid uptake. For 2007, backed by high profile investors, which
example, both Yahoo! and Google can run attempt to bring the advantages of RDF to main-
specialized searches on microformats. stream social networking applications11. Should
2. The RDF data model can be complex and they become successful, then RDF will become
confusing because it mixes metaphors and more prominent in the public eye.
introduces new concepts that can be tricky But RDF is also important as a foundation
to model. For instance the standard notion layer for Ontologies, making it relatively simple
of RDF as composed of subject-predicate- to express higher level ontological constructs.
object is linguistically derived, but its rela- Implementing ontologies in XML and XML
tionship to concepts in other representations Schema without RDF is tricky for several reasons.
is somewhat unclear, e.g., class-property- In describing a procedure for translating an on-
value (object-oriented), node-edge-node tology into an XML Schema, (Klein et al., 2003)
(graph theory), source-link-destination (web note several important problems. First, superclass/
link), entity-relation-entity (database), and subclass inheritance is problematic and has to be
can cause confusion. One of the particularly overcome with artificial workarounds in the XML
tricky constructs is reification, which intro- specification, and defining multiple inheritance is
duces an unproven modeling construct that is not possible at all in XML/S. Second, the possibil-
foreign to most data modeling communities. ity of fully automating the translation process is
Reification can cause confusion because it questionable, limiting its use for large ontologies.
can be used to arbitrarily nest statements, In order to use RDF as a means of representing
possibly negating the stated truth value of knowledge it is necessary to enrich the language
statements (Daconta, Obrst, & Smith, 2003). in ways that fixes the interpretation of parts of
3. The RDF/XML serialization is confusing the language. As described thus far, RDF does
and difficult to work with, especially in the not impose any interpretation on the kinds of
absence of proper tool support. The striped resources involved in a statement beyond the
syntax (Brickley, 2001) can make it difficult roles of subject, predicate and object. It has no
to understand the proper interpretation of way of imposing some sort of agreed meaning
statements. For instance it is often impos- on the roles, or the relationships between them.
sible to tell whether an XML element in The RDF schema is a way of imposing a simple
the RDF serialization represents an edge, ontology on the RDF framework by introducing
or a node. The complexity of the syntax is a system of simple types.
partially responsible for a relative support
of the RSS1.09 specification. RSS1.0 is an rDF schema
RDF based variant of the popular RSS for-
mat, and is probably the most high profile We have seen that RDF provides a means to
use of RDF on the Internet. However, it is relate resources to one another in a graph based
losing ground in terms of popularity to the formalism connecting subjects to objects via
734
predicates. The RDF schema, (RDF/S) provides limiting deficiencies are identified in (Antoniou
modeling primitives that can be used to capture & van Harmelen, 2004):
basic semantics in a domain neutral way. That is,
RDF/S specifies metadata that is applicable to the • Local scope of properties: in RDF/S it is
entities and their properties in all domains. The possible to define a range on properties,
metadata then serves as a standard model by which but not so they apply to some classes only.
RDF tools can operate on specific domain models, For instance the property eats can have a
since the RDF/S meta-model elements will have a range restriction of food that applies to all
fixed semantics in all domain models. The RDF/S classes in the domain of the property, but it
elements are shown in Table 1 and Table 2. is not possible to restrict the range to plants
RDF/S provides simple but powerful modeling for some classes and meat for others.
primitives for structuring domain knowledge into • Disjointness of classes cannot be defined
classes and sub-classes, properties and sub- prop- in RDF/S.
erties, and can impose restrictions on the domain • Boolean combinations of classes are not
and range of properties, and defines the semantics possible. For example Person cannot be
of containers. defined as the union of the classes Male
The simple meta-modeling elements can limit and Female.
the expressiveness of RDF/S. Some of the main • Cardinality restrictions cannot be
expressed.
Table 1. RDF/S classes
Class name comment

rdfs:Resource The class resource, everything.
rdfs:Literal The class of literal values, e.g. textual strings and integers.
rdfs:Class The class of classes.
rdfs:Datatype The class of RDF datatypes.
rdfs:Container The class of RDF containers.
rdfs:ContainerMembershipProperty The class of container membership properties, rdf:_1, rdf:_2, ..., all of which are sub-
properties of ‘member’.
Table 2. RDF/S properties
Property name comment domain range

rdfs:subClassOf The subject is a subclass of a class. rdfs:Class rdfs:Class
rdfs:subPropertyOf The subject is a subproperty of a property. rdf:Property rdf:Property
rdfs:domain A domain of the subject property. rdf:Property rdfs:Class
rdfs:range A range of the subject property. rdf:Property rdfs:Class
rdfs:label A human-readable name for the subject. rdfs:Resource rdfs:Literal
rdfs:comment A description of the subject resource. rdfs:Resource rdfs:Literal
rdfs:member A member of the subject container. rdfs:Resource rdfs:Resource
rdfs:seeAlso Further information about the subject resource. rdfs:Resource rdfs:Resource
rdfs:isDefinedBy The definition of the subject resource. rdfs:Resource rdfs:Resource
735
• Special characteristics of properties like the chances of mutual understanding are greatly
transitivity cannot be expressed. enhanced.
Perhaps the most publicly visible use for
Ontologies simple ontologies is the taxonomies used for site
organization on the World Wide Web. This allows
A good starting point for understanding what ontol- designers to structure information and users to
ogy entails, is to consider Figure 1, adopted from browse and search. Taxonomies can also help
(Daconta, Obrst, & Smith,2003), which places a with sense disambiguation since the context of
number of knowledge models on a continuum. a term is given by the more general terms in the
As you go from the lower left corner to the upper taxonomy.
right, the richness of the expressible semantics Structured ontologies provide more sophis-
increases. This is shown on the right side of the ticated usage scenarios. For instance, they can
arrow with some typical expressions that have provide simple consistency and completeness
some sort of defined semantics for the particular checks. If all products must have a price then web
model. The names for the knowledge models are sites can automatically be checked for missing
given on the left of the arrow. It is important to or conflicting information. Such ontologies can
note that all of the terms on the left hand side have also provide completion where partially speci-
been called “ontology” by at least some authors, fied information can be expanded automatically
which is part of the source for confusion about by reference to the terms in the ontology. This
the word. expanded information could also be used for
Models based on the various points along the refining search, for instance. Ontologies can also
ontology spectrum have different uses (McGuin- facilitate interoperability, by aligning different
ness, 2003). In the simplest case, a group of users terms that might be used in different applications
can agree to use a controlled vocabulary for their (McGuinness, 2003).
domain. This of course does not guarantee that Now we are in a position to see why the on-
they will use the terms in the same way all the tologies on the most formal end of the spectrum
time, but if all the users including database design- are often taken as the default interpretation in
ers chose their terms from an accepted set, then the context of the semantic web, providing the
conceptual underpinning for “ ... making the
Figure 1. The ontology spectrum
736
semantics of metadata machine interpretable” the interpretation of some RDF/S expressions. The
(Staab & Stuber, 2004). But for the semantics of OWL specification defines three sublanguages:
a domain model to be machine interpretable in OWL Full, OWL DL, and OWL Lite: OWL Full
any interesting way, it must be in a format that is upward and downward compatible with RDF
allows automated reasoning in a flexible manner. but OWL DL and OWL Lite are not.
Obviously, taxonomies can specify little in this The names of the three sub languages of OWL
sense. Database schemas are more powerful, but describe their expressiveness, keeping in mind a
limit the interpretation to a single model in terms fundamental tradeoff between expressiveness,
of reasoning over the knowledge base and the only efficiency of reasoning, and support for human
automated reasoning that can be performed is what understanding. OWL Full has constructs that
is allowed by the relational model, i.e., retrieval make the language undecidable. Developers
of tuples actually represented in the database. should therefore only use OWL Full if the other
Formal logic based reasoning about ontologies two sub languages are inadequate for modeling
can consider multiple possible models (Bordiga the relevant domain, or if they wish to maintain
& Brachman, 2003). They are at the same time full compatibility with RDF. Similarly, OWL
more formally constrained and more semantically DL should be used if OWL Lite is not sufficient.
flexible than database schemas. Ontologies based Details of the syntax and semantics can easily be
on different logics can support different kinds of obtained from the technical documentation web
inference, but a minimal set of services should site of the W3C.
include reasoning about class membership, class
equivalence, consistency, and classification (An-
toniou & van Harmelen, 2004). WEb sErVIcEs
The ontology representation language adopted
by the Web Ontology Working Group of the W3C12 There is a great deal of interest about web ser-
is the Web Ontology Language (OWL). OWL is vices (Alonso et al., 2004) and service oriented
a response to a number of requirements (Smith, architectures in general. A useful definition can
Welty, & McGuinness, 2004) including the need be found in (Daconta, Obrst, & Smith, 2003):
for a language with formal semantics that enables “Web services are software applications that can
automated reasoning, and to address the inherent be discovered, described, and accessed based on
limitations of RDF/S as described above. XML and standard Web protocols over intranets,
extranets, and the Internet.” This definition ex-
OWL poses the main technical aspects of web services,
to do with discovery and description, as well as the
According to the original design goal, OWL role of WWW (e.g. XML) technologies for data
was to be a straightforward extension of RDF/S, exchange and communication. Also the definition
guaranteeing downward compatibility such that is abstract enough to exclude low level protocols
an OWL aware processor could also understand like RPC as web services. These core concepts
RDF/S documents without modification. Unfortu- along with the associated technologies are shown
nately this did not succeed because the generality in Figure 2 below.
of some RDF/S elements (e.g. the semantics of It is important to situate the role of Web ser-
class as “the class of all classes”) does not make vices in the real world. (Daconta, Obrst, & Smith,
RDF/S expressions tractable in the general case. 2003) argue that the most important factor for
In order to maintain computational tractability, determining the future of a new technology is not
OWL processors include restrictions that prevent “... how well it works or how “cool” it is ...” but
737
Figure 2. The basic layers of Web services

guage keyword matching, 2) ontology based
keyword matching (increasing precision through
a controlled vocabulary), and 3) semantic match-
making, based on precise semantic descriptions
of services and service needs. Currently, service
descriptions in UDDI, for example, are primarily
text descriptions with no semantic markup, requir-
ing a lot of manual input, and not facilitating the
more advanced approaches to discovery.
service composition
on business adoption. Along this line they see a
bright future for Web services which is being As for service composition (Berardi et al., 2005)
promoted by Microsoft, IBM, Sun, as well as the distinguishes between synthesis, building the
open source community. But why such widespread specification of the composite service from its sub-
support? One reason is the promise of interoper- services, and orchestration, which is the run-time
able systems. Once businesses adopt standardized management of the composite service (scheduling,
web service descriptions, the possibility of ex- invoking sub-services, etc.). Synthesis can be done
changing data and sharing the cost of services either manually or automatically, the latter requir-
increases. In addition, the open standards prevent ing that services have been specified formally. The
monopolization of applications, preventing the orchestration problem for web services has a lot in
dreaded “vendor lock-in” associated with propri- common with similar issues in workflow manage-
etary solutions. Finally, a widespread adoption of ment, which will be discussed in the next section.
Web service protocols means that existing ap- (Dijkman & Dumas, 2004) identify four different
plications can be leveraged by turning them into viewpoints from which the control-flow aspects
Web services. As an example, it is possible for of web services can be described, distinguishing
.NET clients and servers to talk to J2EE servers between choreography, which is a collaboration
using SOAP. between service providers and user to achieve a
The point of all this is that Web services enable certain goal, and orchestration, which is what a
interoperability at the level of business processes service provider performs internally to realize a
without having to worry about interoperating service it provides. (The other two viewpoints
between different applications, data formats, are behavior interface and provider interface).
communication protocols, and so on. We will There are two ongoing standardization efforts
see later in the article that this influences the way related to service composition (Barros, Dumas &
workflows and knowledge based work processes Oaks, 2005), the Web Service Business Process
are modeled and instantiated in particular work Execution Language (WS-BPEL), formerly
environments. known as BPEL4WS, and the Web Service Cho-
As describe above web services must be dis- reography Description Language (WS-CDL).
covered, described, and appropriately connected WS-BPEL (Arkin et al., 2005) is meant to specify
in an implementation independent way. (Berardi both abstract and executable business processes,
et al., 2005) outline 3 different approaches for and the language contains one section of core
web service discovery, on a trade-off between concepts (needed for both kinds of specifications)
ease of provision and accuracy: 1) natural lan- as well as sections with extensions for executable
processes and abstract processes (a.k.a. business
738
protocols), respectively. The main viewpoint taken tool seems to be the only one publicly available
in WS-BPEL is that of orchestration, requiring for download.
centralized control of the business process. WS-
CDL (Cavantzas et al., 2005), takes the alternative WsMO
viewpoint of choreography, meaning that this
language is better suited for describing interplay The Web Service Modeling Ontology (WSMO) is
between several independent parties in a shared a prject undertaken by the WSMO Working group
control domain. under the SDK19 project cluster (EU). WSMO
There are several proposed frameworks to consists of three main components: a modeling
facilitate automatic support for describing, finding framework of core elements for semantic web
and composing web services: METEOR-S13, Web services, a formal description language (Web
Service Modeling Ontology (WSMO)14, Internet Service Modeling Language - WSML) and an
Reasoning Service (IRS)15, and OWL-S16. execution environment (WSMX). The WSMO
core elements are
MEtEOr-s
1. Ontologies – provide the formally specified
The METEOR-S project is run at the LSDIS Lab terminology of the information used by all
at the University of Georgia, as a successor of the other components
METEOR project, whose focus was on workflow 2. Goals – objectives that a client wants to
management in a more traditional transaction- achieve by using Web Services
oriented perspective. METEOR-S take a more 3. Web Services – semantic description of web
semantic and dynamic perspective on workflow services including functional capability and
management. The METEOR-S architecture usage interface
consists of three main components: the process 4. Mediators – connectors between compo-
designer, the configuration module, and the execu- nents with mediation facilities for handling
tion environment. The process designer module heterogeneities
supports the design of abstract work processes
represented inWS-BPEL. The Jena toolkit is Each of these elements is further described
used for building and processing ontologies. The by non-functional properties including the Dub-
process configuration module is responsible for lin Core Metadata Set, versioning information,
dynamically finding and binding services for the quality of service information, and other relevant
defined processes. The METEOR-S Web Service annotations.
Annotation Framework (MWSAF)17 tool is used Together these components are able to define
for semi-automatically annotating web services the terminology of the domain and how it relates
with semantics using the WSDL-S18 language. to applications, and to describe the service in terms
The execution environment consists of a logical of its pre-conditions, post-conditions, effects,
layer over a web process execution engine. The and mediators required during the discovery and
execution engine uses proxies for each virtual execution of the service.
partner of the process. To support run-time and Several tools related to WSMO are publicly
deployment-time binding the configuration mod- available, the most important ones being the Web
ule can change the service bound to the proxies. Service Execution Environment (WSMX), the
At the time of writing (December 2007) the Web Service Modeling Toolkit (WSMT), and the
METEOR-S framework is in version 0.8, with a WSML Validator.
couple of finished tools. Currently, the MWSAF
739
Internet reasoning service in the language OCML (Operational Conceptual

Modeling Language)21.
The Internet Reasoning Service (IRS) is an ongo-
ing project at the Knowledge Media Institute at OWL-s
the Open University. IRS has many similarities
with WSMO, as it actually uses the WSMO’s OWL-S is a W3C initiative to provide an ontol-
ontology as a basis, but then providing several ogy and language to describe web services. It
extensions. In particular, the concepts of goal is less revolutionary than WSMO or IRS, as is
and web service have been extended in IRS rela- evidenced by its closer ties to current standards
tive to the original WSMO definition. The latest like WSDL and UDDI. Its primary role is to as-
implementation is IRS-III, supporting the follow- sist discovery, which it fulfils by specifying three
ing activities for building semantic web services key components of a service as parts of its Upper
(Cabral et al., 2006): Ontology of Services:
• Use of domain ontologies • What does the service provide for prospec-
• Description of client requests as goals tive clients? The answer to this question is
• Semantic description of deployed web given in the Service Profile which is used
services to advertise the service.
• Resolution of conceptual mismatches • How is it used? The answer to this question
• Publication and invocation of the described is given in the “process model.” This per-
web services spective is captured by the ServiceModel
class.
The support for these activities is achieved • How does one interact with it? The answer
by the IRS-III server, having the following com- to this question is given in the “ground-
ponents: ing.” A grounding provides the needed de-
tails about transport protocols.
• The SWS library, where the semantic de-
scriptions of web services are stored. Thus each service presents a Service Profile
• Interpreters for choreography and orches- (what it does), is described by a Service Model
tration, respectively (how it works), and supports a Service Grounding
• The Mediation Handler, supporting bro- (how to access it).
kering in the process of selecting, compos- Available implementations for OWL-S include
ing and invoking web services. OWL files for the Upper Ontology of Services
• The Invoker, which communicates with and other relevant ontologies used by this Upper
the service publishing platform, sends in- Ontology. Moreover, a set of relevant tools have
put from the client to the invoked services, been released, mostly by third parties, for instance:
and returns the results back to the client.
• The OWL-S Protégé-based editor22
The IRS server is written in LISP and is avail- • Another OWL-S Editor23 developed at the
able as an executable file. The publishing platforms University of Malta
for web services are available as Java web appli- • ASSAM Web Service Annotator (Hess et
cations. Also available is the WebOnto20 tool for al., 2004)
visualizing and editing IRS-III ontologies defined • Semantic Web Service Composer24
740
• OWL-S matcher25, to assess the degree of another. In the OWL-S vision this is a step which
correspondence between different service can detract from the primary purpose of discovery.
descriptions To be sure the translation problems still need to be
solved, but OWL-S assumes this will be possible
comparison of the Frameworks through some form of composition (Ankolekar
et al., 2004). But this has some implications for
While OWL-S is a less comprehensive approach, the use of each system in a specific context, such
there are certain similarities between this and the as the system described in subsequent sections.
WSMO-based approaches. The comparison of the four discussed frame-
works is summed up in Table 3, concerning the
• OWL-S Service Profile ≈ WSMO capa- available support for various activities related
bility + goal + non-functional properties. to the development and running of applications
WSMO separates provider (capabilities) based on semantic web services.
and requester points of view (goals) while As can be seen from this comparison, OWL-S
OWL-S Profiles combine existing capabil- is the framework which is most different from the
ities (advertisements) and desired capabili- others, in the sense that it specifies less of the
ties (requests) surrounding architecture for service and applica-
• OWL-S process model ≈ WSMO Service tion development. While the other three include
Interfaces. The process model in the some kind of supporting tools for all the 7 tasks
OWL-S ServiceModel roughly corre- indicated in the left column of the table, OWL-S
sponds to the interfaces in the WSMO Web has not provided any specific support for publish-
Services descriptions of WSMO. ing, composition, invocation and deployment.
• OWL-S Grounding ≈ WSMO Grounding.
Both provide a mapping to WSDL.
WOrKFLOW AND ENtErPrIsE
Nevertheless, clear differences exist in the PrOcEss MODELING
overall architecture as well as the reliance of
WSMO on explicitly defined mediators. A key The unprecedented flexibility of web services
objective of the WSMO is to define a taxonomy provides a particular challenge for how to integrate
of mediators to translate between message pro- their use in enterprise work practices. On the one
duced by one Web service and those expected by hand, demand based service provision promises
Table 3. Comparison of the four frameworks
SWS Activity METEOR-S WSMO IRS-III OWL-S

Publishing Process Designer WSMO Editor / Service Publishing Client / Not detailed
Repository Handler
Discovery Config. Module Matchmaker Mediation Handler Matcher
Composition Config. Module Matchmaker Mediation Handler Not detailed
Selection Config. Module Selector Mediation Handler Matcher
Invocation Execution Engine Communication Manager Invocation Handler / Not detailed
Publishing Platf.
Deployment MWSAF Matchmaker Publishing Platf. Not detailed
Ontology management MWSAF WSMO Editor Mediation Handler OWL-S Editors
741
to be a blessing for facilitating problem solving; Workflow and Process

on the other hand, the instance based variability Modeling Languages
provided through the relatively free range of solu-
tions offered in service composition could result Workflow modeling has been used to learn about,
in a serious challenge to established workflow guide and support practice in a number of different
modeling paradigms. areas including software process improvement
Process modeling implicates a family of (Bandinelli et al., 1995; Derniame, Kaba & Was-
techniques used to document and explicate a set tell, 1998), enterprise modeling (Fox & Grüninger,
of business and work processes in a way that 2000), process centric software engineering (Am-
allows their analysis at various levels, and for briola, Conradi & Fuggetta 1997), and workflow
various purposes. Our specific interest here is systems (Fischer, 2001). The process modeling
to use workflow modeling to analyze the work languages employed in these areas have been use-
contexts that are likely to be involved in the day fully categorized into one of the following types:
to day activities of an enterprise, with the aim transformational, conversational, role-oriented,
of improving the timely delivery of appropriate constraint-based, and systemic (Carlsen, 1997).
information related resources and services. The A summary of each type is given in (Jørgensen,
purpose is to integrate workflow modeling with 2004) where they are considered for their suit-
the potential of web services, to capture the likely ability as interactive modeling paradigms.
usage scenarios under which the services will need Transformational languages represent the
to operate and to model this integrated use. The majority of process modeling languages in use,
aim is that the model of work practices will allow adopting an input-process-output approach. Some
better specification of actual information needs, well known languages adopting this approach are
which will in turn allow for richer requirements Data Flow Diagrams (DFD), IDEF-0, Activity
for the service descriptions expected from a web diagrams, BPMN, Event-driven Process Chains
service, which will facilitate service composition (EPC), and Petri nets. While there are clear dif-
and interoperability. The research problems there- ferences between the formalisms, it is possible
fore complement one another: workflow modeling to generalize in terms of their basic expressive
helps web service design, but the availability of commitments and therefore suitability for model-
these services in turn improves workflow modeling dynamic, flexible workflows (Conradi & Jac-
ing techniques. cheri, 1998; Curtis, Kellner & Over, 1992; Green
The challenge for us is to construct modeling & Rosemann, 2000; Lei & Singh, 1997). The
approaches that maintain sufficient expressive standards defined by the Workflow Management
power for the required points of view as well Coalition (Fischer, 2001), the Internet Engineer-
as to allow flexibility for changing situations. ing Task Force (IETF) (Bolcer & Kaiser, 1999),
(Jørgensen, 2004; Krogstie & Jørgensen, 2004) and the Object Management Group (OMG, 2000)
argue that static workflow models cannot handle are all predicated on a common perspective. We
the changing demands of real world situations, and consider a few languages from this perspective.
that adaptive models, while providing greater flex- The WfMC standards for process definition
ibility, still cannot adequately handle the instance interchange between systems (WfMC, 1999)
based user driven modifications needed in many include a large portion of the primitives involved
situations. They argue for interactive models that in transformational languages. Processes are
can be dynamically configured by users. modeled with hierarchical decomposition, control
flow structures for sequences, iteration, AND and
XOR branching. Activities can be associated with
742
organizational roles and actors, tools and appli- is promising for use in flexible systems. Unfortu-
cations. The core terminology of the WfMC is nately the promise is only partially realized since
found in (Fischer, 2001). Importantly, there is a different primitives are used for the two cases,
distinction between process definition (idealized implying that the automation boundary must
process) and instance (actual work). be defined during process design. Additionally,
From past experience with developing flexible BPML has weak support for local change and
groupware and workflow systems (Carlsen, 1998; unforeseen exceptions. A visual notation, for
Jørgensen, 2001, 2003; Jørgensen & Carlsen, BPML, BPMN has been developed and is getting
1999; Natvig & Ohren, 1999), we have defined increasing support both among researchers and in
an interactive models approach to flexible infor- industrial practice (BPMI.org and OMG, 2006).
mation systems (Jørgensen, 2004). Models are The further standardization of BPMN is taken
normally defined as explicit representations of over by OMG, which have also standardized The
some portions of reality as perceived by some Business Process Definition Metamodel (BPDM)
actor (Wegner & Goldin, 1999). A model is ac- which applies MDA principles to provide a con-
tive if it directly influences the reality it reflects. sistent end-to-end approach for business process
Model activation involves actors interpreting the modeling. The development of
model and adjusting their behavior accordingly. BPMN was based on the revision of other nota-
This process can be tions, including UML, IDEF, ebXML, RosettaNet,
LOVeM and EPCs, and stemmed from the demand
• Automated, where a software component for a graphical language that complements the
interprets the model, BPEL standard for executable business processes.
• Manual, where the model guides the ac- Although this gives BPMN a technical focus, it
tions of human actors, or has been the intention of the BPMN designers
• Interactive, where prescribed aspects of to develop a modeling language that can be ap-
the model are automatically interpreted plied for typical business modeling activities as
and ambiguous parts are left to the users well. The complete BPMN specification defines
to resolve. thirty-eight distinct language constructs plus
attributes, grouped into four basic categories of
We define a model to be interactive if it is in- elements, viz., Flow Objects, Connecting Objects,
teractively activated. By updating such a model, Swimlanes and Artefacts. Flow Objects, such
users can adapt the system to fit their local plans, as events, activities and gateways, are the most
preferences and terminology. basic elements used to create Business Process
The Business Process Modeling Language Diagrams (BPDs). Connecting Objects are used
(BPML) (Arkin, 2002) defines a web service to interconnect Flow Objects through different
interface description language, which presents types of arrows. Swimlanes are used to group
obvious promise concerning the present require- activities into separate categories for different
ments. BPML emphasizes low-level execution functional capabilities or responsibilities (e.g.,
and contains several control flow primitives for different roles or organizational departments).
loops (foreach, while, until), branching (manual Finally, Artefacts may be added to a diagram
choice or rule based switch, join), decomposition where deemed appropriate in order to display
(all, sequential, choice), instantiation (call, spawn), further related information such as processed data
properties (assign), tools (action), exceptions or other comments.
(fault), and transactions (compensate). The ability BPDM on its side acknowledges that busi-
to define manual as well as rule based branching ness process definitions are frequently used for
743
purposes that do not required automation (for INtEGrAtING ENtErPrIsE

example, simulation and optimization of manual AND Is DEVELOPMENt AND
processes). In cases where a business process is INtErOPErAbILIty
to be (partially) automated, the BPDM enables
sufficient detail to be added to a process defini- Different approaches to model-driven develop-
tion to completely specify the process to the level ment are appropriate for supporting different
of detail that is required to generate executable types of processes, from very static, to very
runtime artifacts (e.g. by providing a mapping dynamic, even emergent processes. The differ-
from BPMN to BPEL). ent process types decide the extent to which the
There is some recognition for the need to underlying technology can be based on hard-
separate design components from run-time com- coded, predefined, evolving or implicit process
ponents for increased flexibility. This is realized models. This gives a number of development
in the WfMC’s XML Process Definition Language approaches as illustrated in Figure 3: on one
(WfMC, 2002). But even here, the separation is extreme; systems are manually coded on top of a
focused mainly for facilitating the reuse of design traditional runtime environment, and on the other
components across different workflow engines enterprise models are used directly to generate
and design tools. There is little support for user process-support solutions. In between these, we
driven interaction at run-time. have the approaches typically described in OMGs
It appears that current approaches are not Model-Driven Architecture approach (MDA),
designed with the flexibility required to accom- namely the development of Platform Independent
modate the adaptive workflows that are enabled Models (PIMs) for code-generation (e.g. on top of
by Web Services technologies. One approach a UML Virtual Machine, denoted PIM EE in the
that be recon is worth pursuing is the use of in- figure), or for Platform Specific Models (PSMs)
teractive models, which we will look on in more for more traditional code-generation or manual
detail below. implementation.
In Figure 4, we outline the different types of
interoperability-possibilities between these types
of development. Whereas traditional systems use
special APIs and approaches such as EDI for
interchange of data, on the next level (PSM), we
Figure 3. Overview of different execution environment for different process models
744
Figure 4. Interoperability between different platforms
can identify the standard Web Services Interfaces User environments are generated dynamically
(WSI). Above this level, there is a lot of work based on the definition of tasks using EEML (Ex-
being performed on specific business process tended Enterprise Modeling Language, (Krogstie,
execution platform, with a possibility to exchange 2008). Forms and components for interacting with
directly using a BPI (Business Process Interface). different model objects are selected and composed
Finally, projects such as EXTERNAL26, MAP- based on generic user interface policies and on the
PER27 and ATHENA28 have provided solutions personal and role-oriented preferences of the users.
for how to interoperate on the enterprise model The dynamically generated work management
level, potentially supporting interoperability interface includes services for work performance,
across models developed using different modeling but also for process modeling and meta-modeling.
languages and different tools (Krogstie, 2007). The model-generated workplace (MGWP) is the
The EXTERNAL project developed an in- main component in this interface. In addition to
frastructure to support networked organizations, the services for performing and managing the
defined in three basic layers: the information task, it contains links to all knowledge in the
and communication technology (ICT) layer (1); process models that is relevant for the task. Since
the knowledge representation layer (2); the work the MGWP is dynamically generated, subject
performance and management layer (3). The to personal preferences, the skill levels of task
“business end” of the infrastructure is layer 3 with performers can be taken into account. Similarly,
support for modeling and implementing customer customized MGWPs for project management
solutions, and generating work environments as can support the project management team. The
personalized and context-sensitive user interfaces contents may include an overview of the proj-
available through portals. The task performers may ect, adopted management principles, applicable
access desktop tools, organizational information methodologies, project work-break-down struc-
systems, web services, or automated processes ture, results, plans and tasks, technologies and
through this user environment. resources, status reporting and calculations (see
Jørgensen, 2004; Krogstie & Jørgensen, 2004).
745
The full power of the MGWP is enabled by service-oriented applications, as well as workflow
the first two layers, which enable interoperabil- models including both automated and interactive
ity between applications, services, and data in tasks. The overview of interoperability between
an organizational context. Layer 2 defines how different platforms, together with the example ex-
models and meta-models are represented, used planations of this, illustrates how the combination
and managed and layer 1 defines the execution of these technologies can provide more advanced
platform, software architectures, tools, software interoperability than with current systems.
components, connectivity and communication. We suggest that “interoperability” in the
The key to the integrated functionality of the in- abstract may be an untenable goal, at least in
frastructure is the consistency and interoperability the immediate future. But interoperability in the
of models and service descriptions at all relevant context of dynamic and interactive workflows,
levels. Standards and ontologies can be used across as the next best thing, is very much within our
all levels, and also between levels to make the reach. Important future work on our approach is
interoperation happen smoothly. In addition, the as indicated above related to integrating the results
interactive nature of the models, meaning that the from the current work on semantic web services,
users are free to refine them during execution, in addition to operationalizing the semantic an-
increases their potential as sources of experience notation approach.
and knowledge. As such they document details on
how the work was actually done, not only how it
was once planned. AcKNOWLEDGMENt
Clearly the complexities of such a rich frame-
work must be managed, and this is precisely the Csaba Veres was funded by the Norwegian Re-
role for the fundamental technologies reviewed search Council through the WISEMOD project
in this paper. From a unified data interchange while much of the work behind this paper was car-
format (XML) to a common data model (RDF), ried out. John Krogstie has in addition been funded
and shared conceptualizations (using OWL), it is by the projects ATHENA (http://www.athena-ip.
possible to define services, and the relationships org) and the Norwegian Research Council-project
between services and the tasks they are supposed MONESA. The ideas of this paper were pursued
to support, in a transparent and reproducible way. in the context of the EU NoE project INTEROP
Moreover, the availability of web services can and its continuation INTEROP-VLab, and we
be driven by requirements as documented in real thank other partners of INTEROP for valuable
world workflow models. Vendors could indepen- inspiration. The paper does not represent the view
dently implement solutions with a guarantee that of the funding organizations or project consortia
they will integrate with some existing application and the authors are solely responsible for the
framework. paper’s content.
cONcLUsION AND FUrtHEr WOrK rEFErENcEs
This paper has provided a survey of relevant tech- Alonso, G., Casati, F., Kuno, H., & Machiraju,
nologies for achieving semantic interoperability V. (2004). Web Services: Concepts, Architecture
in the context of enterprise information systems, and Applications. Berlin: Springer.
namely ontologies, service descriptions, and tool
support frameworks for developing and executing
746
Ambriola, V., Conradi, R., & Fuggetta, A. (1997). Berardi, D., Cabral, L., Cimpian, E., Domingue, J.,
Assessing Process-Centered Software Engineer- Mecella, M., Stollberg, M., & Sycara, K. (2005).
ing Environments. ACM Transactions on Software ESWC Semantic Web Services Tutorial. Retrieved
Engineering and Methodology, 6(3), 283–328. 15 Feb, 2006, from http://stadium.open.ac.uk/dip/
doi:10.1145/258077.258080
Berners-Lee, T., Hendler, J., & Lassila, O. (2001).
Ankolekar, A., Martin, D., McGuinness, D., The Semantic Web. Scientific American, 284(5),
McIlraith, S., Paolucci, M., & Parsia, B. (2004). 34–43.
OWL-S’Relationship to Selected Other Technolo-
Bolcer, G. A., & Kaiser, G. (1999). SWAP: Lever-
gies, Technical report, W3C Member Submission
aging the Web To Manage Workflow. IEEE Internet
22 November 2004. Retrieved 1 Feb, 2006, from
Computing, 3(1), 85–88. doi:10.1109/4236.747328
http://www.w3.org/Submission/OWL-S-related/
Borgida, A., & Brachman, R. (2003). Conceptual
Antoniou, G., & van Harmelen, F. (2004). Web
Modeling with Description Logics. In F. Baader,
Ontology Language: OWL. In S. Staab & R.
D. Calvanese, D. McGuinness, D. Nardi, & P.
Studer (Eds.), Handbook on Ontologies (pp. 67-
Patel-Schneider (Eds.) The Description Logic
92). Berlin: Springer.
Handbook: Theory, Implementation and Applica-
Arkin, A. (2002). Business Process Modelling Lan- tions. Cambridge University Press.
guage. Retrieved 23 Aug, 2003, from http://www.
BPMI.org and OMG. (2006). Business Process
bpmi.org/bpmi-downloads/BPML-SPEC-1.0.zip
Modeling Notation Specification. Final Adopted
Arkin, A., Askary, S., Bloch, B., Curbera, F., Specification. Object Management Group, http://
Goland, Y., Kartha, N., et al. (Eds.). (2005) Web www.bpmn.org (February 20, 2006).
Services Business Process Execution Language
Brickley, D. (2001). RDF: Understanding the
Version 2.0, Technical report, OASIS Open, Inc.,
Striped RDF/XML Syntax. Retrieved 25 Sep,
Committee Draft, 21 Dec, 2005. Retrieved 15 Feb
2002, from http://www.w3.org/2001/10/stripes/.
2006 from http://www.oasis-open.org/commit-
tees/download.php/16024/wsbpel-specification- Broekstra, J., Kampman, A., & van Harmelen, F.
draft-Dec-22-2005.htm (2003). Sesame: An Architecture for Storin gand
Querying RDF Data and Schema Information.
Bandinelli, S., Fuggetta, A., Lavazza, L., Loi, M.,
In D. Fensel, J. A. Hendler, H. Lieberman & W.
& Picco, G. (1995). Modelling and Improving
Wahlster (Eds.), Spinning the Semantic Web:
an Industrial Software Process. IEEE Transac-
Bringing the World Wide Web to Its Full Potential
tions on Software Engineering, 21(5), 440–454.
[outcome of a Dagstuhl seminar] (pp. 197-222).
doi:10.1109/32.387473
Barros, A., Dumas, M., & Oaks, P. (2005). A Criti-
Bubenko, J., Jr. (2007). From Information Alge-
cal Overview of the Web Services Choreography
bra to Enterprise Modelling and Ontologies – A
Description Language. BPTrends (www.bptrends.
Historical Perspective on Modelling for Infor-
com), March 2005, pp 1-24.
mation Systems. In J. Krogstie, A. L. Opdahl &
Batzarov, Z. (2004). Orbis Latinus: Linguistic S. Brinkkemper (Eds.): Conceptual Modelling
Terms. Retrieved 3 Apr, 2005, from http://www.or- in Information Systems Engineering. Berlin:
bilat.com/General_References/Linguistic_Terms. Springer, pp. 1-18.
html
747
Butler, H. (2002). Barriers to real world adoption Daconta, M., Orbst, L., & Smith, K. (2003). The
of Semantic Web technologies. Hewlett-Packard. Semantic Web: A guide to the future of XML,
Web Services and Knowledge Management.
Cabral, L., Domingue, J., Galizia, S., Gugliotta,
London: Wiley.
A., Tanasescu, V., Pedrinaci, C., & Norton, B.
(2006). IRS-III: A broker for semantic web services Derniame, J.-C., Kaba, B. A., & Wastell, D. G.
based applications. In I. Cruz et al. (Eds.). Proc. (Eds.). (1998). Software Process: Principles,
ISWC’06, LNCS 4273,201-214. Methodology, Technology. Berlin: Springer
(LNCS 1500).
Carlsen, S. (1997). Conceptual Modelling and
Composition of Flexible Workflow Models. PhD Dijkman, R., & Dumas, M. (2004). Service-
thesis, Dept of Computer and Inforamtion Science, Oriented Design. A Multi-Viewpoint Approach.
Norwegian University of Science and Technology, International Journal of Cooperative Informa-
Trondheim, Norway. tion Systems, 13(4), 337–368. doi:10.1142/
S0218843004001012
Carlsen, S. (1998). Action Port Model: A Mixed
Paradigm Conceptual Workflow Modeling Lan- Fischer, L. (Ed.). (2001). The Workflow Handbook
guage. Proceedings of the 3rd IFCIS International 2001. Lighthouse Point, FL: Workflow Manage-
Conference on Cooperative Information Systems ment Coalition (WfMC).
(CoopIS’98), pp. 300-308. Los Alamitos, CA:
Fox, M., & Grüninger, M. (1998). Enterprise
IEEE CS Press.
Modeling. AI Magazine, 19(3), 109–121.
Cavantzas, N., Burdett, D., Ritzinger, G., Fletcher,
Green, P., & Rosemann, M. (2000). Integrated
T., Lafon, Y., & Barreto, C. (Eds.). (2005). Web
Process Modelling: An Ontological Evaluation.
Services Choreography Description Language
Information Systems, 25(2), 73–87. doi:10.1016/
Version 1.0. Technical report, W3C Candidate
S0306-4379(00)00010-7
Recommendation, 9 Nov, 2005. Retrieved 10
Feb, 2006, from http://www.w3.org/TR/2005/ Haake, J. M., & Wang, W. (1997). Flexible sup-
CR-ws-cdl-10-20051109/ port for business processes: extending cooperative
hypermedia with process support. In Proceedings
Coalition, O. W. L.-S. (2004). OWL-S 1.1 Release.
of GROUP’97, International Conference on Sup-
Retrieved 9 Aug, 2005, from http://www.daml.
porting Group Work. The Integration Challenge.
org/services/owl-s/
Hayes, P. (2004). RDF Semantics. Technical re-
Conradi, R., & Jaccheri, L. (1998). Process
port, W3C, 10 Feb 2004. Retrieved Mar 3, 2005,
Modelling Languages. In J.-C. Derniame, B. A.
from http://www.w3.org/TR/rdf-mt/
Kaba & D. G. Wastell (Eds.), Software Process:
Principles, Methodology, and Techniques (pp. Hess, A., Johnston, E., & Kushmerick, N. (2004).
27-52). Berlin: Springer LNCS 1500. ASSAM: A Tool for Semi-automatically Annotat-
ing Semantic Web Services. In S.A. McIlraith et
Curtis, B., Kellner, M., & Over, J. (1992). Process
al. (Eds.). Proc. ISWC’04, Springer LNCS 3298,
Modeling. Communications of the ACM, 35(9),
320-334.
75–90. doi:10.1145/130994.130998
748
Jørgensen, H. D. (2001). Interaction as a Frame- Krogstie, J. (2007). Modelling of the People, by

work for Flexible Workflow Modelling. In: C. Ellis the People, for the People. In J. Krogstie, A. L.
& I. Zigurs (Eds.), Proceedings of the International Opdahl & S. Brinkkemper (Eds.), Conceptual
ACM SIGGROUP Conference on Supporting Modelling in Information Systems Engineering.
Group Work 2001. September 30 - October 3, Berlin: Springer, pp. 305-318.
2001, Boulder, Colorado, USA. p.32-41.
Krogstie, J. (2008). Integrated Goal, Data and
Jørgensen, H. D. (2003). Model-Driven Work Process modeling: From TEMPORA to Model-
Management Services. In R. Jardim-Goncalves, Generated Work-Places. In, Johannesson and
H. Cha, A. Steiger-Garcao (Eds.), Proceedings Søderstrøm, (eds.), Information Systems Engi-
of the 10th International Conference on Concur- neering. IGI Publishing 2008.
rent Engineering (CE 2003), July 2003, Madeira,
Krogstie, J., & Jørgensen, H. (2004). Interactive
Portugal. A.A. Balkema Publishers.
Models for Supporting Networked Organizations.
Jørgensen, H. D. (2004). Interactive Process In A. Persson & J. Stirna (Eds.), Advanced Infor-
Models. PhD thesis, Department of Computer mation Systems Engineering, 16th International
and Information Science, Norwegian University Conference (CAiSE’04), Berlin: Springer (LNCS
of Science and Technology, Trondheim, Norway. 3084).
Jørgensen, H. D., & Carlsen, S. (1999) Emergent Kunz, J., Christiansen, T., Cohen, G., Jin, Y.,
Workflow: Integrated Planning and Performance & Levitt, R. (1998). The virtual design team: A
of Process Instances. In J. Becker, M. zur Mühlen, Computational simulation model of project orga-
M. Rosemann (Eds.) Proceedings of the 1999 nizations. Communications of the ACM, 41(11),
Workflow Management Conference: Workflow- 84–92. doi:10.1145/287831.287844
based Applications, 9 Nov, Univ., Münster, Ger-
Lei, Y., & Singh, M. (1997). A Comparison of
many, pp 98-116.
Workflow Metamodels. Paper presented at the
Klein, M., Broekstra, J., Fensel, F., van Harmelen, ER’97 Workshop on Behavioral Models and De-
F., & Horrocks, I. (2003). Ontologies and Schema sign Transformations: Issues and Opportunities
Languages on the Web. In D. Fensel, J. A. Hendler, in Conceptual Modeling, Los Angeles, CA.
H. Lieberman & W. Wahlster (Eds.), Spinning the
Lillehagen, F. (1999). Visual extended enterprise
Semantic Web: Bringing the World Wide Web to
engineering embedding knowledge management
Its Full Potential [outcome of a Dagstuhl seminar]
systems engineering and work execution. IFIP
(pp. 95-139). Cambridge, MA: MIT Press.
International Enterprise Modeling Conference
Krogstie, J. (2004). Integrating Enterprise and IS (IEMC’99), Verdal, Norway.
Development Using a Model-Driven Approach.
Linthicum, D. (2003). Next Generation Applica-
In O. Vasilecas, A. Caplinskas, G. Wojtkowski,
tion Integration: From Simple Information to Web
W. Wojtkowski, J. Zupancic (Eds.) Informa-
Services. Boston: Addison-Wesley.
tion Systems Development Advances in Theory,
Practice, and Education. (Proc. ISD’04). Boston, Loos, P., & Allweyer, T. (1998). Process Orienta-
MA: Kluwer. tion and Object Orientation - An Approach for In-
tegrating UML (Technical Report). Saarbrücken,
Germany: Institut für Wirtschaftsinformatik,
University of Saarland.
749
Manola, F., & Miller, E. (2004, 10 Feb). RDF Smith, M., Welty, C., & McGuinness, D. L. (Eds.).
Primer. Retrieved 15 Aug, 2005, from http://www. (2004, 10 Feb). OWL Web Ontology Language
w3.org/TR/rdf-primer/ Guide. Retrieved 25 Feb, 2005, from http://www.
w3.org/TR/owl-guide/
McGuinness, D. L. (2003). Ontologies Come of
Age. In D. Fensel, J. A. Hendler, H. Lieberman & Staab, S., & Studer, R. (Eds.). (2004). Handbook
W. Wahlster (Eds.), Spinning the Semantic Web: on Ontologies. Berlin, Germany: Springer.
Bringing the World Wide Web to Its Full Potential
van der Aalst, W. M. P. (1999). Formalization
[outcome of a Dagstuhl seminar] (pp. 171-195).
and Verification of Event-driven Process Chains.
Information and Software Technology, 41(10),
Miles, A. (2006, February). 639–650. doi:10.1016/S0950-5849(99)00016-6
RDFMolecules:Evaluating Semantic Web Tech-
Wegner, P., & Goldin, D. (1999). Interaction
nology in a Scientific Application. Available at
as a Framework for Modeling. In P. P. Chen, J.
http://www.w3c.rl.ac.uk/SWAD/ papers/RDF-
Akoka, H. Kangassalo, & B. Thalheim (Eds.),
Molecules_final.doc (Feb 20, 2006)
Conceptual Modeling. Berlin: Springer (LNCS
Natvig, M. K., & Ohren, O. (1999). Modeling 1565), pp. 100-114
shared information spaces (SIS). In GROUP
WfMC. (1999). Workflow Management Coali-
‘99: Proceedings of the international ACM SIG-
tion Interface 1: Process Definition Interchange
GROUP conference on Supporting group work,
Process Model (Technical report No. WfMC
Phoenix, AZ, Nov 14-17, pp. 99-108. New York:
TC-1016-P). Lighthouse Point, FL: Workflow
ACM Press.
Management Coalition.[gs1]
OMG. (2000). Workflow Management Facil-
WfMC. (2002). Workflow Process Definition In-
ity Specification, v.1.2. Needham, MA: Object
terface - XML Process Definition Language (Tech-
Management Group.
nical report No. WFMC-TC-1025). Lighthouse
Opdahl, A. L., & Sindre, G. (2007). Interoperable Point, FL: Workflow Management Coalition.
Management of Conceptual Models, In J. Krogstie,
WSMO Working Group. (2005). The Web Service
A. L. Opdahl & S. Brinkkemper (Eds.), Conceptual
Modeling Language WSML. Retrieved Aug 20,
Modelling in Information Systems Engineering.
2005, from http://www.wsmo.org/wsml/wsml-
Berlin: Springer, pp. 75-90.
syntax
Powers, S. (2003). Practical RDF. Sebastopol,
CA: O’Reilly.
Scheer, A.-W., & Nuttgens, M. (2000). ARIS ENDNOtEs
Architecture and Reference Models for Business
Process Management. In W. M. P. van der Aalst,
1
http://is.uib.no/wiki/UEML/UEML
J. Desel & A. Oberweis (Eds.), Business Process
2
http://www.w3.org/TR/rdf-syntax-gram-
Management (pp. 376-390). Berlin, Germany: mar/
Springer (LNCS 1806).
3
http://www.w3.org/DesignIssues/Notation3
4
http://www.w3.org/TR/2004/REC-rdf-
testcases-20040210/#ntriples
5
http://www.dajobe.org/2004/01/turtle/
750
6
http://www.w3.org/2001/ sw/BestPractices/ 20
http://kmi.open.ac.uk/projects/webonto/
HTML/ 2006-01-24-rdfa-primer 21
http://kmi.open.ac.uk/projects/ocml/
7
http://www.mindswap.org/2005/SMORE/ 22
http://www.daml.org/services/ owl-s/tools.
8
http://microformats.org/ html, http:// owlseditor.semwebcentral.org/
9
http://web.resource.org/rss/1.0/spec 23
http://staff.um.edu.mt/cabe2/ supervising/
10
http://www.rssboard.org/rss-specification undergraduate/ owlseditFYP/OwlSEdit.
11
http://www.radarnetworks.com/ html
12
http://www.w3.org/2001/sw/WebOnt/ 24
http://alphaworks.ibm.com/tech/ettk
13
http://lsdis.cs.uga.edu/projects/meteor-s/ 25
http.//owlsm.projects.semwebcentral.org/
14
http://www.wsmo.org/ 26
http://research.dnv.com/external/default.
15
http://kmi.open.ac.uk/projects/irs/ htm
16
http://www.daml.org/services/owl-s/ 27
http://193.71.42.92/ websolu-
17
http://lsdis.cs.uga.edu/ projects/meteor-s/ tion/UI/ Troux/07 /Default.
downloads/mwsaf/ asp?WebID=260&PageID=1
18
http://www.w3.org/Submission/WSDL-S/ 28
http://www.athenaip.org
19
http://sdk.semanticweb.org/index.html
This work was previously published in Global Implications of Modern Enterprise Information Systems: Technologies and
Applications, edited by Angappa Gunasekaran, pp. 172-194, copyright 2009 by Information Science Reference (an imprint of
IGI Global).
751
752
Chapter 3.13
In-House vs. Off-the-Shelf
e-HRM Applications
Nawaf Al-Ibraheem
KNET, Kuwait
Huub Ruël
University of Twente, The Netherlands & American University of Beirut, Lebanon
AbstrAct make the most effective choice between the two

systems. It was found through this research that fac-
Companies new to the e-HRM technologies are tors such as continuous user involvement, effective
overwhelmed by the dilemma of choosing either communication, and strong change management are
the ready-made, off-the-shelf e-HRM systems, or most considered by companies that develop e-HRM
develop their own e-HRM systems in house in order in house, while advocates of off-the-shelf e-HRM
to implement the e-HR transformation. Therefore, systems are most affected by success factors such as
this research was done to shed some light on the business process reengineering, planning and vision,
differences and similarities between off-the-shelf and project management. Another finding was that
e-HRM systems and in-house developed ones, with increasing efficiency, providing customer-oriented
regards to some elements developed in a preliminary service excellence, and improving self services were
framework, such as the implementation and develop- top goals accomplished by companies developing
ment approaches, e-HRM activities they facilitated, their e-HRM system in house. These findings, be-
application types and characteristics, and e-HRM side many other ones discovered in this research,
outcome and benefits. This comparison provided would help companies decide which system best
insightful information that could help companies fits their needs and accomplish high levels of ef-
fectiveness gained from the transformation of their
DOI: 10.4018/978-1-60566-304-3.ch006 HR function to e-HR.
In-House vs. Off-the-Shelf e-HRM Applications
INtrODUctION through internet-technology based applications.

Ruël et al (2004) define e-HRM as a way of con-
Human Resources Management has always been sciously implementing HRM practices, policies
the first business process to use the emerging tech- and strategies supported by or fully delivered
nologies of the new era. As a matter of fact, Payroll through internet-technology based applications.
Administration is known to be one of the earliest Terms like Management Self-Service (MSS) and
business processes to be automated (Lengnick- Employee Self Service (ESS) started to be used
Hall and Mortiz 2003, p.365). Through the use as well. The purposes of implementing e-HRM
of technology and information systems, human systems broadened in comparison to those con-
resources departments of companies around the nected to HRISs.
world are able to use computers to log employ- During the e-commerce era in the 1990s, the
ees’ data and interact electronically with them. term e-HRM emerged to basically refer to con-
These functions and more are offered to Human ducting Human Resources Management functions
Resources personnel and other employees through using the Internet (Lengnick-Hall and Mortiz 2003,
what is called the Electronic Human Resources p. 365). Nowadays, many organizations are imple-
Management, or e-HRM. menting e-HRM within the strategic design of their
Since the 1960s many firms (large and small) all core businesses. Some Organizations require the
over the globe have been implementing IT-based use of standardized HR management tools such as
Human Resources Management applications in payroll, employee benefits, recruitment, training,
order to reduce the amount of associated costs. etc. These organizations mainly refer to off-the-
Ball (2001) pointed out that by 1998, 60 percent shelf e-HRM solutions offered by third party
of the Fortune 500 companies used a Human such as Oracle, PeopleSoft, SAP, or IBM, where
Resources Information System (HRIS) to support they perceive efficiency and fulfilment in these
daily HRM operations. The main benefits of HRISs ready-made systems. Others require customized
are focused on improved accuracy, the provision Human Resources Information Systems (HRIS)
of timely and quick access to information, and tailored to best fit their business’s needs. One of
the savings of costs (Ngai & Wat, 2006 based on the key advantages of in-house developed HRISs
Lederer, 1984; Tetz, 1973; Wille & Hammond, is that they save HR staff time in dealing with
1981). For the human resources activities, e-HR the elements of the application as they already
has the potential to enhance efficiency by reducing know and understand the parameters of their own
cycle times for processing paperwork, increasing software (Thaler-Carter 1998, p.22).
data accuracy, and reducing human resources One objective therefore becomes the main
workforce (Lengnick-Hall and Mortiz, 2003). purpose of this research and that is to compare and
Effectiveness can also be influenced by empower- contrast the in-house developed and off-the-shelf
ing both employees and managers to make better, e-HRM systems with regards to their implementa-
accurate, and timely decisions (Lengnick-Hall tion, associated costs, usage, and effectiveness.
and Mortiz, 2003).
HRIS started to be more internet-technology
based since the second half of 1990’s, where its LItErAtUrE bAcKGrOUND
aim was not only to support the HR department
itself, but to target managers and employees’ ef- An essential step in e-HRM systems implementa-
fectiveness. The term electronic Human Resources tion is the decision whether to buy an off-the-shelf
Management (e-HRM) was coined and has become e-HRM software package (in many cases part of
a dominating label for HRM services delivered an Enterprise Resource Planning (ERP) package
753
like Oracle/Peoplesoft or SAP) or to develop e- system, while off-the-shelf software would cause
HRM systems in-house. As Kovach et al. (2002) difficulties during the integration phase.
describe, a small company might prefer to depend In this study, the authors will concentrate on the
on consultants and outsource all aspects of design, issue of the differences and similarities between
implementation, and operation, while a large com- off-the-shelf and in-house developed e-HRM
pany could prefer to opt for in-house development implementations. We define implementation
of e-HRM systems by their own IT department. broadly, from the first initiation to the evaluation
Widely believed benefits of off-the-shelf systems of results of e-HRM, and semi-technical (i.e.
in general are time-saving, proven system quality, type of applications/functionality/design) and
and availability of expertise through the provider. organizational issues (i.e. change management,
On the contrary, disadvantages of off-the-shelf HR restructuring, etc).
systems are the lack of customization/flexibility Technology plays the role of enabler that helps
and dependency on vendor-expertise. In order HR to deliver the workforce support and manage-
to avoid these down-sides companies might opt ment based on business needs (Gueutal & Stone
for in-house development of an HRIS system. 2005, p. 2). New web technologies have enabled
Exact figures regarding the differences/similari- HR to perform certain paperless tasks that used
ties between off-the-shelf e-HRM and in-house to consume time and human resources in the old
developed e-HRM, do not really exist. ages of HR. Furthermore, technology-enabled
To take further the point issued by Kovach et HR functions have evolved into human capital
al (2002), in case of e-HRM, the acceptance and management (HCM), as Gueutal and Stone (2005,
usage of applications by employees and managers p. 2) agree with Ruël et al. in stating that HCM
is critical to successful implementation and the and e-HR are the responsibilities of everyone in
generation of results. That makes the decision an organization, from employees to executives.
whether to ‘buy or make’ even more important, Nevertheless, Hendrickson (2003, p. 383) stresses
and seems to favor in-house development over the fact that this change in information technology
off-the-shelf implementation. However funded has created significant challenges for HR profes-
knowledge on this issue is lacking so far. sionals to get updated with the latest information
The annual CedarCrestone survey (2006) technology while simultaneously transforming
shows that many organizations lay out more HR processes to e-processes.
expenses for outsourced solutions (mainly off- From these definitions arises a general un-
the-shelf) than they do for in-house development. derstanding of what e-HRM is. In this research,
Three main reasons are explained: organizations e-HRM is the management of an HR network
with in-house software development mostly have that connects employees, managers, and HR staff
fragmented system operations which cause them through web-technology-enabled channels that
to overlook embedded costs within distributed provide them with an electronic access to human
operations; outsourcing assures better view and resources transactions, strategies, policies, and
control of costs; and outsourcing is a solution for practices.
organizations that struggle to control costs due to
the complexity of their operations (HR Technology benefits, Goals, and
Trends to Watch in 2007 2007, p. 1). On the other Objectives of e-HrM
hand, the CedarCrestone survey (2006) notes that
some respondents sense that in-house HR solutions The fast changing world of business stipulates
positively affect employees’ productivity and can organizational flexibility, capability, and rapid
be easily and swiftly integrated into the current response in order to succeed (Lepak & Snell
754
1998, p. 215). Therefore, e-HRM must attain costs it incurs during and after the implementa-
objectives such as strategy-oriented focus, flex- tion stage such as employee turnover, resistance,
ibility, cost containment, and service-providing and confusion.
excellence (Lepak & Snell 1998, p. 215). Strategic A typical agreement by most authors proves
orientation can be achieved by concentrating on that efficiency is a top goal considered by many
the transformational HR activities mentioned in organizations that use e-HRM. Cost reduction
section 2.2, while flexibility can be reached by is also agreed upon to be the main benefit of
updating HR processes, policies, and practices e-HRM. Other benefits might not be as notice-
to counter the rapidly changing business world. able or tangible as efficiency. Lengnick-Hall and
Efficiency, on the other hand, can be achieved by Moritz (2003) divide the outcome of e-HRM into
many ways such as staff reduction, elimination tangible and intangible benefits. Tangible benefits
of manual processes, and timeliness. However, include process and administrative cost reduction,
cost reduction can only be considered beneficial staff reduction, transaction processing speed-up,
if there is no loss of effectiveness or deterioration elimination of information errors, and improve-
of customer service excellence (Schramm 2006, ment in tracking and controlling HR activities
p. 3). Hendrickson (2003) agrees that efficiency (Lengnick-Hall & Moritz 2003, p. 369). Intangible
is one of the goals of implementing e-HRM that benefits are mostly associated with the indirect
can be rendered by increasing number of HR effect of e-HRM on the HR function. Such ben-
transactions and timeliness without affecting efits are improvement in employee productivity
needed resources. and morale, decision making, and information
Another benefit of using e-HRM is the in- sharing. Innovation enhancement and time-to-
creased effectiveness of human resources manage- market acceleration might also be considered as
ment through the use of technology which sim- intangible benefits of e-HRM (Lengnick-Hall &
plifies HR processes and enhances performance Moritz 2003, p. 369).To summarize this section,
(Hendrickson 2003, p. 385). Beer et al. (1984) the author will consider the following main goals
distinguish four different benefits or outcomes of implementing e-HRM within organizations:
from e-HRM: commitment, competence, con-
gruence, and cost effectiveness. Commitment is • Increase efficiency by reducing cost and
achieved by motivating employees to interact with eliminating unnecessary functions.
managers cohesively during changes to achieve • Provide management and employee self-
organizational goals (Ruël et al. 2004, p. 369). service mechanisms.
High competence is when employees learn new • Improve HR’s strategic focus.
tasks and gain competitive skills to enhance their • Achieve client service excellence.
contribution, effectiveness, and creativity. Con-
gruence is also another benefit of implementing
e-HRM that involves structuring the business from On the other hand, different outcomes can be
the inside out in the best interest of stakeholders gained from implementing e-HRM within an
(Ruël et al. 2004, p. 369). Cost effectiveness aims organization. For this research, the following
at increasing the level of economical value based outcomes will be considered:
on the tangible and intangible benefits produced
by the amount of resources spent (i.e. money, time, • Commitment.
human resources, etc). A cost effective e-HRM • Competence.
system, however, is one that provides adequate • Cost Effectiveness.
e-HRM services for the reasonable HR-related • Congruence.
755
e-HrM Applications E-HRM applications are the software programs

that offer a useful and easy-to-use electronic
e-HRM applications (as depicted in Figure 1) medium, through which the e-HRM goals are
represent the outmost layer of the human re- accomplished by performing different types of
sources management model suggested in this human resources activities electronically to yield
research. This model suggests that the outer layers the desired outcome and benefits.
comprise all of the elements of the inner ones.
Therefore, the application layer includes all the This definition explains the layered approach
basic requirements and functions of HRM and shown in Figure 1 but, nevertheless, it doesn’t
e-HRM including its types, goals, and benefits. explain the types of these software applications.
E-HRM applications are wide in range and they The following section suggests a few types of
provide a variety of automated HR activities that e-HRM applications that have been discussed in
enchant the HR function with flexibility and ease previous literature.
of use. In an exclusive survey by IOMA (2002)
professionals reported that e-HRM applications types of e-HrM Applications
which are most preferred by users are those that
possess high levels of integration and processing e-HRM applications can be specified under dif-
capabilities, user friendliness, robustness, and ferent categories according to the HR activities
reporting. Furthermore, e-HRM studies with they facilitate electronically. Seven types have
emphasis on the Technology Acceptance Model been identified by Florkowski and Olivas-Luján
(Davis et al. 1989) has exhausted the use and (2006) and divided into two predominant groups:
outcome of e-HRM applications and determined software applications that target HR staff as end
that they are mainly dependant on the usefulness users, and those that target internal staff (manag-
and ease of use of such applications (Ruta 2005). ers and employees) as end users. The types that
On one hand, usefulness is defined by Davis et al. target HR staff are HR functional applications
(1989, p. 985) as “the prospective user’s subjective and integrated HR suite applications. Those that
probability that using a specific application system focus on managers and employees as end users
will increase his or her job performance within are Interactive Voice Response (IVR) systems, HR
an organizational context”. This means in order intranet applications, self-service applications,
for the e-HRM system to be perceived by users HR extranet applications, HR portal applications
as useful, positive impact on the organizational
performance of an employee must be observed Figure 1. e-HRM layers
during or after the use of the e-HRM applications.
Davis et al. (1989, p. 985), on the other hand,
define the perceived ease of use as “the degree
to which the prospective user expects the target
system to be free of effort.” E-HRM applica-
tions could be perceived by employees as “easy
to use” if it doesn’t require effort to be operated.
The combination of e-HRM definition produced
in section 2.1 with the general understanding of
e-HRM applications mentioned here yields a
predominant meaning for e-HRM applications:
756
(Florkowski & Olivas-Luján 2006, p. 687).The of its HR activities with the help of the application
CedarCrestone annual report (2007), on the other sets mentioned above (CedarCrestone annual re-
hand, divides over 30 e-HRM applications into port 2007). On the contrary, CedarCrestone reports
the following groups: that the least business categories to adopt e-HRM
Administrative applications, including core technologies are Retail and Public Administration.
HR functions such as payroll, benefits, and HR The top players in e-HRM application software
management system. This category handles most market are Oracle/PeopleSoft, ADP, SAP, and
of the operational type of HR activities—men- Kronos, as most of the applications mentioned
tioned under section 2.2—that are administered above are supported by these giants and favored
by the HR staff. The report identifies three major by a plethora of HR departments around the world.
activities that belong to this group as payroll, HR
management system, and benefits administration e-HrM Applications Implementation
(CedarCrestone annual report 2007). and Development Approaches
Employee and manager productivity appli-
cations which provide self service transactions The implementation and development of the
that aim at improving service delivery, reducing e-HRM application doesn’t require a set of criti-
cost, and enabling employees and managers to cal success factors or approaches different than
concentrate more on core processes (CedarCre- those constituted by Enterprise Resource Plan-
stone annual report 2007). These applications ning (ERP) projects or large enterprise portal
also provide access to operational HR activities projects, as e-HRM is considered a part of ERP
performed only by managers and employees implementation itself. These large projects differ
such as benefit-related self service (BSS), simple from traditional information system projects in
management reporting (SMR), pay-related self scale, scope, complexity, organizational changes,
services (PSS), employee self services (ESS), time costs, and need for business process reengineering
management self service (TMSS), total benefits (Remus 2007, p. 538). As a result, technical and
statements (TBS), HR-oriented help desk, and business difficulties during the implementation
manager self service (MSS). and development of these systems (including
Strategic HCM applications, such as talent e-HRM systems) are imminent. Literature has
management, training enrollment, competency it that these difficulties are widely spread, but
management, e-learning, compensation, perfor- only fragmented research was done regarding the
mance, succession planning, and career planning. critical success factors and approaches to ERP
These applications transform relational HR activi- implementation (Nah et al. 2001, p. 286). These
ties into their electronic version. approaches were narrowed down to the ones shown
Business intelligence applications comprise in Table 2, which have been previously identified
HR data warehousing, operational reporting, HR by academic articles as shown in the table.
scorecard, and analytics that help develop the stra-
tegic focus of an organization. These applications User Involvement
are the protagonists of e-HRM. They facilitate the
transformational activities that help develop the User involvement in the early stages of the imple-
strategic orientation of the HR function. mentation and development phases of the e-HRM
Looking at the adoption level of e-HRM ap- system allows users to make adjustments to the
plications, Financial Services seem to top other system to satisfy their needs. Consequently, orga-
businesses in leading the e-commerce market of nizational resistance to the new changes implied
human resources management by automating most by the use of the e-HRM applications is minimized
757
Table 1. e-HRM approaches
Lee & Lee (2001) Nah et al. (2001) Al-Sehali (2000) Remus (2007)
User involvement   
Business process reengi-
  
neering
Planning & strategy    
Training &
  
education
Change
  
management
Top management support   
Effective
 
communication
Project
 
management
Table 2. Research constructs
Constructs/Variables Description
Overview This strategy is used to start a case study research by collecting general information about the organiza-
tion to be researched and the e-HRM system it uses.
Implementations and Develop- The factors affecting the implementation and development of e-HRM system and how the company
ment Approaches deals with them
E-HRM Goals The main goals of implementing the e-HRM system and how they coincide with corporate strategy
E-HRM Activities The HR-related activities offered by the new e-HRM system and how employees work with them
E-HRM Application Charac- The characteristics of the e-HRM software application as defined by ISO 9126 standards
teristics
E-HRM Outcomes The four C’s of HR: commitment, competence, congruence, and cost effectiveness and how they are
impacted after the implementation of the e-HRM system
HRIS Benefits Usefulness and ease of use of the new e-HRM system
and customer satisfaction is increased (Lee & Lee for the e-HRM system to be more effective. Such
2001, p. 208). Change management processes in reengineering mechanism is applied when trans-
this case are minimized and easily controlled. forming HR manual processes to paperless forms.
Reengineering should begin before choosing the
Business Process Reengineering software system to make sure changes are ac-
cepted by the stakeholders and the processes can
Achieving benefits through the implementation actually be aligned with the new system (Nah et
of an enterprise system is impossible without the al. 2001, p. 294).
inevitable alignment of processes and activities
with the new system requirements (Remus 2007; Planning and Vision
Bingi et al. 1999). This means that when a com-
pany implements a new e-HRM system, some of A solid business plan needs to be defined in
the HR processes must be reengineered in order order for the ERP project to be directed toward
758
the proposed strategic and tangible benefits, re- would positively assist the implementation effort
sources, costs, risks, and timeline (Buckhout et (Nah et al. 2001, p. 293). Training and education
al. 1999). This means for e-HRM implementation is a critical step in managing change itself, as
to be successful, a plan must be agreed upon by employees must be educated about the new system
the project manager or the responsible parties to to understand how it changes business processes
follow during the project life cycle. This plan will (Nah et al. 2001, p. 293).
guarantee the alignment of the e-HRM goals and
strategy with the HR and corporate strategies to Top Management Support
ensure maximum effectiveness, integration, and
alignment. Lee and Lee (2001) insist that good One of the most critical success factors for imple-
planning consumes a considerable amount of menting an ERP system is the support and involve-
time prior to implementation. This ensures that ment of top managers in the project during its life
the e-HRM system is thoroughly exploited and cycle (Al-Sehali 2000, p. 79). In order for e-HRM
efficiently implemented to coincide with the implementation to be successful, top managers have
corporate strategies. to approve and continuously support the responsible
parties during the implementation stage to make
Training and Education sure no obstacles prevent or delay the progress of
the project. Also, an executive sponsor should be
Since the e-HRM system offers new methods appointed to coordinate, communicate, and integrate
of processing transformed or new HR activities, all aspects of the project between the development
proper training must be given to all users of the team and top management (Remus 2007, p. 544). The
system. This becomes crucial since the new in- executive sponsor should communicate, integrate,
terface provides functionality that has never been and approve the shared vision of the organization
used before and needs to be related to the newly and the responsibilities and structures of the new
reengineered business processes (Remus 2007, e-HRM system (Nah et al. 2001, p. 291).
p. 544). Education is the catalyst that brings the
knowledge of the users up to the point where they Effective Communication
can familiarize themselves with the new e-HRM
system quickly and sufficiently. Interdepartmental communication as well as com-
munication with customers and business partners
Change Management is a key element in the success of implementing an
ERP system (Remus 2007, p.544). Communica-
Managing change within the organization could tion helps employees and involved parties better
be a full time job by itself, as it requires the understand the new e-HRM system to keep up
management of people and their expectations, with the development and implementation stages
resistance to change, confusion, redundancies, of the project. Employees should also be informed
and errors (Remus 2007, p. 541). For the e-HRM in advance the scope, objectives, activities, and
to be successfully implemented, the organization updates implemented by the new system in order
should realize the impact of this new change on to meet their expectations (Nah et al. 2001, p. 291)
employees, managers, and HR staff and under-
stand its dimensions in order to manage the effects Project Management
with a corporate strategy that is open to change.
Furthermore, emphasis on quality, computing abil- Managing the implementation and development
ity, and willingness to accept the new technology of e-HRM system is a crucial step toward suc-
759
cessful results. The scope of the project must and human resources. Second, risk is minimized
be clearly defined, including aspects such as the by the fact that COTS systems are professionally
amount of systems implemented, involvement of provided and supported by experienced entities.
business units, and the amount of business process Third, COTS vendors take responsibilities of
reengineering needed (Nah et al. 2001, p. 292). A supporting and maintaining the software, which
company must assign a project manager to lead leads to reducing the cost of supporting the soft-
the project of developing and implementing an e- ware itself. Fourth, COTS systems are virtually
HRM system professionally according to profound up to date as vendors keep investing in develop-
business rules. The project itself must have clearly ing the software itself (Silverman 2006, p 10).
defined business and technical objectives and goals It is also known as a fact that most IT managers
corresponding to the project deliverables (Remus first evaluate COTS software when money and
2007, p. 543). Such E-HRM goals (mentioned in time-to-market are at the top of the priority list
section 2.3) are embedded within the different (Traylor 2006, p 20). This evidence makes it clear
e-HRM activities (mentioned in section 2.2) as that off-the-shelf software saves on time and
part of the project management scheme. money resources. On the contrary, disadvantages
of off-the-shelf systems are illustrated by the lack
recent studies on e-HrM of customization/flexibility and the dependency
Application Implementation on vendor-expertise. Amons and Howard (2004)
insist that buying an already-made software pack-
An essential step in e-HRM implementation and ages will increase time and other resources spent
development is the decision whether to buy an on the integration and deployment phases of the IT
off-the-shelf e-HRM systems (in many cases project. Indeed, COTS systems need to be perfectly
part of an Enterprise Resource Planning (ERP) aligned and integrated within the practices of the
package such as Oracle/Peoplesoft or SAP), or to organization, and deployed in a matter that best fits
develop such systems in-house. As Kovach et al. the company’s vision. One particular disadvantage
(2002) describe, a small company might prefer to of COTS is of an intangible nature; when most
depend on consultants and outsource all aspects companies adjust their practices to conform to
of design, implementation, and operation, while the COTS software standard offerings they lose
a large company could prefer to opt for in-house their competitive advantage, as their technology-
development of e-HRM applications by their based processes become replicas of each other
own IT department. Widely believed benefits of (Carr 2003, p. 6). For e-HRM, this means that
commercial off-the-shelf (COTS) systems in gen- companies cannot strategically position their HR
eral are time-saving, proven system quality, and activities or improve its strategic orientation by
availability of expertise through the provider. Fur- the use of electronic means.
thermore, COTS systems help reduce the amount Although it is believed that off-the-shelf soft-
of time and resources used during the design and ware usually contains industry’s best practices,
implementation phases of the software project, as but Yeow and Sia (2007) posit that the notion that
the software is already professionally designed “best practices” are context-free and can be con-
and implemented to fit most of customer’s needs veniently acquired, stored, used, and transferred
(Amons & Howard 2004, p. 50). Silverman (2006) across organizations is questionable. For example,
offers a few reasons why most organizations opt implementing off-the-shelf ERP systems such
for COTS solutions. First, avoiding lengthy and as Oracle, PeopleSoft, and SAP, in the context
complex software design and development cycles of non-US/European and non-private organiza-
(by using COTS solutions) saves time, money, tions (i.e. governmental institutes in Kuwait) can
760
be quite challenging and might not even fit the To take further the point issued by Kovach et
environment’s needs. Any discrepancies between al (2002), in case of e-HRM the acceptance and
the organization’s needs and the off-the-shelf usage of applications by employees and managers
software’s features will most likely affect the is critical to successful implementation and the
overall project success (Lucas et al., 1988; cited generation of results. That makes the decision
by Yeow and Sia 2007, p. 2) whether to ‘buy or build’ even more important,
In order to avoid these down-sides, companies and seems to favor in-house development over
might opt for in-house development of an HRIS off-the-shelf implementation. However funded
system. Building e-HRM applications in house knowledge on this issue is lacking so far. The
requires a special team of developers that is fully annual CedarCrestone survey (2006) shows that
skilled and dedicated to do the job of designing many organizations lay out more expenses for
and implementing the software. One of the rea- outsourced solutions than they do for in-house
sons companies choose to develop their HRIS development. Three main reasons are explained:
in-house is that they can design the software organizations with in-house software development
precisely to fit the organizational goals, as they mostly have fragmented system operations which
know their own processes better than anyone else cause them to overlook embedded costs within
(Zizakovic 2004, p. 4). This way companies gain distributed operations; outsourcing assures better
the competitive advantage and have full control view and control of costs; and outsourcing is a
of their proprietary software. solution for organizations that struggle to control
Pearce (2005) provides three key elements costs due to the complexity of their operations
for software implementation and development. (HR Technology Trends to Watch in 2007 2007, p.
First, organizations must consider the function 1). On the other hand, the CedarCrestone survey
of the IT application (i.e. e-HRM applica- (2006) notes that some respondents sense that
tion) to determine the level of standardization in-house HR solutions positively affect employ-
required by the operations of the application ees’ productivity and can be easily and swiftly
and whether COTS software can provide these integrated into the current system, while off-the-
standard operations (Pearce 2005, p. 93). For shelf software would cause difficulties during the
example, e-HRM activities such as payroll and integration phase.
benefits can be suitably offered by COTS sys-
tems. Second, organizations must consider the theoretical Framework
life cycle of the application and how it will be
modified, updated, maintained, and supported The suggested framework for this research is
over the years to come (Pearce 2005, p. 94). depicted in Figure 2. This framework combines
This consideration helps organizations to set the main elements discussed in the literature
forth a long-term strategic plan for its e-HRM review. Since studies on the different e-HRM
practices in order to continue sharpening its applications are scarce, this framework is used
competitive edge. The third key element in primarily to study different elements of imple-
deciding on the best software implementation menting software applications and to focus pre-
strategy is to consider the return on investment, dominantly on implementing e-HRM software
or ROI. Pearce (2005) insists that companies applications and the effect they might bring onto
must practice the strategy of weighing the as- the HR function.
sociated costs of implementing and developing This framework suggests three dependant
different software methods against the potential stages of e-HRM systems’ implementation and
outcome of the IT project itself. deployment. First stage is the e-HRM software
761
Figure 2. Theoretical framework
implementation and development approaches, means of less value than the other stages. It repre-
which is affected by the type of system used by sents the outputs and benefits an HR department
the organization—whether it is a commercial off- aims for during the e-HRM transformation. These
the-shelf e-HRM system or one that is developed outputs consist of the “four C’s of HR”—namely
in house. Another factor affecting this stage is commitment, competence, cost effectiveness, and
the predetermined e-HRM goals as suggested congruence—and user benefits suggested by the
in the literature review. E-HRM goals constitute Technology Acceptance Model—such as useful-
the implementation and development approaches ness and ease of use.
and set a course of action to be followed during
all three stages. The second stage consists of the
main elements of e-HRM application deployment MEtHODOLOGy
such as e-HRM activities handled by the system,
application types, and application characteristics. Deciding which research strategy to adopt depends
These application criteria help an HR department solely on the nature of the research question and how
to provide the necessary e-HR functions to accom- the author plans to answer it. The strategy consists
plish the goals mentioned in the first stage and meet of a plan with clear objectives derived from the
the needs of the stakeholders. Note that each of the research question and specifies the sources from
e-HRM activity types fit under at least one of the which to collect data (Saunders et al. 2003, p. 90).
e-HRM application types. Furthermore, e-HRM Since the aim of this research is to compare the two
application and activity types, and the application types of e-HRM system implementation methods,
characteristics are interdependent (represented the author assumes a comparative, descriptive case
by the up-down arrow). For example, relational study strategy. A Comparative study scrutinizes the
e-HRM activities such as training, recruitment, differences and similarities between the two e-HRM
and performance management depend mainly on implementation methods, whilst a descriptive study
the level of reliability and usability of the e-HRM attempts to construct a precise profile of the different
system, while functionality and efficiency of the e-HRM system implementation methods (Robson
e-HRM system might change with the type of e- 2002; cited in Saunders et al. 2003, p. 97). Further-
HRM activity types. The last stage is not in any more, the descriptive study is a perfect strategy
762
that helps with the proliferation of the inductive by Oracle, called “Oracle ERP”. The rea-
research approach. This strategy helps to reach a son this company was chosen because of
theoretical conclusion about the outcome and level the amount of experience it had with the
of effectiveness induced by both implementation e-HRM system and the strong reputation it
methods. A case study research strategy implies possessed in the Kuwaiti marketplace.
the use of detailed and empirical investigation • Case Study 2: The second case study was
within the real context of a particular phenomenon conducted on a governmental institute that
(Cassell & Symon 2004, p. 323). While the case had gained its competitive advantage main-
study strategy provides a good challenge and a ly through leveraging its IT capabilities to
source of new hypotheses of an existing theory, provide the latest technological trends to
it can also help with the creation of a new theory its customers. The reason this organization
using different means of data collection methods was chosen is that it was one of the few
such as questionnaires, interviews, observation, local companies that had developed their
and documentary analysis (Saunders et al. 2003, p. e-HRM system in house.
93). This, indeed, makes a perfect strategy that fits
the research philosophy and approaches suggested The main constructs used in this research are
in the previous sections, as Cassell and Symon shown in the table below. These variables can then
(2004) agree that the case study strategy begins be used to create a set of questions that help the
with a primitive framework which then will be researcher with gathering appropriate research
reproduced or developed under a new theoretical data for analysis. This process is called the “op-
framework that makes sense of the collected data erationalization of construct” and is explained in
and can be systematically examined for plausibil- the next section.
ity. This strategy also facilitates the collection of
data through multiple research instruments such Operationalization of constructs
as participant observation, direct observation, in-
terviews, focus groups, and documentary analysis Boiling down the constructs into research ques-
(Cassell & Symon 2004, p. 325). Consequently, this tions helps the researcher with the design of
means that a variety of research instruments can the appropriate research instruments to be used
be utilized to understand how the different e-HRM to collect and test enough data about the com-
implementation methods affect the HR function. mercial off-the-shelf and in-house developed
All in all, case study strategy can be considered e-HRM systems. By doing so, the constructs are
a prudent choice for this research as it offers a “operationalized” so that they can be measured
flexible environment for the researcher to work accordingly. Table 3 shows the operationalization
with and attempts to collect as much information of constructs mentioned in the previous section.
as necessary to conclude how the two E-HRM
systems can be compared and contrasted.For the research Instruments
purpose of this research, two descriptive case
studies were conducted on local organizations Case study is a strategy that includes multiple
that have been chosen according to the e-HRM methods (or instruments) suited to research
system they used: questions that require detailed understanding of
a particular phenomenon due to the large amount
• Case Study 1: This case study was con- of data collected in the context (Cassell & Symon
ducted on a well known company that uses 2004, p. 323). For this research, the following data
a commercial off-the-shelf e-HRM system collection instruments were used:
763
Table 3. Operationalization of constructs
Constructs/ Definition/Dimension Operationalization of Constructs

Variables
Overview
Background General information about the company - Business establishment
- Number of employees
- Type of Business
Current HR practices Where the HR practices currently stand - Main function/mission of HR department
- Level of employee/manager involvement
E-HRM The e-HRM application currently in use - e-HRM system brand/name
Application - Reason to chose this system
- criteria used to make the decision
E-HRM HR activities utilized/offered by the new - HR activities handled by the system
Activities e-HRM system - New HR activities introduced by the system
- level of employee/manager involvement in e-HRM activities
Implementation & Development Approaches
User Involving employees and managers in the - Type of users involved
Involvement e-HRM implementation and development - Stage(s) at which users were involved
process - Level of involvement
Business Reengineering/redesigning of HR business - HR processes reengineered
Process processes - HR processes eliminated by BPR
Reengineering - New HR processes introduced
Planning & Vision The planning process and strategy used to - Type and quality of business plan
implement the e-HRM system - Plan outline
- Project strategy alignment with corporate strategy and vision
Training & Education Plan taken to train and educate users of the - Level of training/education
new e-HRM system - Training material
- Training delivery
Change How the change/conflict implied by the - Resistance, conflict, or confusion by staff/users
Management new e-HRM system is managed - Handling of conflict
- Corporate strategy toward change
Top The support of the e-HRM system imple- - Level of top management involvement
Management Support mentation by top managers
Effective Inward/outward communication needed - Level and type of communication between different parties
Communication during the implementation phase involved
- communicating project scope,
objective, activities, updates
Project Managing and administering the e-HRM - Implementation management
Management implementation project - Responsible PM team
- Project milestones
e-HRM Goals
Increase Make HRM more efficient by reducing - Achieving cost reduction
Efficiency time, costs, human resources, etc - Cost, time, staff, other resources reduction
- Processes eliminated to achieve efficiency
Provide Provide a mechanism for employees and - Employee Self Services (ESS)
Self Service managers to perform HR functions - Manager Self Services (MSS)
Achieve Service Excel- Provide HR high quality services to em- - Customer orientation of HR department
lence ployees and managers - How customers perceive the quality of services
Improve The strategic orientation and focus of the - The effect of the e-HRM implementation on HR strategy
Strategic Focus organization should be improved through - Changes made to improve HR strategy
the use of the e-HRM system
764
Table 3. continued
Constructs/ Definition/Dimension Operationalization of Constructs
Variables
e-HRM Activities
Operational Activities Routine HR tasks such as payroll, atten- - Transformed operational activities
dance, vacation, etc - New operational activities introduced by e-HRM
- Effect of e-HR operational activities on HR function
Relational Activities Internal and external relational activities - Transformed relational activities
such as recruitment and selection, training, - New relational activities introduced by e-HRM
performance management, etc. - Effect of e-HR relational activities on HR function
Transformational HR activities aimed at improving the - Transformed transformational activities
Activities strategic orientation of the HR department - New transformational activities introduced by e-HRM
such as knowledge management, compe- - Effect of e-HR transformational activities on HR function
tence management, strategic redirection,
downsizing, upsizing, rightsizing, etc
e-HRM Application Characteristics
Functionality How the e-HRM system functions with - Fulfilment of stated needs
regards to suitability, accuracy, interoper- - Appropriate set of HR functions
ability, and security - Expected results
- Information and data security
Reliability How reliable the e-HRM system in deliver- - System maturity
ing HR needs - Fault tolerance
- Recoverability
Usability How easy is it to understand, learn, and - Understandability
operate the e-HRM system - Learnability
- Operability
- Attractiveness
Efficiency Performance of the e-HRM system with re- - Time utilization
gards to time and other resources utilization - Resources utilization
Maintainability The stability of the e-HRM system when - Analyzability
analyzed, changed, and tested - Changeability
- Stability
- Testability
Portability How the e-HRM system can adapt new - Adaptability
requirements, be installed in different - Installability
environment, coexist with other software - Co-existence
programs, and can be replaced - Replaceability
e-HRM Outcomes
Commitment Employees commitment to organization - Change in level of employee commitment to organization
due to the implementation of the new e-
HRM system
Competence Development of employees competencies - Influence of e-HRM on employee competencies
due to the implementation of e-HRM
Cost Benefits of implementing e-HRM such as - Cost of delivering e-HRM practices and services such as staff
Effectiveness providing adequate e-HRM services should turnover and resistance
outweigh costs such as employee resistance - Benefits versus costs
and turnover
Congruence The effect of e-HRM on the level of con- - Relationship between managers and employees
gruence between employees’ own goals and - Relationship between employees
those of the organization
HRIS Benefits
Usefulness How useful the e-HRM system is in en- - Enhancement of employee performance
hancing employee performance
Ease of Use Easiness of performing e-HRM activities - Effort needed to perform e-HRM tasks
765
• Semi-structured interviews: These types sufficient amount of different feelings and

of non-standardized interviews use list of experiences about the e-HRM system at
questions that vary between interviews the same time.
according to the context (Saunders et al.
2003, p. 246). These interview questions sampling Methods
are the outcome of the operationalization
of constructs process mentioned in the pre- The criteria for selecting case study samples were
vious section. Table 4 shows the position mainly based on the type of e-HRM system used
of the interviewees and the related ques- by an organization—whether it is off-the-shelf or
tions used during the interviews. in-house developed e-HRM system. The author
• Documentary analysis: Company docu- also wanted to conduct intensive, comparative, and
mentation of the e-HRM system imple- descriptive investigations of both e-HRM systems
mentation is analyzed to understand how and, therefore, selected only two case studies to
the system works. This data, if available, concentrate on. Secondly, the requirement to col-
provides a secondary source of informa- lect large amount of data to analyze the systems put
tion that could potentially assist the re- a constraint on the author to select samples with
searcher to reach a final conclusion about vast experience in the context of electronic HR.
the research question. This, indeed, narrowed down the search for perfect
• Participant’s observation: This method samples to those organizations with a minimum
allows the researcher to observe first-hand of 2 years of e-HRM experience. Thirdly, the two
the day-to-day e-HRM experiences, activi- sample organizations operate in different markets
ties, feelings, and interpretations of the us- and have different cultures, as to generate more
ers in order to collect information related interesting and variable results. For example, the
to some of the constructs mentioned in the first sample organization was a mid-size, privately-
previous section (Cassell & Symon 2004, owned company with a market culture that is profit
p. 154). oriented, while the second sample organization
• Group interviews: The purpose of group was a government-owned entity with a culture that
interviews is to collect data from feed- is more oriented toward bureaucracy. Fourthly, the
backs and interpretations of multiple inter- size of the sample organizations was not a major
viewees (with similar experiences) during criterion; however, it gave the organization more
one gathering. An advantage of the group of an initiative to implement an e-HRM system.
interviews is to hear different opinions of For example, middle to large size organizations
the phenomenon at the same time (Cassell feel more pressure to implement an HR system
& Symon 2004, p. 143). This gives the that would keep human resources management
researcher an opportunity to observe a under control. Lastly, technologically advanced
Table 4. Interviewees & related questions
Interviewee position Related questions (Appendices)

HR manager / administrator A1, A2, B2-B5, C1-C5, D1-D4, G1-G2, G4-G5
End users (employees & managers) B1, B7, C2-C3, F3-F4, G1-G2, G4, H1-H2
Project Manager of e-HRM project A3-A4, B2-B9, C1, C4-C5, D1-D4, G1, G3-G5
IT personnel A3-A4, B1-B2, B8, C2-C3, D1-D4, E1-E5, F1-F5
766
organizations were preferred by the author as they structuring and business process reengineering
are more likely to adopt e-HRM technologies at took place. The aim of this BPR was to transform
early stages of their life cycle. all business processes into the IT era, including
the transformation of the HR practices into e-HR
Data Analysis Methods technologies. The main steps of reengineering
were diagnostics, analysis, and implementation.
Intensive interviews were first conducted at both HR reengineering (i.e. transformation to e-HR
sites and recorded on a digital recorder, and then using Oracle ERP) was 30% of the whole BPR
transcribed into separate sets of data for each process of the organization. It took about 3 years
case. The data was then categorized into mul- to progress through the diagnostics, analysis, and
tiple sections according to the operationalization implementation stages, and about one year to
Table 3. complete the transformation of the HR function.
In order to implement such a big change within a
large organization, the bureaucratic and centralized
FINDINGs culture needed to be changed as well. This culture
was changed to reflect the open environment that
The data analysis starts with transcribing the re- accepted change and forward movement.
corded interviews to capture as much necessary As for why this organization opted for Oracle
data as possible. The transcripts are then studied ERP, it was recommended by the Chairman to
and paragraphs rephrased and rearranged under choose this system as the best offer because of
the corresponding instruments depicted in the its flexibility, changeability, adaptability, and
framework. reputation between other vendors such as SAP
The following sections contain those analyzed and Anderson. The second reason was that the
transcripts from both case studies, where case 1 IT department at this company didn’t have the
represents the company that uses off-the-shelf qualified resources to implement such a fully in-
Oracle ERP, and case 2 represents the governmen- tegrated enterprise system that could support the
tal institute that has developed its own e-HRM new structure and culture. As a matter of fact, the
system in house. Data collected from each case is IT department was mostly outsourced to a third
displayed side by side according to the theoretical party that was able to arrange and manage the
instruments to show a clear comparison between e-commerce transformation. Oracle ERP needed
the two different systems. lots of customizations to remove unwanted pro-
cesses such as taxation and add or modify the HR
e-HrM Overview processes to fit the company’s HR needs.
About 80% of manual work was eliminated
The aim of this section is to explain some of the by the new system. The BPR, along with the ERP
general issues associated with the implementation implementation, has caused the reduction of man-
of the e-HRM systems of both cases. The overview power from 900 to approximately 200, as most
allows the researcher to get a feel of the researched processes were eliminated, reduced, or outsourced.
systems before getting into the detailed interviews The other 20% of manual and paperwork cannot
that would confirm or refute the propositions sug- be changed due to the nature of the work that
gested in the theoretical framework. requires to be done in professional and manual
Case 1: The implementation of Oracle ERP way such as memorandums and official letter.
system started in 2004. Before that, in 2001, an Case 2: The e-HRM system was developed
initiative from the Board to perform a total re- by the Systems Development team in this or-
767
ganization’s IT department. The reason behind Presentation of results

developing this e-HRM system in house was that
there was no system out in the market that could While the data collected might specifically de-
fulfill the complex HR requirements implied by scribe the contextual situation of e-HRM systems
the country’s laws and by this governmental insti- used by these two cases, some general remarks
tute. The IT department manager indicated that if about the differences and similarities between off-
they had chosen an off-the-shelf e-HRM system, the-shelf and in-house developed E-HRM systems
customization would have cost them incredible can be drawn. Referring back to the main question
amount of time and money. Second, this system of this research—which compares off-the-shelf
was developed and implemented in order to fit it with in-house developed e-HRM with regards to
to the user’s needs and not to adjust the user to the implementation and development approaches,
the system’s offerings. e-HRM goals, e-HRM activities they facilitate,
The system development started in 2002 and application types and characteristics, and e-HRM
took about 2 years to complete. It is totally web outcomes—one can provide conclusive evidence
based using Oracle database, Microsoft .NET about the differences and similarities at hand.
framework, and object oriented programming In general, organizations transform their HR
methods. Long before that, in 1976, there was an function to e-HR in the hope to reduce the amount
initiative by top management to automate most of time spent on processing HR activities. The au-
of the HR processes in order to minimize time tomation of HR processes, therefore, is a necessary
and associated costs. Paperwork was costing the requirement and a major step in the transforma-
organization lots of time and money to store the tion process, as mentioned by Lengnick-Hall and
voluminous data in multiple cabinets. Also, data Moritz (2003, p. 367). A minimum of 90% of HR’s
integrity and security was a top priority and major paperwork at both cases has been eliminated and/
concern that would justify the need to transform or replaced by automated functions. It is also true
all HR processes into the e-commerce era. At that that companies evaluate the capabilities of their
time, a mainframe system called “Tandem” was IT department and accordingly decide whether
used to build the infrastructure that would sup- they can develop and support enterprise-quality
port the renovated business. The current system software in house. Furthermore, the point of
was therefore developed and implemented in ac- Pearce (2005) about the organizations’ concern of
cordance to the vision of top management which whether off-the-shelf software can provide stan-
encouraged the IT department to integrate the dard e-HRM functions is touched upon by both
latest and most flexible and reliable electronic cases. On one hand, the first case saw that Oracle
technologies. This upgrade required the replace- ERP would provide just about the right e-HRM
ment of the old mainframe system with the more functions with minimal customization needed.
flexible and compact Microsoft system. On the other hand, the second case believed that
The e-HRM system provides user with many fa- no single COTS system could provide standard
cilities such as attendance, which is totally secured e-HRM functions that best fit their business needs.
and tied up to a biometric system that scans users Also, to confirm the point of Amons and Howard
fingerprints for authentication. Also, the main HR (2004), off-the-shelf e-HRM systems dramatically
functions provided by COTS systems have been reduce the amount of time spent on design and
extracted and implemented within this in-house implementation stages, as the system has already
developed system, with a touch of flexibility to been professionally designed and only needs to
fit those functions to the organizational needs. be integrated and deployed. However, the integra-
768
tion and deployment stages are overwhelmingly Factors that affect off-the-shelf e-HRM imple-
critical due to the fact that off-the-shelf systems mentation the most are business process reengi-
need to be customized to properly bolt onto the neering, planning, and project management. The
company’s environment. It took case 1 a whole reason being is that HR process reengineering and
year to completely integrate and deploy Oracle project management are two essential elements
ERP. On the other hand, as Amons and Howard handled professionally by the vendor within a
(2004) suggest, design and development stages well organized plan to implement off-the-shelf
are lengthy using in-house developed software, e-HRM software. In other words, vendors do not
especially with the second case in this research just deliver the software, but they rather manage
where the development team’s experience was the whole project through planning, reengineering,
limited. training, and implementation. From a realist point
Looking at the development and implementa- of view, one can also conclude that implement-
tion approaches, a general theme can be depicted. ing off-the-shelf e-HRM system is a professional
When implementing either e-HRM system (i.e. experience that must be handled by those who
COTS or in-house developed), the organizations are most skilled and knowledgeable about the
look for satisfying the HR’s department needs product itself, where this knowledge needs to be
as the main user, and the needs of all other us- transferred to the acquiring party in an organized
ers such as employees and managers. However, and well planned manner.
some approaches might be emphasized more A new predominant factor has been extracted
than others according to the type of e-HRM during this research, which affected the implemen-
system. For example, when implementing in- tation of e-HRM systems at both cases. Teamwork
house developed e-HRM systems, the users are plays an essential role in implementing e-HRM
continuously involved, trained, and educated systems as it triggers a synergistic effect between
throughout the product life cycle to ensure that the teams involved in the transformation project.
the software works for the best interest of its Cross-functional teams work cohesively in both cases
users, and to build a high level of trust in the to achieve planned targets and meet project gates.
IT department. This new experience for the IT Flexibility, timeliness, and elimination of
department meant a new challenge and a new manual processes are advocates of efficiency when
chance to prove its competence. Also, change implementing both e-HRM systems. Increasing
management and communication management efficiency, therefore, is the most critical goal of
had to be perfected to provide the proper support transforming the HR function to e-HR. Efficiency
for the development team and to make sure no can also be measured by the amount of HR staff
conflict or confusion would cause the deteriora- reduced after the implementation. In both cases,
tion of trust between key players. As for project less staff members could handle more tasks us-
management, the second case believes that or- ing the e-HRM systems. Providing employee and
ganizations who develop their e-HRM software manager self services also came on the top of
in house must appoint a project manager from most effective e-HRM goals list. ESS and MSS
within the development team for two reasons. are implemented within both e-HRM systems to
First, it is best to manage this first experience in provide easy access for employees to perform
house to avoid conflict of interest and to impose multiple e-HRM services swiftly and without
more control on decision making. Second, no interaction with the HR office. Third, achieving
one knows how to orchestrate the development service excellence is considered crucial for orga-
team and leverage their capabilities better than nizations that develop e-HRM software in house.
a person who works closely with them. It is concluded through this research that when
769
an e-HRM system is developed in house, users’ Comparing the e-HRM outcomes of the two
expectations of results are at peak, due to the fact cases, an apparent theme emerges. Because of the
that this software is developed under the same amount of trust, confidence, and support shown to
organizational culture, by staff that commemorate the in-house development team during the design,
this same culture. Last and perhaps least consid- development, and implementation stages, employ-
ered by both cases is the goal of improving the ees’commitment may substantially increase. As the
strategic focus. It is evident that due to the lack of HR senior staff at case 2 phrases it: “supporting
market competition in Kuwait, companies spend the hometown team eventually pays off.” Another
less effort to improve their strategic orientation. reason for such a phenomenon is due to the amount
Operational, relational, and transformational of customer satisfaction implemented into the
e-HRM activities are common between the two system, as most of the users’ needs are catered for
systems. However, these types of e-HRM activi- to create a system that best fits those needs. Cost
ties become the most competitive advantage of effectiveness, on the other hand, is most favored by
in-house developed e-HRM systems. It is con- those organizations that use off-the-shelf system.
cluded through this research that when developing Traylor (2006) agrees that although the amount of
an e-HRM system that best fits the users’ needs, money spent on acquiring, integrating, and support-
many of those needs become reality in an attempt ing the enterprise off-the-shelf system is incredibly
to achieve maximum customer satisfaction and high, more cost effectiveness is achieved for both
meet those high expectations. The outcome of implementation and ongoing maintenance. Case 1
this phenomenon is a system that harbors the best felt that using the renowned Oracle ERP system and
practices of off-the-shelf e-HRM systems with the support package offered by the vendor avoided
a competitive twist tailored to the last details of software obsolescence in the long run, as profes-
users’ needs. sional software providers keep on implementing
As for the e-HRM application characteristics, the latest technologies into their provided services.
supporters of off-the-shelf systems (i.e. case 1) Another outcome observed by the first case was
believe that corporations such as Oracle and SAP changing the attitude of employees, which remains
implement the markets best practices into their questionable. The change in attitude might have
software solutions to gain a competitive advan- been partially affected by the open environment
tage and increase market share. This leads to the created during the e-HRM transformation, but it
creation of state-of-art enterprise solutions such is believed that the organizational-level business
as Oracle ERP. Also, case 1 agrees with Silverman process reengineering had the bigger effect on
(2006) that risk of implementing off-the-shelf e- people’s attitude.
HRM systems is minimized by the fact that those Finally, usefulness and ease of use become
systems are professionally provided and supported natural effects of implementing both e-HRM sys-
by experienced entities. On the other hand, soft- tems. However, both benefits are most noticed by
ware quality characteristics as depicted by ISO the users of in-house developed e-HRM systems
9126 might not have been seriously considered because those systems are made to specifically
by the amateur development team of the second fit the user’s needs.
case. This might have been caused by the fact that The Table 5 provides a summary of those
software development is not in conjunction with findings and a clear comparison between the two
this organization’s profession. However, features systems with regards to the variables in the sug-
like accuracy, suitability, security, reliability, gested framework.
usability, and maintainability become adhesive
during the design and development phases.
770
Table 5. Off-the-shelf vs. in-house developed e-HRM systems
Variables Off-the-shelf In-house

Implementation & Development Approaches
User Involvement - Moderate - High
- Users continuously involved throughout all stages
Business Process - Considered a major factor since HR processes - Some processes reengineered to eliminate time-
Reengineering need to be reengineered to conform to off-the-shelf consuming manual work
standards
Planning & - More planned - Informal plan and strategy
Vision - Strategy and vision are professionally outlaid
Training & - Standardized - More emphasized to gain proper organizational
Education - Delivered by vendor support
Change Management - Handled professionally with the help of the vendor - Handled by project manager to increase effective-
- More rejection of the foreign system ness
- Less rejection and more support of the friendly
system
Top Management - Moderate support - High level of support for the home-made product
Support - High trust in professional services
Effective - Formal communication through awareness sessions - Different informal communication channels for
Communication maximum effectiveness
Project Management - Outsourced to experienced entity for more effec- - In-house for more effectiveness
tiveness
Teamwork - Highly emphasized - Highly emphasized
e-HRM Goals
Increase Efficiency - Time and cost reduction are guaranteed by vendor - Efficiency accomplished by utilizing resources,
and their professional e-HRM minimizing time and errors, and eliminating manual
work
Provide Self Service - ESS and MSS as provided by vendor - ESS and MSS shaped to better fit users’ needs
Achieve Service - Customer-oriented culture imposed by the system - Crucial due to the high level of users’ expectations
Excellence and not necessarily by organization
Improve Strategic - Strategic solutions offered professionally by the - Not critical
Focus system
e-HRM Activities
Operational Activities - Standardized - Tailored to the last details of users’ needs
Relational Activities - Standardized - Tailored to the last details of users’ needs
Transformational - Standardized - Tailored to the last details of users’ needs
Activities
e-HRM Application Characteristics
Functionality - Very secure, accurate, and interoperable - High suitability
- Moderately suitable with a number of customiza- - Very secure and accurate
tions - Limited interoperability
Reliability - High maturity, fault tolerance, and recoverability - Immature but recoverable
- Limited errors
Usability - Easy to understand, learn, and operate - High usability levels due to the implementation of
- Normal attractiveness due to the standard graphical most users’ needs
interface - Very attractive GUI
Efficiency - High levels of timeliness and resource utilization - High efficiency
Maintainability - Professionally maintained by vendor through ser- - Easy to analyze, install, and test
vice and support agreements - Limited changeability
771
Table 5. conitnued
Variables Off-the-shelf In-house

Portability - Easily adopted in multiple environments - Limited to Windows-based environments
e-HRM Outcomes
Commitment - Mildly imposed by the software’s culture - Substantially high due to the amount of users’ needs
implemented in the system
Competence - Moderate increase - Moderate increase
Cost Effectiveness - Substantially high due to the professionalism of the - Normal
system
Congruence - High - High
HRIS Benefits
Usefulness - Normal - Meets or exceeds expectations
Ease of Use - Normal - Meets or exceeds expectations
FUtUrE rEsEArcH cONcLUsION
Further research can be done to determine the best The comparison between off-the-shelf and in-
practices for implementing e-HRM systems in the house developed e-HRM applications has been
Kuwaiti context. Yeow and Sia (2007) found out, exploited in this research to find out the dif-
from a social constructivist point of view, that “best ferences and similarities between development
practices” are socially enacted knowledge that and implementation approaches, e-HRM goals,
requires organizational power through politics, e-HRM activities, application types and charac-
and discourse to influence different technological teristics, and the outcome of implementing both
assumptions, expectations, and knowledge until systems. Realism philosophy was used to extract
this difference is resolved into what is logically facts about the suggested theoretical framework
called “best practices”. This means that politics introduced in chapter three. Further, two cases
and discourse in the context of Kuwaiti organiza- were used as the best candidates for this research
tions can be studied to determine what the best because of the different corporate cultures and
practices are for implementing e-HRM systems. markets they operated in, which lead to more
Another augmented approach to future research interesting remarks.
could be done after the fact that e-HR transforma- Looking back at the theoretical framework
tion in Kuwait has reached its maturity and the developed in chapter three, one can suggest a
market is saturated with diverse organizations few minor adjustments. One of the elements that
that use different e-HRM systems. Infusing e- became irrelevant to this comparative study was
HRM technologies in Kuwait’s organizations is the e-HRM application types. E-HRM systems
inevitable, however it could be long before any are divided into four main types as described in
such technologies are enforced by cultural values the literature, namely administrative, employee/
and/or organizational needs due to the fact that manager productivity, strategic HCM, and busi-
legislations for electronic technologies have yet ness intelligence applications. Each of these sets
to be considered by Kuwaiti laws. handles a number of e-HRM activities (i.e. opera-
tional, relational, and transformational activities).
However, these application types do not provide
relevant information and, therefore, cannot be
772
used as a credible variable of the theoretical as- the most due to the fact that these systems are
sumptions that are made about the differences and handled professionally by the vendors.
similarities between off-the-shelf and in-house Increasing HR efficiency is a critical objective
developed e-HRM systems. Another adjustment behind implementing both e-HRM systems. This
would be to add “teamwork” as one of the major goal can be mainly achieved through increasing
factors affecting the implementation and develop- the flexibility and timeliness of HR processes and
ment of the e-HRM systems, as it was observed eliminating a big portion of the manual processes.
greatly by both cases. Reducing HR manpower is a consequent effect of
Generally, the decision of choosing between the e-HR transformation that can also be counted
off-the-shelf and in-house developed e-HRM is as an efficiency measure.
dependent on factors such as the IT department The fact that an e-HRM system is developed
capabilities, amount of time and resources spent and supported internally by the IT department
on development and implementation phases, and increases the level of pride and support by users
the purpose of e-HR transformation. We saw that toward the system and the team who develops it.
when companies plan a total business process re- This high level of support instantiates high con-
engineering, for example, the best practice would gruence between employees and line managers,
be to seek the assistance of professional services and improves the alignment between departmental
that can provide enterprise-level qualities. and organizational goals. Furthermore, employees’
One of the most interesting findings in this re- commitment increases as the level of confidence
search is that when e-HRM systems are developed and support of the “home-made” e-HRM system
in house, users’ expectations are high. Also, trust increases. On the contrary, companies who imple-
and confidence levels in the development team ment off-the-shelf e-HRM systems increase cost
must be increased in order for the development effectiveness during the life cycle of the product
to meet those high expectations. On the other due to the fact that those systems are “built to last”.
hand, organizations who implement a certain off-
the-shelf e-HRM system automatically develop a
high level of trust in the professional expertise of rEFErENcEs
the provider. This is due to the fact that software
providers implement industry’s best practices Al-Sehali. Saud, H. (2000). The Factors that Af-
and software quality (i.e. ISO 9126) into their fect the Implementation of Enterprise Resource
solutions to gain a competitive advantage and Planning (ERP) in the International Arab Gulf
increase market share. States and United States Companies with Special
Teamwork and cross-functional team coop- Emphasis on SAP Software. D.I.T. dissertation,
eration are on the top list of factors affecting University of Northern Iowa, Iowa.
the development and implementation of both Amons, P., & Howard, D. (2004). Buy it, Build it,
systems, as they create a synergistic environment or Have it Built? Catalog Age, 21(3), 50.
that leads to more efficiency and effectiveness.
User involvement, effective communication, and Ball, K. S. (2001). the use of human resource infor-
change management are the most critical success mation systems: a survey. Personnel Review, 30(6),
factors for implementing in-house developed 677–693. doi:10.1108/EUM0000000005979
E-HRM software. On the other hand, planning Beer, M., Spector, B., Lawrance, P., Mills, Q.,
and vision, project management, and business & Walton, R. (1984). Managing Human Assets.
process reengineering are factors that affect the New York: The Free Press.
implementation of off-the-shelf e-HRM systems
773
Bingi, P., Sharma, M. K., & Godla, J. (1999). Hendrickson, A. R. (2003). Human resource
Critical issues affecting an erp implementation. information systems: backbone technology of
Information Systems Management, 16(3), 7–14. contemporary human resources. Journal of Labor
doi:10.1201/1078/43197.16.3.19990601/31310.2 Research, 24(3), 381–394. doi:10.1007/s12122-
003-1002-5
Buckhout, S., Frey, E., & Nemec, J. Jr. (1999).
Making ERP succeed: turning fear into prom- HR Technology Trends to Watch in 2007 (2007).
ise. IEEE Engineering Management . RE:view, HR Focus, 84(1), 1.
116–123.
International Organization for Standardization.
Carr, N. G. (2003). IT doesn’t matter. Harvard (ISO) (2001). ISO/IEC: 9126 Software Engineer-
Business Review, 81(5), 41–49. ing – Product Quality – Part 1: Quality Model –
2001. Retrieved November 21, 2007 from http://
Cassell, C., & Symon, G. (2004). Essential
www.iso.org.
Guide to Qualitative Methods in Organizational
Research. London: SAGE Publications. Kovach, K. A., Hughe, A. A., Fagan, P., & Magitti,
P. G. (2002). Administrative and strategic advan-
CedarCrestone 2006 HCM Survey: Workforce
tages of HRIS. Employment Relations Today,
Technologies and Service Delivery Approaches
29(2), 43–48. doi:10.1002/ert.10039
– 9th Annual Edition.
Lee, C., & Lee, H. 2001. Factors Affecting En-
CedarCrestone 2007-2008 HR Systems Survey:
terprise Resource Planning Systems Implementa-
HR Technologies, Service Delivery Approaches,
tion in a Higher Education Institution. Issues in
and Metrics – 10th Annual Edition.
Information Systems, 2(1), 207-212. Retrieved
Davis, F. D., Bagozzi, R. P., & Warshaw, P. R. November 24, 2007 from http://www.iacis.org.
(1989). User acceptance of computer technology:
Lengnick-Hall, M. L., & Mortiz, S. (2003). The
a comparison of two theoretical models . Man-
impact of e-HR on the human resource manage-
agement Science, 35(8), 982–1004. doi:10.1287/
ment function. Journal of Labor Research, 24(3),
mnsc.35.8.982
365–379. doi:10.1007/s12122-003-1001-6
Exclusive IOMA Survey: What do Users Like (and
Lepak, D. P., & Snell, S. A. (1998). Virtual HR:
Dislike) About Their HRIS? (2002, December).
strategic human resource management in the
IOMA’s Payroll Manager’s Report, 02(12), 1.
21st century. Human Resource Management
Florkowski, G. W., & Olivas-Luján, M. R. Review, 8(3), 215–234. doi:10.1016/S1053-
(2006). The diffusion of human resource infor- 4822(98)90003-1
mation technology innovations in US and non-
Madill, A., Jordan, & A., Shirley, C. (2000). Objec-
US firms. Personnel Review, 35(6), 684–710.
tivity and reliability in qualitative analysis: realist,
doi:10.1108/00483480610702737
contextualist and radical constructionist episte-
Gueutal, H. G., & Stone, D. L. (2005). The mologies. The British Journal of Psychology,
Brave New World of eHR: Human Resources 91(1), 1–20. doi:10.1348/000712600161646
Management in the Digital Age. San Francisco:
Nah, F. F., Lau, J. L., & Kuang, J. (2001). Critical
Jossey-Bass.
factors for successful implementation of enterprise
systems. Business Process Management Journal,
7(3), 285–296. doi:10.1108/14637150110392782
774
Ngai, E. W. T., & Wat, F. K. T. (2006). Human re- Wright, P. M., & Dyer, L. (2000). People in the
source information systems: a review and empiri- E-business: New Challenges, New Solutions.
cal analysis. Personnel Review, 35(3), 297–314. Working paper 00-11. Center for Advanced Human
doi:10.1108/00483480610656702 Resource Studies, Cornell University.
Pearce, J. (2005). In-house or out-source? three Yeow, A., & Sia, S.K. (2007). Negotiating “Best
key elements for it development. Franchising Practices” in Package Software Implementation.
World, 37(4), 93–95. Information and Organization. Retrieved from
doi:10.1016/ j.infoandorg.2007.07.001.
Remus, U. (2007). Critical success factors
for implementing enterprise portals: a com- Yumiko, I. (2005). Selecting a software package:
parison with erp implementations. Business from procurement system to e-marketplace. The
Process Management Journal, 13(4), 538–552. Business Review, Cambridge, 3(2), 341–347.
doi:10.1108/14637150710763568
Zizakovic, L. (2004). Buy or Build: Corporate
Ruël, H., Bandarouk, T., & Looise, J. K. (2004). Software Dilemma. Retrieved November 11, 2007,
E-HRM: innovation or irritation. an explorative from http://www.insidus.com.
empirical study in five large companies on
Web-based HRM. Management Review, 15(3),
364–380.
KEy tErMs AND DEFINItIONs
Ruta, C. D. (2005). The application of change
management theory to the HR portal implementa- BPR: Business Process Reengineering
tion in subsidiaries of multinational corporations. BSS: Benefit-related Self Service
Human Resource Management, 44(1), 35–53. COTS: Commercial Off-The-Shelf
doi:10.1002/hrm.20039 E-HRM: Electronic Human Resources Man-
agement
Saunders, M., Lewis, P., & Thornhill, A. (2003). ERP: Enterprise Resource Planning
Research Methods for Business Students, 3rd edi- ESS: Employee Self Services
tion. Essex: Pearson Education Ltd. HCM: Human Capital Management
Schramm, J. (2006). HR technology Competen- HRIS: Human Resources Information Systems
cies: New Roles for HR Professionals. HRMaga- HRM: Human Resources Management
zine, 51(4), special section, 1-10. ISO: International Organization for Stan-
dardization
Silverman, R. (2006). Buying better to buy better. IVR: Integrated Voice Response
Contract Management, 46(11), 8–12. MSS: Manager Self Services
Thaler-Carter, R. E. (1998). Do-it-yourself soft- PSS: Pay-related Self Service
ware. HRMagazine, (May): 22. ROI: Return on Investment
SMR: Simple Management Reporting
Traylor, P. (2006). To buy or to build? That is the TBS: Total Benefits Statement
question. InfoWorld, 28(7), 18–23. TMSS: Time Management Self Service
Valverde, M., Ryan, G., & Soler, C. (2006). Dis-
tributing HRM responsibilities: a classification of
organisations. Personnel Review, 35(6), 618–636.
doi:10.1108/00483480610702692
This work was previously published in Handbook of Research on E-Transformation and Human Resources Management Tech-
nologies: Organizational Outcomes and Challenges, edited by Tanya Bondarouk, Huub Ruel, Karine Guiderdoni-Jourdain and
Ewan Oiry, pp. 92-115, copyright 2009 by Information Science Reference (an imprint of IGI Global).
775
776
Chapter 3.14
Enterprise Resource Planning
Under Open Source Software
Ashley Davis
University of Georgia, USA
Open source software is becoming more prevalent Open source software (OSS) is becoming a promi-
in businesses today, and while still a relatively im- nent part of the business infrastructure landscape.
mature offering, open source enterprise resource However, open source application software is still
planning (OS-ERP) systems are becoming more in its infancy. Success of open source enterprise
common. However, whether or not an OS-ERP pack- resource planning (OS-ERP) systems will signify
age is the right software for a given organization is a coming of age of open source applications. There
a little researched question. Building on the current are many factors that will determine if OS-ERP
real options thinking about platform acquisitions, systems are a valuable option for corporations,
this chapter proposes the five most critical factors to and thus whether OS-ERP systems will become as
consider when evaluating an OS-ERP package. To prominent as other open source offerings like Linux
adequately do this, a great deal of detail about the or JBOSS. This chapter will inform the reader of
current offerings in OS-ERP software is presented, the current state of OS-ERP in the global context,
followed by a review of the real options theory and and explain to potential adopters of OS-ERP the
thinking behind using these factors to evaluate OS- important factors to consider in evaluating an OS-
ERP options. The international implications of OS- ERP option.
ERP are presented in the “Future Trends” section. First, a common language for defining OS- ERP
systems will be developed. Second, the current
state of OS-ERP software will be explored. Third,
DOI: 10.4018/978-1-59904-531-3.ch004
Enterprise Resource Planning Under Open Source Software
the business models of OS-ERP vendors will be initiatives of the past, it does not adequately
exposed. Fourth, the advantages and disadvantages cover all that “open source” includes today. This
of customization of OS-ERP software will be is in contrast to proprietary software where the
explained. Fifth, the factors for valuing OS-ERP license generally does not allow for distribution
options using real options theory (Fichman, 2004) of the source code and is not free of cost. Evolu-
are defined. Finally, the global adoption of ERP tion and commercialization of OSS have led to
software is explored. many products being labeled “open source” that
are not free of cost. As well, proprietary software
(software controlled and offered by vendors for
bAcKGrOUND a price) that give access to the code are termed
open source, while there is no licensing to support
The first necessary requisite for understanding OS- the open source model of software development.
ERP systems is to define a common language for Proprietary software that allows access to the
talking about OS-ERP applications. This includes source code still leaves the control of the source
defining exactly what an OS-ERP application encode, what is included in the source code in future
tails and whether the software meets the definition versions, in the hands of the vendor, who may be
of open source software. There is much ambigu- less accepting of contributions of code than an
ity in the popular press about what is and is not open source community.
OSS; this is only confounded when ERP systems However, even under the most stringent of
claim to be open source. To clarify these issues, open source (OS) definitions, there have been
the next section will explain historical context of many great open source successes. For example,
OSS. Secondly, open source licensing issues are MySQL is an open source database server that
explained. Then, the issue of open source ERP has grown phenomenally since its inception in
functionality is addressed. Lastly some examples 1995. MySQL AB is the company that supports
of OS-ERP software are provided. the MySQL product. This product is free and the
source code is available to everyone under the
History of Open source software GNU General Public License (GPL). Licensing
will be discussed in more detail in the next section.
Open source software has a rich history, from an MySQL is currently backed by several venture
initial chaotic beginning out of a hacker culture capitalists and is without debt (MySQL, 2007).
(Raymond, 1999) to its current manifestation as There were over 12 million downloads of MySQL
a foundation for profit-seeking corporations such in 2006, and 2,500 new customers started using
as JBOSS (Watson, Wynn, & Boudreau, 2005), MySQL to power Web sites, critical applications,
Compiere Inc., and Red-Hat Linux. As open packaged software, and telecommunications in-
source has evolved, the definition of open source frastructure. MySQL is just one example of the
software has changed and the open source ecosys- success of OS software in the infrastructure space.
tem has grown. Previously, open source software Other examples in infrastructure offerings include
was defined in terms of two characteristics: (1) JBOSS and Linux.
licenses that give programmers the ability to view, In terms of business applications, there are
change, enhance open source code, and distribute fewer major success stories. SugarCRM, however,
the source code without discrimination (Feller & is one of the most successful open source business
Fitzgerald, 2000; Open Source Initiative, 2005); applications. SugarCRM is a customer relationship
and (2) the software is free of cost. While this management software package (packaged soft-
definition was sufficient, for pure open source ware is that which offers a “set of functionality” in a
777
complete state that does not require programming) 1. Apache License Version 2.0: “Subject to
that is available under a “custom” open source the terms and conditions of this License,
license.1 SugarCRM has over a million downloads each Contributor hereby grants to You a
and about 1,000 customers (SugarCRM, 2007). perpetual, worldwide, non-exclusive, no-
ERP systems—large packaged integrated business charge, royalty-free, irrevocable copyright
applications that include intensive functionality license to reproduce, prepare Derivative
in the following areas: marketing and sales, ac- Works of, publicly display, publicly perform,
counting and finance, production and materials sublicense, and distribute the Work and such
management, and human resources2—are relatively Derivative Works in Source or Object form.”
less prolific than these major success stories in the As with all licenses, it is recommended that
open source space. However, as the rest of this a full review of the license is conducted be-
chapter outlines, there are several open source fore working with any open source software
ERP offerings becoming available. It waits to (Apache, 2007).
be determined which of these applications will 2. Mozilla Public License 1.1: This license
become widely used in businesses. allows for access to the source code for
review and modification. This is a copy-
Open source Licenses left license, meaning that all modifications
involving the original source code are sent
Licensing plays a big role in definitions of OSS. back to the originator of the software. As
As of 2006, the Open Source Initiative (OSI) with all licenses, it is recommended that a
recognizes nearly 60 different open source li- full review of the license is conducted be-
censes (Open Source Initiative, 2006). Given fore working with any open source software
this number of licenses, and that OSI is not the (Mozilla, 2007).
only organization that provides accreditation of 3. GNU GPL:“GNU is a recursive acronym
licenses,3 it is understandable that the definition for ‘GNU’s Not Unix’; it is pronounced
of open source is not stable. Many companies guh-noo, approximately like canoe…the
have turned to a dual-licensing strategy, making GNU General Public License is intended to
their code available under general public license guarantee your freedom to share and change
(GPL) but also offering a commercial license free software—to make sure the software is
(Rist, 2005). Given this large number of licensing free for all its users.” Users of software under
options and business models, it is understandable the GNU GPL can use, modify, improve,
that there is confusion as to what software really and redistribute such software. As with all
is open source software. licenses, it is recommended that a full review
Open source software still must allow access of the license is conducted before working
to the source code, as mentioned above. So, OS- with any open source software (GNU, 2007a,
ERP systems are defined to allow access to the 2007b).
code. However, under what terms this is available,
including for what cost, is a question. For the However, some software companies create
purposes of understanding OS-ERP software, one their own licenses with their own implications
must assume that the cost is not necessarily free. for what “open source” means to their software.
For the OS-ERP packages in this chapter, three In the OS-ERP space, examples include:
general license types are utilized:
1. OpenMfg:“…the license OpenMFG uses
allows companies to view and modify source
778
code, as well as make contributions to the recommended that a full review of the license
source code” (Caton, 2006). However, this is conducted before working with any open
license does not allow for distribution of the software.
source code. So, by definition, this is not
what is generally meant by “open source” Table 1 outlines current OS-ERP offerings,
software. As with all licenses, it is recom- licensing, and Web sites where more can be learned
mended that a full review of the license is about the software.
conducted before working with any software. The popular press has referred to all of the
2. avERP: This license includes no license software listed in Table 1 as OS-ERP, even those
fees, no charge for updates, no fees charged with cost. ERP systems are the backbone of an
for own programming, no obligation to pay organization, and the cost of the initial license is
or purchase anything at anytime, no obliga- only part of the total cost of ownership (TCO) of
tion for use of services, and it is possible an ERP system or any software package. Organi-
to alter all program modules if wished and zations cannot rely on vendors and analysts to
includes source code (HK-Software, 2007). provide accurate TCO estimates. One practitioner
It does not appear that redistribution of the notes, “…no matter how honest they try to be,
software is allowed. As with all licenses, it neither vendor nor analyst can ever fill in all the
is recommended that a full review of the variables of the TCO formula…The unique com-
license is conducted before working with bination of resources, both machine and human,
any software. at work within your organization is something
3. OpenPro: This is software built on open only you can fully understand” (McAllister, 2005).
source technologies that allows access to One estimate of the costs that make up TCO in-
the source code. However, there is a license cludes hardware and software, technical services,
fee involved, and the right to distribute the planning and process management, finance and
software is not explained on the Web site administration, training, user support, peer sup-
(OpenPro, 2007). As with all licenses, it is port, and application development (DiMaio, 2004).
Table 1. Open source ERP licensing
Open Source Offering License Web Sites

Compiere GNU GPL http://www.compiere.org/
OpenMFG Custom http://www.openmfg.com/
OfBiz Apache License Version 2.0 http://ofbiz.apache.org/
Tiny ERP GNU GPL http://sourceforge.net/projects/tinyerp
OpenPro Undisclosed Cost Structure http://www.openpro.com/
WebERP GNU http://www.weberp.org/
ERP5 GPL http://www.erp5.org/
Adempiere GNU GPL http://www.adempiere.com/
avERP Custom http://www.hk-software.net/h-k.de/content/doc_138099-3-5-0.php
Fisterra GNU GPL http://community.igalia.com/twiki/bin/view/Fisterra
OpenBravo Mozilla Public License 1.1 http://www.openbravo.com/
GNUe GNU GPL http://www.gnuenterprise.org/
Value ERP GNU http://www.valueerp.com/
779
However, migration costs, testing, system integra- the functionality of each ERP application is also
tion, and so forth must be included in determining important. The maturity of the proprietary ERP
the total cost of switching to any software pack- market has allowed convergence as to what con-
age. For a breakdown of some of the costs to stitutes ERP functionality. For most, an acceptable
consider, please see the cost comparison pro- general definition of ERP systems is as follows:
vided by Nolan (2005). The comparison of costs ERP systems are complex, modular software in-
between proprietary ERP and OS-ERP systems tegrated across the company with capabilities for
is not readily known, as “the lack of large-scale most functions in the organization. As mentioned
implementations (of OS-ERP) eliminate a direct in the “History of Open Source” section, ERP
comparison with enterprise solutions, such as systems include functionality in the following
mySAP Business Suite or Oracle E-Business suite” areas: marketing and sales, accounting and finance,
(Prentice, 2006). However, even getting an ac- production and materials management, and human
curate TCO is not adequate when deciding resources. Many times supply chain management
whether to use open source software (DiMaio, and customer relationship management are also
2004, 2007). For OS-ERP software, the decision included in ERP packages. It is beyond the scope
is more complex because this decision crosses of this chapter to list and define whether each open
the bounds of the entire organization. Decision source system that claims to be ERP actually con-
makers involved in evaluating ERP systems are tains ERP functionality. For practitioners looking
aware that even for proprietary systems, the cost to acquire OS-ERP software and academicians
estimates vary wildly (Cliff, 2003), and thus when looking to do OS-ERP research, a thorough as-
considering OS-ERP solutions, TCO (as complex sessment of whether a particular system contains
as it is) will be but one evaluative criterion. ERP full ERP functionality is required. This is more of
evaluative criteria are often handled by complex an issue in OS-ERP systems since most of these
request for proposal processes involving outside systems are fairly new and a consensus has not
experts. Beyond this expertise, this chapter pro- been achieved as to what is actually an OS-ERP
poses factors for valuing OS-ERP options (see system. In the OS-ERP market, many of the claims
the “Real Options Value of OS-ERP” section) that of what functionality constitutes an OS-ERP
should be considered in addition to these complex system is blurry:
TCO estimates.
As well, the business model (see the “Open One characteristic of open source is that differ-
Source Vendor Models” section) behind open ent projects define their category’s feature sets
source offerings will explain why cost is not as in different ways. This is especially true of ERP
much of an issue in the definition of OS-ERP packages. Linux-Kontor, for example, defines
systems. However, a pure definition of OS-ERP ERP without accounting, focusing instead on
would not include AvERP, OpenPro, or OpenMFG customer management, order entry, invoicing,
because, although all allow access to modify and and inventory. TUTOS, on the other hand, calls
change the code, these packages either charge a itself ERP but more closely resembles a groupware
license fee or do not allow for re-distribution of suite. Clearly, some research is needed to make
the software (as mentioned above). sure you’re really getting what you expect in this
category. (Rist, 2005)
Functionality of Os-ErP
Another concern with open source software
For those interested in understanding the OS-ERP is that the offerings are changing rapidly, so for
phenomena, a solid foundation of understanding up-to-date and complete information, the Web
780
Table 2. OS-ERP platform and functionality
Software Platform Functionality

Compiere Independent Quote to Cash, Requisitions to Pay, Customer Relations, Partner Relations, Supply
Chain, Performance Analysis, Web Store/Self Service
OpenMFG Linux, Apple Mac OS X, Manufacturing, Materials Management, Supply Chain (Sales Order, Purchase Order,
Windows CRM), Accounting
OfBiz Linux, Berkeley Software Supply Chain Management, E-Commerce, Manufacturing Resource Planning, Cus-
Distribution tomer Relationship Management, Warehouse Management, Accounting
Tiny ERP Independent Finance and Accounting, CRM, Production, Project Management, Purchasing, Sales
Management, Human Resources
OpenPro Linux, Windows Financials, Supply Chain, Retail and Manufacturing, CRM and E-Commerce,
Warehousing, EDI
WebERP Independent Order Entry, Accounts Receivable, Inventory, Purchasing, Accounts Payable, Bank,
General Ledger
ERP5 Linux, Windows (coming Customer Relationship Management, Production Management, Supply Chain Man-
soon) agement, Warehouse Management, Accounting, Human Resources, E-Commerce
Adempiere Independent or Linux on Point of Sale, Supply Chain Management, Customer Relationship Management,
its site Financial Performance Analysis, Integrated Web Store
avERP Linux Sales, Manufacturing, Purchasing, Human Resource Management, Inventory Control,
CAD Management, Master Data Management, Business Analyses
Fisterra GNU/Linux Point of Sale, Other Business Processes Specific to Automotive Glass Repair Busi-
nesses
OpenBravo Linux, Windows Procurement, Warehouse, Project Management, Manufacturing, Sales and Financial
Processes, Customer Relationship Management, Business Intelligence
GNUe Linux, Microsoft Windows Human Resources, Accounting, Customer Relationship Management, Project Man-
agement, Supply Chain, E-Commerce
Value ERP (Proj - Linux, Solaris, Berkeley General Ledger, Payable and Receivable, Invoicing, Purchase and Receiving, Time
ect Dot ERP) Software Distribution, Mi- Sheet Management for HR, Inventory Management and Manufacturing
http://www.valueerp. crosoft Windows
net/catalog/index.php
sites listed in Table 1 are the best sources for Os-ErP Examples
information about the offerings and the platforms
on which the software is built/supports. However, There are many different flavors of OS-ERP pack-
just to provide a flavor of the offerings, Table 2 ages, mostly because each package grew from a
provides a cursory look at the functionality being particular need (as OSS offerings often do). For
offered by these software products (OS-ERP) at example, Fisterra grew from a custom applica-
the time of this work. Again, since functionality is tion built for an automotive glass replacement
hard to pin down, (for example, if a package only and repair company (Fisterra, 2007). Versions
lists “HR” as its offering, does this mean benefits of the Fisterra product specifically intended for
administration is included?), those interested in this industry are now called Fisterra Garage.
this software should contact the company directly This new name allows distinction from Fisterra
for the most accurate and up-to-date information. 2—a generic ERP released in 2004. In this case,
proprietary ERP definitions apply, in that the
package has what most would define as full ERP
functionality. However this is not always the case
(as noted above).
781
The industry origins of the ERP package are Augustin is founder of VA Linux (VA Software)
important, as many times these are industry solu- and launched SourceForge.net, the largest open
tions that work best for the original industry. This source development site on the Internet (Compiere,
is akin to SAP providing excellent manufacturing 2006b). These new additions to the board indicate
functionality because of its origins in the manu- an interest in growth by Compiere.
facturing industry. Another example of OS-ERP However, Compiere faces many challenges,
starting in a specific industry is OfBiz, which for example a group of developers decided to
started with an emphasis on the retail industry “fork”4 and created Adempiere ERP, CRM, &
(Adamson, 2004). SCM Bazaar (commonly referred to as Adem-
piere). Forking (Raymond, 1999) is not unusual
in the open source community. OFBiz also has a
AssEss tHE Os-ErP LANDscAPE forked version called Sequoia. Sequoia in Febru-
ary 2006 was renamed Opentaps, meaning Open
In the OSS arena, infrastructure products like Source Enterprise Application Suite. The issue of
Linux, Perl scripting language, and My SQL forking is not new in open source software and
database management systems have been very thus is an issue in open source ERP applications
successful and thus very prominent. Given that as well. As mentioned earlier, forking usually
the developers of open source software were occurs when a group of developers decides that
many times the users of open source software, the current direction of the project is not as they
this success is predictable. Less predictable is would like; take for example what happened with
the success of open source software when used to Compiere’s fork Adempiere. Adempiere was
develop applications. Most of the time, develop- started in September 2006 after “a long running
ers are developing the application for users that disagreement between Compiere and the commu-
are very different from themselves. The users of nity that formed around the project” (Adempiere,
applications are not technical, and their require- 2006). The developers behind Adempiere felt that
ments are very different from those of technical Compiere was focusing on “commercial/lock-in”
users—thus the criticism that open source software aspects of the project, and decided to create a new
is not “user friendly” or has a usability problem version that could focus more on the community
(Nichols & Twidale, 2003). This issue is impor- and the “sharing/enriching” aspects of the project.
tant to OS-ERP, as possibly the entire organiza- Jorg Janke refutes these allegations, nonetheless
tion will interface with the OS-ERP application, forking has occurred.
rather than only IT people directly working with
infrastructure products. Geographic Origins, Age, and
OS-ERP is a quickly changing environment. Networks of Os-ErP Projects
For example, Compiere, one of the most well-
known open source ERP companies, announced At this time, there is very little research (academic
Andre Boisvert as chairman of the board and chief or practitioner) into the global status of open source
business development officer in May 2006. Jorg ERP. It is clear that there are many vendors out
Janke (founder of Compiere) noted that Boisvert’s there offering open source packages in multiple
success in “applying the open source business languages—indicating an international audience.
model to markets traditionally monopolized by Also, the packages analyzed for this research have
proprietary software vendors is definitely a plus” geographically dispersed locations, if any location
(Compiere, 2006a). Then, in July 2006, Compiere is listed at all. Many times in an open source ERP
added Larry Augustin to its board of directors. Mr. project, the Web serves as the primary location for
782
Table 3. Geographic origins of OS-ERP projects

are often offering consulting services, implemen-
Open Source Project Founded Where tation services, maintenance, or training. The
Company number of customers is also important in assess-
Compiere Inc. 2001 Santa Clara, CA ing the maturity of the OS-ERP. As can be seen
OpenMFG 2001 Norfolk, VA from Table 4, Compiere currently has more the
OfBiz 2001 (2003 migrated to Web three times the number of customers as any other
java.net) OS-ERP application. However, it is hard to assess
Tiny ERP 2002 Belgium the number of customers because many of these
OpenPro 1998 (1999 first release) Fountain Valley, open source projects have no central repository
CA
for keeping track of this information. Although
WebERP 2001 (2003 first release) Web
partners that provide services for OS-ERP systems
ERP5 2002 Web
have some count of how many customers they
Adempiere 2006 Web
currently have, they do not necessarily share this
avERP 1998 (2001 first instal- Bayreauth, Ger-
lation) many
information. For assessing the value of OS-ERP
software, the number of partners, customers, and
Fisterra 2002 (2003 first release) Web
developers provides a strong network and thus
Openbravo 2001 Pamplona, Spain
should provide greater value. More will be dis-
GNUe 2003 (first release) Web
cussed on this topic in the “Real Options Value
Value ERP 2005 Sparta, NJ
of OS-ERP” section.
Another important consideration in assessing
the landscape of OS-ERP offerings is that most
the project, and there is no real geographic orien-
of these packages are targeted to specific audi-
tation associated with the project, as is indicated
ences and may only be proven for specific sizes
by “Web” in Table 3.
of organizations. Table 5 outlines the size of or-
Most of the open source ERP companies are
ganization targeted by OS-ERP packages accord-
relatively new (see Table 3), though many have
ing to their Web sites.
formidable numbers of partners (see Table 4).
From Table 5, it can be gleaned that most OS-
Partners are important in sustaining an open source
ERP projects are targeting small and medium
project. Partners have some monetary interest in
enterprises, with some targeting large organiza-
sustaining a viable open source offering, as they
tions. With this target audience, the question of
Table 4. OS-ERP network (partners, customer, and developers)
Customers Partners Developers Data Collected From

OpenBravo 15 * * Serrano & Sarriegi, 2006
Compiere Inc. 240 70 50 Ferguson, 2006
ERP5 10 8 * Serrano & Sarriegi, 2006; www.erp5.org
avERP 60 * * http://de.wikipedia.org/wiki/AvERP
OpenMfg 35 20 100 Ferguson, 2006; www.openmfg.com
OFBiz 59 21 * www.ofbiz.org
TinyERP * 34 * http://tinyerp.com
OpenPro * Over 75 * www.openpro.com
WebERP * 9 * www.weberp.org
783
multinational implementation success with OS- ERP, CRM, & SCM Bazaar at #11; and Compiere
ERP systems is raised. At this time, the limited at #14. There is clearly activity happening at the
and varying functionality (see Table 2), and lack site that hosts these projects (http://sourceforge.
of large-scale implementations precludes direct net/top/mostactive.php?type=week). This is a
comparison with solutions like mySAP Business sign of strength for these OS-ERP projects. Also,
suite or Oracle E-Business suite (Prentice, 2006). in trying to assess the value of OS-ERP options,
VC funding and project activity would certainly
Venture capital Funding and Project positively impact the prospect for network domi-
Activity in Oss and Os-ErP nance of the OS-ERP. More will be discussed on
this topic in the “Real Options Value of OS-ERP”
The open source market is heavily funded by section of this chapter.
venture capital firms, Guth and Clark (2005)
in the Wall Street Journal estimated that $290
million was invested in open source in 2004. OPEN sOUrcE VENDOr
However, open source ERP companies are lag- bUsINEss MODELs
ging in maturity and thus their numbers for VC
funding are much lower. Currently, two open Open source vendors have many different business
source ERP companies are funded by venture models. When working with a piece of open source
capitalists: OpenBravo (Serrano & Sarriegi, 2006) software, it is important to understand which type
and Compiere (Hoover, 2006). Both companies of organization/business model is in use by any
are funded with $6 million. particular vendor. Table 4 defines the business
Given the importance of the community to models that have been used to describe what is
open source project value, SourceForge.com tracks happening with open source firms in general. The
the most active projects daily. As of November most simple understanding of these models can be
5, 2006, three OS-ERP projects fall in the top 15 drawn from Bonaccorsi, Giannangeli, and Rossi
most active: Openbravo ERP at #7; Adempiere (2006) with Pure OS and Hybrid business models.
The Pure OS model includes firms that only offer
OS products or OS solutions. The Hybrid busi-
Table 5. OS-ERP
ness model includes all the others that play and
Software Size of Organization Targeted profit in the OS space. Hybrid business models
Compiere Small, Medium mix products, types of licenses, and sources of
OpenMFG Small, Medium Manufacturers revenues. As academicians and practitioners, it
OfBiz * is important to understand what type of vendor
Tiny ERP *
is included in any OS project undertaken. As
OpenPro Small, Medium, Large (1-1,000+ users)
can be seen from the plethora of business model
WebERP *
types, the objectives and goals of each type of
business will differ and may impact the quality
ERP5 Small, Medium, Large
of service or the types of products that the vendor
Adempiere Small, Medium
offers. Organizations should consider the business
avERP Small, Medium (1-300 employees)
model of any OS vendor with whom they engage
Fisterra Small, Medium
in business, as the model and the viability of the
OpenBravo Small, Medium
vendor will impact the value of the OS-ERP op-
GNUe Small, Medium, Large
tion. Proprietary vendors do not have this range
Value ERP Small, Medium, Large
of business models, so this is not an issue with
784
Table 6. Open source business models
Business Models Definition Source

FOSS Value-added service-enabling—sell support services and comple- Fitzgerald, 2006
mentary software products
Loss-leader/market creating—goal is enlarging the market for
alternative closed source products and services
OSS 2.0 Includes FOSS, adds the following: Fitzgerald, 2006
Dual product/licensing—can download for free, small percentage
of downloads purchase a commercial license
Cost reduction—proprietary companies offer OS as part of their
solution
Leveraging the open source brand—as government agencies man-
date that open source be a priority option for software solutions, the
open source brand becomes more valuable
Software Producer GPL Model Entirely open source offerings Krishnamurthy, 2005
Software Producer Non-GPL Model Incorporate source code into a larger code base and create a new Krishnamurthy, 2005
product or take an entire open source product and bundle it with
existing products
The Distributor Derives benefits by providing it on CD, providing support services Krishnamurthy, 2005
to enterprise customers, upgrade services only to open source
product
Third-Party Software Provider Provide services for all types of products Krishnamurthy, 2005
Pure OS Business Model Firms that offer only OS products and OS solutions Bonaccorsi et al., 2006
Hybrid Business Model “They (hybrids) distribute OS products but also develop custom- Bonaccorsi et al., 2006
ized solutions using OS software, for which they presumably offer
installation, support, and maintenance. The large majority also ac-
tively supply complementary services such as consulting, training,
and to a lesser extent research and development.” (p. 1090)
Professional Open Source (POS) “POS combines the benefits of open source (OS) with the develop- Watson et al., 2005
ment of methodologies, support, and accountability expected from
enterprise software vendors.” (p. 329)
Three features of POS: (1) separation of product adoption and pur-
chase, (2) seed and harvest marketing strategy, and (3) dual growth
of firm and ecosystem
Proprietary Inability to view and modify the source code (regardless of price) Watson, Boudreau, York,
Greiner, & Wynn, 2006
Open Community Volunteers develop and support code with limited or no commercial Watson et al., 2006
interest
Corporate Distribution Organizations create value by providing complementary services Watson et al., 2006
such as: interacting with the community for support, supporting the
software for customers, identifying appropriate OSS for customers
Sponsored OSS Corporations act as primary sponsors of OSS projects. These corpo- Watson et al., 2006
rations provide funding and/or developers to the project.
Second-Generation OSS (OSSg2) This is a combination of corporate distribution and sponsored OSS Watson et al., 2006
(the OSSg2 company provides complementary services around the
products, but OSSg2 companies also provide the majority of the re-
sources needed to create and maintain their products). These types
of corporations strive to provide accountability, talented program-
mers, and a healthy ecosystem.
proprietary ERP systems. OS-ERP option value more in the “Real Options Value of OS-ERP”
and its relation to business model will be discussed section of this chapter.
785
cUstOMIZAtION OF Os-ErP the adopter’s organization. Code of questionable

quality will be harder to modify and maintain.
Proprietary ERP system customization has been
explored extensively in the ERP literature (Gat-
tiker & Goodhue, 2004; Levin, Mateyaschuk, & rEAL OPtIONs VALUE OF Os-ErP
Stein, 1998; Nah & Zuckweiler, 2003; Soh & Sia,
2004, 2005; Soh, Siew, & Tay-Yap, 2000). OS- Much research has been done on real options theory
ERP applications are different from proprietary as a way of evaluating platform change decisions
ERP applications in that they are closer to custom (Fichman, 2004; Fichman, Keil, & Tiwana, 2005;
applications than to packaged software. For ex- McGrath, 1997; McGrath & MacMillan, 2000).
ample, packaged software has limitations in terms In fact, a switch from SAP R/2 to SAP R/3 was
of customization because the source code is not considered a platform decision and evaluated
available. The issues with packaged software, as in terms of real options (Taudes, 2000). These
described by Gross and Ginzberg (1984) of “uncer- models and methods are useful, but seem to be
tainty about package modification time and cost, missing some key variables that would influence
vender viability, and the ability of the package to the option value of OS-ERP.
meet the user needs,” will apply differently to open Real options theory rests on the belief that
source applications. Open source software has the limited commitment can create future decision
code available, so the user is free to change (and rights. Real options theory has a rich history
many times is encouraged to change) the code as borne from the finance and economics literature,
needed. Therefore, the issue of uncertainty about and has been applied to technology in a variety of
package modification time and cost depends solely ways. Real options theory, in terms of technology
on the skill of the programmers, not on some con- positioning projects, proposes that technologies
straint that the modification may not be allowed are desirable if they provide opportunities for
by the proprietary software vendor. However, future rent creation. These investments require
the ability to change source code raises a new set less commitment than if a full plan was created
of problems. The efficiencies that we reap from that did not allow for quitting midstream. Beyond
packaged software (standards, easy maintenance, the specific real options literature that supports
and upgrades) will not necessarily be available the logic of committing to software with the in-
with open source software. Thus customization tent of exercising some future option, the project
of OS-ERP is a double-edged sword. Although it management literature on management informa-
is beneficial to adopters from a flexibility of the tion systems (MIS) also supports this notion. In
software perspective, maintenance, upgrades, and the case of ERP systems and large systems in
version control become new issues with which the general, organizations implementing technology
adopting organization must contend. have decided that a “staged” or incremental ap-
Another issue that comes with OS-ERP is that proach is preferable to “big bang” implementa-
the quality of the code will not always be similar tions. The staged approach allows for less radical
to the acceptable quality of code required by the change introduction in the organization. A logical
implementing organization (Spinellis, 2006). With extension of this trend in project implementa-
proprietary code this is not as much of an issue, as tion approaches to valuing technologies in the
you do not usually have access to all that code, and selection decision seems clear. If companies can
maintenance is definitely the duty of the vendor. commit to a technology, and yet have the option
In the OSS industry, it is many times expected to proceed, as well as options for future rents, the
that the skill for maintaining the code resides in technology is more valuable to the organization.
786
ERP systems—and more specifically, OS-ERP Benaroch & Kauffman, 1999; Clemons & Gu,
systems—tend to allow for future options. 2003; Santos, 1991; Taudes, 1998, 2000). Table
Option valuation will be performed multiple 7 provides important background information
times throughout the life of a project. This is not about these studies.
a one-time pre-implementation exercise. This is For valuing the option of OS-ERP, I propose
important: as implementation techniques have thinking more in line with the thinking behind
changed, so have the options that are afforded McGrath (1997), McGrath and MacMillian
by technology positioning investments. For ex- (2000), Chen and Chen (2005), and Fichman
ample, a company may look at the option value of (2004). These studies apply real options thinking
implementing specific modules of an ERP system. to a qualitative options valuation method. For
Future options may be to continue to implement example, McGrath (1997) provides factors that
other modules of this system, to implement an influence the option value of a technology posi-
integrated best-of-breed addition, to use these tioning option. This framework was expanded
modules as a stand-alone part of their system, with explicit items to characterize each factor in
to integrate this module back to existing legacy later work (McGrath & MacMillan, 2000). This
systems, and so forth. This approach would be theory of real options for technology positioning
in congruence with current project management focuses on the factors that are necessary for spe-
practices, where the project is analyzed at different cific domains: new product development and
steps to ensure the project is proceeding satisfac- R&D type research. Although there are many
torily. Likewise, the option value of investment similarities in R&D investments and IT platform
should be analyzed periodically to determine how investments, there are also many differences
to proceed with the option5—the emphasis being (Fichman, 2004).
here that option value does change with time and These differences are explained by drawing
thus should be examined over time. from four complementary perspectives: technol-
Most of the studies of real options in a tech- ogy strategy, organizational learning, innova-
nology context have used a rigorous finance- tion bandwagons, and technology adaptation to
influenced quantitative methods for evaluating develop 12 factors identified as antecedents of
the value of real options (Benaroch, 2000, 2002; option value in IT platform investments (Fich-
Table 7. Finance-based studies of IT platform decisions as real options
Authors Major Theme Context Case or Conceptual

Clemons & Gu, 2003 Strategic options in IT infrastructure Credit card rates for Case
Capital One (industry level)
Benaroch, 2002 Manage IT investment risk that that helps to Internet sales channel Case
choose which options to deliberately embed in
an investment
Benaroch & Kauffman, 2000 Investment timing POS debit market Case
Taudes, 2000 Evaluates ERP platform change SAP R/2 to SAP R/3 Case
Benaroch & Kauffman, 1999 Investment timing POS debit market Case
Taudes, 1998 Evaluates software growth options EDI growth option Case
Dos Santos, 1991 Applies real options theory to IT investments None Conceptual
using two stages, where the first stage creates
the option for the second stage
787
man, 2004). As the focus of this book chapter is FUtUrE trENDs

on option value of OS-ERP systems, the origi-
nal factors proposed to explain option value of Global Adoption of
platform decisions (Fichman, 2004) are pared Open source ErP
down to the most influential in this context, and
the short definition is contextualized to OS-ERP International research on open source firms is
option valuation. Then, several factors (see Table scarce (Bonaccorsi et al., 2006). Research in the
8) are added based on the previous discussion of area of OS-ERP packages is scarcer. Even scarcer
OS-ERP systems. still is international OS-ERP research. For that
According to real options theory, options are reason, some generalizations about the current
valued more highly if limited commitment is state of “adoption of open source” research are
required to take advantage of the option. In terms attempted based on what is happening in open
of OS-ERP options, less commitment will be source software in general. As well, several open
required if the organization already possesses the source ERP participants were asked their opinions
resources to customize, maintain, and upgrade of the international open source ERP market.
the OS-ERP software. Future rents may be cre- There is clearly international activity in the
ated by customizing and upgrading the OS-ERP open source arena, as is evidenced by a study
software. Less resources are required if the qual- performed by Lancashire (2001) where he showed
ity of the source code is high. As well, a fit between that contributors to open source software devel-
the business model of the OS vendor and the type opment were shifting to an international origin.
of services and future solutions the organization However, even this study had a very small sample
might need would require less commitment in (two projects) and there are questions as to the
terms of gaining acceptance of the vendor. Future generalizability of these findings. More recently,
rents will be created if the business model of the evidence of the activity internationally in open
vendor proves to be viable and the vendor is able source projects comes from the samples taken
to survive. This phenomenon is not unlike the dot. for some open source research. Bonaccorsi et al.
com era where the business model was as impor- (2006) performed a survey of 175 partners and
tant as the technology. As was discussed earlier, system administrators of Italian open source6
these factors cannot be ignored in evaluation of firms. Their sample was drawn using a snowball-
an OS-ERP software option. ing technique where initial contacts refer other
Table 8. Factors for valuing OS-ERP options
Factor Short Definition

Susceptibility to network externalities (Fich- The extent to which a technology increases in value to individual adopters with the size
man, 2004) of the adoption network. In the case of OS-ERP, this is particularly relevant as future
versions of the software depend on the adoption network, the partners, and developers.
Prospects for network dominance of the technol- The extent to which the technology instance being adopted is likely to achieve a dominant
ogy instance (Fichman, 2004) position relative to competing technology instances within the same class. In the case of
OS-ERP, a dominant position may be achieved by a large network of adoption, partners,
and developers, as well as venture capitalists.
Customization The extent to which an organization possesses the resources to customize,
maintain, and upgrade OS-ERP software.
Quality of Source Code The extent to which the source code associated with an OS-ERP is of high quality.
Business Model of OS Vendor The extent to which the business model of the OS vendor is acceptable and viable to
the organization.
788
potential participants. This sample hints at the cONcLUsION

breadth of OSS in European countries.
Currently, several open source ERP partici- This chapter outlines the current state of several
pants based in the United States see potential for OS-ERP packages (see Table 1) along the dimen-
international growth. One open source ERP com- sions of licensing, geographic origins, number of
pany estimates that about 15% of their business customers, number of partners, targeted organiza-
is from outside the United States and the CEO tion size, and functionality. Then the several fac-
sees “enormous growth opportunity internation- tors that are important to decision making about
ally.” Another participant involved with an open OS-ERP packages are discussed: susceptibility
source ERP package based in the United States to network externalities, prospects for network
reports that in 2005 about 80% of his income dominance, customization, quality of the source
came from international sources. Then there are code, and business model of the OS vendor.
open source vendors that are based all over the These factors will be important to OS-ERP
world. The small sample of ERP vendors studied packages crossing the credibility gap. Currently,
for this research shows that the base country for as companies scour the landscape looking for an
these organizations varies greatly (see Table 3). ERP package, there are very few open source of-
There are many reasons that open source ERP is ferings that have established credibility as reliable,
being developed internationally and development maintainable, and scalable. Once several OS-ERP
services sold internationally. The requirements of packages cross this hurdle, more widespread use of
ERP systems vary from country to country. For OS-ERP can be expected. As well, since the cost
example, China has different requirements of ERP of ownership of an ERP package is so complex,
packages than the U.S. (Liu & Steenstrup, 2004) . marketing OS-ERP packages as more affordable
So, where certain packages dominate the U.S. ERP is probably not going to gain much traction or
market, they are less diffused in the Asian market. gain the attention of major decision makers in
Similar considerations will occur with open source the ERP space. Other features of OS-ERP, like
ERP applications internationally. There is more customizability (given an adequate upgrade path)
room for competition from open source ERP pack- and a strong OS-ERP network, most specifically
ages in countries where the requirements are not customers and consulting partners, will be desir-
as mainstream. This observation helps explain the able to decision makers.
proliferation of open source ERP developments This chapter hopes to enlighten practitioners
around the world. and academia about the growing field of OS-ERP
As for the services of U.S. developers being systems and their role in the international ERP
sold around the world, some U.S.-based open community. There are significant differences in
source ERP participants feel that the international the offerings of open source ERP packages, and
opportunity is greater because the dollar is weaker, theoretically grounded evaluation techniques of
and so development by U.S. programmers is more OS-ERP are little researched or published. The
affordable to those outside the U.S. The econom- aforementioned topics are important for creating
ics of this assumption also work in the reverse. a clear picture of what is currently on the market
There are many instances of international partners and the future directions of the OS-ERP market.
to U.S.-based ERP projects that are called on to Given the data presented in this chapter, it is clear
develop customizations for U.S.-based users of that while OS-ERP is entering the ERP market,
the software because this international labor is OS-ERP is far from a strong player. Also, OS-ERP
more affordable. is currently marketed as a small and medium en-
terprise option, with some governmental agencies
789
also considering its use. Future research should at- be tested through rigorous research. This could
tempt to explore the value gained and experiences be done by looking at how much customization
of organizations actually using OS-ERP systems. is actually done to OS-ERP packages, and how
upgrades and maintenance are handled on these
customized systems. ERP customization litera-
FUtUrE rEsEArcH DIrEctIONs ture would benefit from such research. This is a
partial list of possible future research directions.
OS-ERP is a very little researched area. Future All of the topics covered in this chapter: factors
research should be performed that rigorously infor evaluating options, global OS-ERP systems,
terviews OS-ERP participants about motivation, and OS-ERP organizations require much more
as this will add to the current literature about OSS research for academia to build a complete and
motivation (several motivation articles are listed coherent picture of OS-ERP and its impacts.
in the Additional Reading section). Many open
source software initiatives fill a gap in functional-
ity and thus are interesting to programmers. ERP rEFErENcEs
systems are mature enough to lack such gaps, and
thus figuring out what interests programmers that Adamson, C. (2004). Java.net: The source for
take part in OS-ERP projects could shed more light Java technology collaboration. Retrieved No-
on the issue of motivation. Also, OS-ERP in the vember 4, 2006, from http://today.java.net/pub/a/
global context is an area ripe for research. Case today/2004/06/01/ofbiz.html
study research of a large multinational implemen- Adempiere. (2006). Adempiere: It’s just a com-
tation of OS-ERP would inform academia as to munity—Nothing personal. Retrieved from http://
how large organizations work with open source adempiere.red1.org/
communities and whether the functionality of-
fered by OS-ERP systems is adequate in such a Apache. (2007). Apache license, version 2.0. Re-
setting. Survey research with global companies trieved March 8, 2007, from http://www.apache.
might shed light on the differences in compa- org/licenses/LICENSE-2.0.html
nies that adopt OS-ERP solutions and those that Benaroch, M. (2000). Justifying electronic bank-
choose proprietary solutions. Global public sector ing network expansion using real options analysis.
research is also needed, as the motivations and MIS Quarterly, 24(2), 197. doi:10.2307/3250936
evaluation criteria for public sector organizations
is noted in the popular press to be different from Benaroch, M. (2002). Managing information
private companies. technology investment risk: A real options per-
To build on the ideas in this chapter, the five spective. Journal of Management Information
factors should be included in a survey of those Systems, 19(2), 43–84.
that have adopted and not adopted open source
Benaroch, M., & Kauffman, R. (1999). A case
research to determine how these factors impacted
for using real options pricing analysis to evalu-
the decision, and whether there were future impli-
ate information technology project investments.
cations of any of these factors on the success of the
OS-ERP implementation. Each factor should be
doi:10.1287/isre.10.1.70
further researched in terms of options evaluation in
the OS-ERP domain. For example, the author has
posited that OS-ERP is closer to custom software
than to packaged software. This assertion should
790
Bonaccorsi, A., Giannangeli, S., & Rossi, C. Fichman, R. G. (2004). Real options and IT
(2006). Entry strategies under competing stan- platform adoption: Implications for theory and
dards: Hybrid business models in the open source practice. Information Systems Research, 15(2),
software industry. Management Science, 52(7), 132–154. doi:10.1287/isre.1040.0021
1085–1097. doi:10.1287/mnsc.1060.0547
Fichman, R. G., Keil, M., & Tiwana, A. (2005).
Caton, M. (2006). OpenMFG: ERP basics and Beyond valuation: “Options thinking” in IT project
more. eWeek, 23, 44-45. management. California Management Review,
47(2), 74–96.
Clemons, E. K., & Gu, B. (2003). Justifying
contingent information technology investments: Fisterra (2007). Fisterra.org: A short history.
Balancing the need for speed of action with Retrieved March 8, 2007, from http://community.
certainty before action. Journal of Management igalia.com/twiki/bin/view/Fisterra/ProjectHistory
Information Systems, 20(2), 11–48.
Fitzgerald, B. (2006). The transformation of open
Cliff, S. (2003, 3/6/2003). Survey finds big varia- source software. MIS Quarterly, 30(3), 587–598.
tion in ERP costs. Computer Weekly, 8.
Gattiker, T., & Goodhue, D. (2004). Understand-
Compiere. (2006a). Andre Boisvert joins Compiere ing the local-level costs and benefits of ERP
team. Retrieved from http://www.compiere.org/ through organizational information processing
news/0522-andreboisvert.html theory. Information & Management, 41, 431–443.
doi:10.1016/S0378-7206(03)00082-X
Compiere. (2006b). Compiere appoints open
source thought leader to its board of directors: GNU. (2007a). GNU general public license, ver-
Larry Augustin will help drive continued growth sion 2, June 1991. Retrieved March 8, 2007, from
of leading open source ERP and CRM provider. http://www.gnu.org/licenses/gpl.txt
Retrieved November 3, 2006, from http://www.
GNU. (2007b). GNU’s not Unix! Free software,
compiere.org/news/0724-augustin.html
free society. Retrieved March 8, 2007, from http://
DiMaio, A. (2004). Look beyond TCO to judge www.gnu.org/
open source software in government. Gartner
Gross, P. H. B., & Ginzberg, M. J. (1984). Barriers
(G00123983).
to the adoption of application software packages.
DiMaio, A. (2007). When to use custom, propri- SOS, 4(4), 211–226.
etary, open-source or community source software.
Guth, R., & Clark, D. (2005). Linux feels growing
Gartner (G00146202).
pains as users demand more features. Wall Street
Feller, J., & Fitzgerald, B. (2000). A framework Journal, (August 8), B1.
analysis of the open source software development
HK-Software. (2007). HK-Software features and
paradigm. In Proceedings of the 21st International
modules, reasons for using AvERP. Retrieved
Conference in Information Systems (ICIS 2000).
March 8, 2007, from http://www.hk-software.
Ferguson, R.B. (2006). Open-source ERP grows net/h-k.de/content/doc_138099-3-5-1.php
up. eWeek, 23(27), 26-27.
Hoover, L. (2006). Compiere is on the move—
again. NewsForge The Online Newspaper for
Linux and Open Source.
791
Krishnamurthy, S. (2005). An analysis of open Nolan, S. (2005). Knowing when to embrace open
source business models. In J. Feller, B. Fitzgerald, source. Baseline, (48), 76.
S. Hissam, & K.R. Lakhani (Eds.), Making sense
Open Source Initiative. (2005). Open source defi-
of the bazaar: Perspectives on open source and
nition. Retrieved from http://www.opensource.
free software. Boston: MIT Press.
org/
Lancashire, D. (2001). Code, culture and cash:
Open Source Initiative. (2006). OSI Web site.
The fading altruism of open source development.
Retrieved October 24, 2006, from http://www.
First Monday, 6(12).
opensource.org/licenses/
Levin, R., Mateyaschuk, J., & Stein, T. (1998).
OpenPro. (2007). OpenPro: The open source ERP
Faster ERP rollouts. Information Week.
software solutions that give you more value and
Liu, L., & Steenstrup, K. (2004). ERP selection more features. Retrieved March 8, 2007, from
criteria for Chinese enterprises. Gartner (COM- http://www.openpro.com
22-0114).
Prentice, B. (2006). The advent of open-source
McAllister, N. (2005). You can’t kill TCO. Info- business applications: Demand-side dynamics.
World, (August): 29. Gartner (G001412).
McGrath, R. G. (1997). A real options logic for Raymond, E. S. (1999). The cathedral and the
initiating technology positioning investments. bazaar: Musings on Linux and open source by
Academy of Management Review, 22(4), 974–996. an accidental revolutionary. Sebastopol, CA:
doi:10.2307/259251 O’Reilly and Associates.
McGrath, R. G., & MacMillan, I. C. (2000). As- Rist, O. (2005). Open source ERP. InfoWorld,
sessing technology projects using real options 27(32), 43–47.
reasoning. Research-Technology Management,
Santos, B. L. D. (1991). Justifying investments in
43(4), 35–49.
new information technologies. Journal of Manage-
Mozilla. (2007). Mozilla public license, version ment Information Systems, 7(4), 71.
1.1. Retrieved March 8, 2007, from http://www.
Serrano, N., & Sarriegi, J. (2006). Open source
mozilla.org/MPL/MPL-1.1.html
software ERPs: A new alternative for an old
MySQL. (2007). The world’s most popular open need. IEEE Software, 23(3), 94–97. doi:10.1109/
source database: About MySQL AB. Retrieved MS.2006.78
March 8, 2007, from http://www.mysql.com/
Soh, C., Kien, S. S., & Tay-Yap, J. (2000). Cul-
company/
tural fits and misfits: Is ERP a universal solution?
Nah, F. F. H., & Zuckweiler, K. M. (2003). ERP Communications of the ACM, 43(4), 47–51.
implementations: Chief information officer’s per- doi:10.1145/332051.332070
ceptions of critical success factors. International
Soh, C., & Sia, S. K. (2004). An institutional per-
Journal of Human-Computer Interaction, 16(1),
spective on sources of ERP package-organization
5–22. doi:10.1207/S15327590IJHC1601_2
misalignments. The Journal of Strategic Infor-
Nichols, D., & Twidale, M. (2003). The usability mation Systems, 13(4), 375–397. doi:10.1016/j.
of open source software. First Monday, 8(1). jsis.2004.11.001
792
Soh, C., & Sia, S. K. (2005). The challenges of Crowston, K., & Howison, J. (2005). The social
implementing “vanilla” versions of enterprise structure of free and open source software devel-
systems. MIS Quarterly Executive, 4(3), 373–384. opment. Retrieved from http://www.firstmonday.
org/issues/issue10_2/crowston/index.html
Spinellis, D. (2006). 10 tips for spotting low-
quality open source code. Enterprise Open Source Fitzgerald, B., & Kenny, T. (2004). Developing
Journal, (September/October). an information systems infrastructure with open
source software. IEEE Software, 21(1), 50–55.
Sugar, C. R. M. (2007). SugarCRM: Commercial
doi:10.1109/MS.2004.1259216
open source. Retrieved March 8, 2007, from http://
www.sugarcrm.com/ Gallivan, M. J. (2001). Striking a balance between
trust and control in a virtual organization: A con-
Taudes, A. (1998). Software growth options.
tent analysis of open source software case studies.
Information Systems Journal, 11(4), 277–304.
15(1), 165–185.
doi:10.1046/j.1365-2575.2001.00108.x
Taudes, A. (2000). Options analysis of software
Gattiker, T. F., & Goodhue, D. L. (2005). What hap-
platform decisions: A case study. MIS Quarterly,
pens after ERP implementation: Understanding
24, 227. doi:10.2307/3250937
the impact of inter-dependence and differentiation
Watson, R., Wynn, D., & Boudreau, M. C. (2005). on plant-level outcomes. MIS Quarterly, 29(3).
JBOSS: The evolution of professional open source
Hars, A., & Ou, S. S. (2002). Working for free?
software. MIS Quarterly Executive, 4(3), 329–341.
Motivations for participating in open-source
Watson, R. T., Boudreau, M.-C., York, P., Greiner, projects. International Journal of Electronic
M., & Wynn, D. (2006). (forthcoming). The busi- Commerce, 6(3), 25–39.
ness of open source. Communications of the ACM.
Hertel, G., Niedner, S., & Herrmann, S. (2003).
Motivation of software developers in open source
projects: An Internet-based survey of contribu-
ADDItIONAL rEADING tors to the Linux kernel. Research Policy, 32(7),
1159–1177. doi:10.1016/S0048-7333(03)00047-7
Al Marzoug, M., Zheng, L., Rong, G., & Grover,
V. (2005). Open source: Concepts, benefits, and Krishnamurthy, S. (2002). Cave or community?
challenges. Communications of the AIS, (16), An empirical examination of 100 mature open
505-521. source projects. First Monday, 7(6).
Benkler, Y. (2002). Coase’s penguin, or, Linux Lacy, S. (2005). Open source: Now it’s an eco-
and the nature of the firm. The Yale Law Journal, system. Business Week, (October): 7.
112(3), 369–446. doi:10.2307/1562247 Lerner, J., & Tirole, J. (2002). Some simple eco-
Brown, C. V., & Vessey, I. (2003). Managing the nomics of open source. The Journal of Industrial
next wave of enterprises systems: Leveraging Economics, 50(2), 197.
lessons from ERP. MIS Quarterly Executive, Liang, H., & Xue, Y. (2004). Coping with ERP-
2(1), 65–77. related contextual issues in SMEs: A vendor’s
perspective. The Journal of Strategic Informa-
tion Systems, (13): 399–415. doi:10.1016/j.
jsis.2004.11.006
793
Mabert, V. A., & Watts, C. A. (2005). Enterprise Zhao, L., & Elbaum, S. (2003). Quality assurance
applications: Building best-of-breed systems. under the open source development model. Journal
In E. Bendoly & F.R. Jacobs (Eds.), Strategic of Systems and Software, 66, 65–75.
ERP extension and use. Stanford, CA: Stanford
University Press.
Madanmohan, T. R., & Krishnamurthy, S. (2005). ENDNOtEs
Can the cathedral co-exist with the bazaar? An
analysis of open source software in commercial
1
http://www.sugarcrm.com/crm/SPL-The
firms. First Monday, (Special Issue #2). Retrieved SugarCRM Public License Version (SPL)
from http://firstmonday.org/issues/special10_10/ consists of the Mozilla Public License
madanmohan/index.html Version 1.1, modified to be specific to
SugarCRM. Please see the actual license to
Nelson, M., Sen, R., & Chandrasekar, S. (2006). understand the terms of this license.
Understanding open source software: A research 2
Intensive and complete functionality for
classification framework. Communications of the ERP vendors includes the aforementioned
AIS, 17(12), 266–287. functional areas, though the specifics of
Niederman, F., Davis, A., Greiner, M. E., Wynn, what is included in each functional area may
D., & York, P. (2006). A research agenda for be termed differently by different vendors
studying open source II: View through the lens (i.e., Oracle and SAP); however, both offer
of referent discipline theories. Communications much of the same functionality and this
of the AIS, 18. base functionality is what is intended by the
author when discussing ERP systems. For
Niederman, F., Davis, A., Wynn, D., & York, P. details about this functionality, see SAP.com
(2006). A research agenda for studying open source or Oracle.com.
I: A multi-level framework. Communications of 3
See the Free Software Foundation Web site
the AIS, 18. at http://www.fsf.org/licensing/licenses/ for
Stewart, K., & Gosain, S. (2006). The impact of another source for OSS licensing resources.
ideology on effectiveness in open source soft-
4
“Forking” is not unusual in open source com-
ware development teams. MIS Quarterly, 30(2), munities. Forking refers to taking the code
291–314. in a separate direction, usually a direction
not intended by those managing the open
Watson, R. T., Boudreau, M.-C., Greiner, M., source project. There are dissenting opinions
Wynn, D., York, P., & Gul, R. (2005). Governance as to whether this is cause for alarm for the
and global communities. Journal of International original open source project.
Management, 11(2), 125–142. doi:10.1016/j.int- 5
For a review of potential option outcomes,
man.2005.03.006 see Fichman et al. (2005).
6
Open source is defined as firms that supply
OS-based offerings, even if the offering
includes proprietary solutions.
of IGI Global).
794
795
Chapter 3.15
Real Time Decision Making
and Mobile Technologies
Keith Sherringham
IMS Corp, Australia
Bhuvan Unhelkar
AbstrAct device anywhere anytime. This chapter discusses

the role of mobile technologies in real time deci-
For business decision making to occur, data needs sion making.
to be converted to information, then to knowl-
edge and rapidly to wisdom. Whilst Information
Communication Technology (ICT) solutions INtrODUctION
facilitate business decision making, ICT has not
always been effective in providing the critical As society exits the industrial age and enters the
“data to wisdom” conversion necessary for real- knowledge era, society suffers from data overload,
time decision making on any device anywhere information is lacking, knowledge is scarce and
anytime. This lack of effectiveness in real-time wisdom is wanting (Balthazard, & Cook 2004).
decision making has been further hampered by Instead of having the right information, presented
a dependence upon location and time. Mobile at the right time in the right way to make decisions,
technologies provide an opportunity to enhance society is epitomised by people spending large
business decision making by freeing users from amounts of time trawling and sifting through data
complex information management requirements to try and find what is needed to make decisions
and enabling real-time decision making on any (Adair 2007). This practice of searching and sifting
through data in an effort to find information poses
DOI: 10.4018/978-1-60566-156-8.ch016 a huge in-built inefficiency with higher costs and
Real Time Decision Making and Mobile Technologies
un-assured service delivery. Advances in mobile The familiar example is a hall porter, who by
technology are likely to create further challenges putting a favourite wine in a hotel room collects
to these searches and sorts as they bring in addi- a reward. The hall porter takes the data elements
tional dimensions of location-independence and (arrival at that hotel, to the appropriate room, at
personalization. that correct time and with the wine) and because
The need for the rendering of information the elements are required and understood they are
in context, as part of work-flow, to any device information. Context is given to the information
anywhere anytime to enable real time decision (managing the relationships between pieces of
making is the goal of many organisations. The information and used with work-flow) to achieve
mobile enablement of business, as discussed by knowledge, which is then applied at the right time
Sherringham and Unhelkar (2008a) in a separate in the right way to realise a profit (Ghanbary and
chapter in this book, is expected to further drive Arunatileka 2006).
the demand for real time decision making ser- The business imperative of the data – wisdom
vices (Ghanbary 2006). The emergence of real conversion has been widely noted, e.g. Macmanus
time decision making, the elements required to et al. (2005), but the significance of the value
achieve that process and the implications of mo- stack lies in the importance of providing context
bile technologies in real time decision making are through managing relationships between pieces
discussed in this chapter. of information and by the integration of informa-
tion with work-flow. Within mobile business,
the significance of location provides additional
DAtA – WIsDOM VALUE stAcK context to the information.
Resolution of the data – knowledge conversion
A record in a database, a marketing video or a allows the right information to be presented at the
company’s financial report are all data. Data only right time in the right way to the right audience,
becomes information when it is analysed, under- providing two advantages. Firstly, the need for
stood and needed. With the application of experi- users to have advanced information management
ence and skills information becomes knowledge skills to complete the most rudimentary of tasks
and when such knowledge is applied at the right is reduced. Secondly, the difficulty of managing
time in the right way, knowledge becomes wisdom information on the small screen of current mo-
(power /profit). This data to wisdom conversion bile devices is removed. Resolution of the data–
is illustrated in Figure 1. knowledge conversion to service both business
and mobile business will allow mobility to realise
its true significance through the provision of real
Figure 1. Data – Wisdom value-stack
time decision making (Raton 2006) and provide
business with a major competitive advantage
(Ekionea 2005).
ELEMENts OF rEAL tIME

DEcIsION MAKING
With real time decision making, our favourite

restaurants can bid in real time to achieve our
patronage on any device anywhere anytime. The
information needed to take the best route home
796
is supplied dynamically as the road is traversed pieces of information often need to be drawn to-
and everything that is needed to make a foreign gether, i.e. the relationships between elements of
exchange trade is rendered to any device anywhere information need to be provided to give context.
anytime for decision and execution (Gupta 2006). Information relationships are managed using a
The key elements of real time decision mak- metadata framework that provides all of the sup-
ing are summarised in Figure 2 and are discussed porting details to give context and by linking it
as follows. to steps in a work-flow. A metadata framework
Consolidated Repository – Information for includes classification schema, versioning details,
real time decision making is accessed from vir- role based access, security and privileges, as well
tual consolidated repositories that combine spatial as device specific information and spatial needs.
data, database data, transactional data and docu- Use of metadata to manage information re-
ments (including images). To stop the duplication lationships is critical in the data to knowledge
of effort, information is single sourced from vir- conversion and is a prerequisite to the provision
tual consolidated repositories. Within these re- of more sophisticated mobile business services.
positories, content is separated from presentation Work-flow – By defining and integrating
and from mechanism of delivery. Archiving, information to process, knowledge can be pre-
backup, recovery and version control, are all sented in context for real time decision making
performed on the repositories on behalf of the (Ekione and Abou-Zeid 2005). The user is taken
user, freeing up both the end user and the end- through a series of recipes (process and informa-
device (Sherringham 2005). tion combined in sequence - recipes) to realise the
The benefits of distributed computing power required outcomes. The combination of informa-
shall continue within the mobile computing en- tion relationships and work-flow together provide
vironment but unlike the desktop environment context and allows the conversion of data into
where data were trapped locally, consolidated information and to knowledge. The challenge is
data storage ensures that data are single sourced the development and maintenance of the standard
and that data are not stored for extended periods recipes by audience and of the virtual consolidated
of time on the end device (Sherringham 2008). contextual information bases.
To effectively support mobile computing, large When providing mobile business services and
centralised databases shall be replaced with virtual products, it is ease of use that influences user ac-
consolidated contextual information bases (Yang ceptance and product adoption. The use of clearly
and Wang 2006). defined recipes to guide the user to the assured
Information Relationships – To achieve outcome is part of the mobile business solution.
knowledge and to make decisions, different
Figure 2. Elements of real time decision making
797
Unified search – Search shall be unified processing or archiving for compliance shall also
across all data types and tightly integrated with be part of the messaging environment.
work-flow. Search harnesses the information Presentation – With a resolution of the infor-
relationships to provide context and uses role- mation to be provided, the provision of context
based access and other usability information to and work-flow, the remaining element is the pre-
provide focused results. The effectiveness of a sentation to the end of device. The interface will
search capability is a pivotal tool to the provision consist of a series of intuitive icons that provide
of services on mobile devices. the user with the required business functionality.
Artificial intelligence – Work-flow, search Invoking an icon changes the interface to reflect
and information relationships are all expected to the task on-hand with the work-flow embedded.
merge into one layer that is artificial intelligence. The ability to customise and personalise the inter-
It is the presence of artificial intelligence that will face would also exist, together with an integrated
liberate the end user and empower mobile business. search (Sherringham and Unhelkar 2008b).
Security – As part of the overall security
issue, a metadata framework facilitates the se-
curity provision and management because of the Ict INDUstry trENDs AND
application of user access rights to the specific rEAL tIME DEcIsION MAKING
information elements.
Messaging – Real time decision making Involving telecommunications companies, hard-
requires a unified messaging environment to ware and software suppliers, content providers
combine voice, data, text, images and video. To and consulting services, real time decision making
guarantee service delivery the messaging environ- on any device anywhere anytime is an emerg-
ment needs to be architected around the FedEx ing trillion dollar business opportunity that will
model, where the quality of hand-off, the message, underpin business operations. The emergence of
the delivery and storage of the message are all real time decision making from the evolving use
separated and utility infrastructure underpins the of the Internet is shown in Figure 3.
message processing. Brochureware remains a major use of the In-
Supporting the messaging environment are ternet and will transition to mobile devices. Al-
the device and asset management functionality though the Internet is increasingly being used to
and capability, including locational information. support transaction processing, many opportuni-
Tagging of messages for application specific ties for development still exist because the trans-
Figure 3. Evolution of the Internet and the emergence of real time decision making
798
action processing capability is still in its infancy development, enterprise architecture, business
and advanced and complex transaction processing process and information management (Alag 2006).
is currently not well supported. Basic transaction Iconic interface – Currently, an iconic inter-
processing is starting to be seen on mobile de- face is the most effective approach for providing
vices but complex transaction processing requires a common and intuitive interface on the small
some significant changes in business processes screen of a mobile device through to the more
and resolution of security and authentication spatial screen real estate of the plasma television
before a wide adoption on mobile devices occurs. set. Iconic interfaces are increasingly used for
The emergence and growth of collaboration many features of applications with the display
is the focus for much of the ICT industry and is automatically changing to reflect those most
increasingly providing other business opportuni- relevant features to the needs of the user, e.g.
ties. Collaboration includes Web 2.0, with social Apple’s I-phone and the latest release of Microsoft
networking, business networking and the need for Office. The ability to personalise the interface and
entertainment. The greater use of the Internet for to be independent of location are extra features
messaging and the convergence of telecommuni- to be seen within mobile devices.
cations with the computer are also aspects of the Features to functionality - Whilst the trend
collaboration stage in the evolution of the Internet for more and more features to be supplied within
and mobile business. Whilst mobile devices will software applications that are used by fewer
play a significant role within collaboration, the people is still strongly present, this trend is set
maturity of business processes and of manage- to change because of the need for a functionally
ment frameworks remains an issue for business driven iconic interface on mobile devices. An
collaboration. iconic interface that includes work-flow and which
The natural extension of collaboration is real changes to reflect the functionality required by
time decision making. Effective collaboration a user to achieve an outcome, removes the user
requires that the right information be shared with from the need for intimate knowledge of feature
participants so that informed decisions can be driven software. The successful software of the
made and then executed. With the provision of future delivers functionality and NOT features.
information in context linked to work-flow, real Application consolidation - The use of an
time decision making promises to add significantly iconic interface that invokes functionality in-
to business capabilities. stead of features creates a blurring in the need
Whilst many business talk about collaboration for distinct killer applications because it is about
and the provision of collaboration services, the bringing elements of functionality together to
value lies in real time decision making because meet a need. The focus will no longer be on
of its higher value and range of opportunities. launching an application to complete a task but it
Compared with the Internet wave that swept is about conducting a task and bringing together
through society when Web sites and the Internet the functionality necessary to achieve the required
came to the fore, the changes seen in society as a outcomes. Whilst Microsoft’s use of Outlook as a
result of real time decision making will transform master application to drive desktop functionality
humanity. Strategically aligning both business and is another step in the seamless integration of ap-
ICT now to support and adopt real time decision plications, this approach is still evolving.
making is required. Microsoft PowerPoint allowed people to be a
Real time decision making brings together presenter. Microsoft Word allowed people to be
some key and innovative technologies (Table I) a typist and Microsoft Excel allowed people to
whilst providing a unified approach to software be accountants. It was Microsoft’s close linking
799
Table I. Summary of ICT trends impacting upon real time decision making
Trend Description Opportunity

Iconic interface Iconic interface of business functionality that includes work- Control the framework that manages the
flow that changes to meet user need. Spans both the desktop and interface and control access to the rest.
mobile devices.
Features to function- Move away from feature rich software used by few people to The software of the future is development
ality functionality driven applications used by many people across all of functionality driven software, includ-
devices. ing work-flow, as object of functionality,
delivered as required to any device.
Application consoli- Focus is not on application specific software but bringing together Standardisation and dominance over
dation objects of functionality. The killer applications that standardised mobile computing, desktop computing and
the desktop will be replaced by integration of elements of func- enterprise computing shall come from stan-
tionality rendered on any device. dardised seamless integration of elements
of functionality.
Sound practice Return to consolidated data storage and work-flow; separation of Position now and implement best practices
content from presentation and mechanism of delivery and use of in future developments ready to support
smart end device with the load being taken by the server. mobile computing.
Software as a Service Real time decision making is a natural extension of SaaS; with Real time decision making on mobile de-
(SaaS) only those services required being rendered as needed. vices will drive the definition of new SaaS
opportunities.
End device & operat- In the emerging mobile market, the de-facto standard for hard- Own the operating system of the mobile
ing system ware and end device operating system are still to be realised. device and define the standards for the end
device.
Data storage Increasing demand for consolidated data storage is required to Standardise and become the market leader
support mobile business. in the development and provision of global
virtual consolidated databases that Google
has already started.
Consolidated con- The advances required are in information in context integrated The software to support consolidate con-
textual information with work-flow – contextual information bases. These will be textual information bases requires a new
bases virtual consolidated information bases. generation of database software that will
underpin access from millions of users on
mobile devices.
Context based Google became a billion dollar company searching data in an Realsie the value in searching and manag-
searching effort to find information. Searching in context is the evolving ing knowledge. Realise the knowledge
opportunity. utility.
of killer applications with its operating system end device but also some bad trends occurred.
that allowed for standardisation of the desktop Firstly, the end device was burdened with more
environment and led to market dominance. In the and more applications which has caused update
world of real time decision making and mobile and coordination problems. Secondly the benefits
devices, it is the seamless integration of elements of consolidated data storage and work-flow were
of functionality that shall lead to standardisation lost. Furthermore, content was no longer separated
and dominance over mobile computing, desktop from presentation and mechanism of delivery.
computing and enterprise computing (Armstrong Through real time decision making and the
2006). demands from mobile business, some of the best
Return of sound practices – In the mainframe practices can be returned to ICT (Kaliszewski
environment, the end device was dumb and ev- 2006). To work effectively, the end device (mo-
erything was done centrally on a mainframe. The bile device) will be smart but it cannot become
introduction of the PC realsied the benefits of overloaded like the desktop PC. The bulk of the
distributed computing power and of an intelligent work will be done at the server end with results
800
only being displayed on the mobile device and the computer, with television, with the gaming console
prompt for the next stage of the process. The mobile and with the music and video player.
end device remains simple and easily managed. Whilst the PC environment was characterised
Using mobile devices to access contextual in- by an integration of killer applications with an
formation from consolidated contextual informa- operating system that led to standardisation, the
tion bases means that the benefits of consolidated mobile business environment is different. Within
data storage linked to work-flow shall be seen the mobile business environment, there is a de-
within mobile business. The separation of content cline in the importance of killer applications but
from presentation and mechanism of delivery sees a greater significance in elements of business
the return of another good practice and a requisite functionality. Close integration of elements of
for real time decision making is implemented. In business functionality through an iconic interface
addition, real time decision making requires that with the mobile device operating system is the
the business logic and processing rules (Lucas path to standardisation and market dominance
2005) be stored in contextual databases and not in for software vendors.
the source code – a return to good coding practices. Data storage – The demand for data stor-
Software as a Service (SaaS) – The provi- age capability is set to rapidly increase. It is not
sion of Software as a Service is in its infancy and so much in the storage of transaction data and
the mobile opportunities for SaaS are still to be documents that will further significantly pressure
realised. Much of SaaS is still very application data storage but it is the growth in imaging for
specific and there have been concerns about the work-flow, results for simulation, images from
quality and applicability of code supplied with surveillance and services for entertainment. The
SaaS. Real time decision making and the elements storage will not be on the end device but on the
of business functionality in the interface is a natural server in global virtual consolidated databases
extension to current SaaS. Mobile business will (Raisinghani 2006).
provide many new opportunities for SaaS. Consolidated contextual information bases
End device and operating system – Like – Real time decision making and mobile business
any other market, the evolution of the desktop depends upon the use of contextual information
environment led to a highly diversified market bases to underpin its operation. These databases
with many players in hardware, software and shall function along distributed lines and shall be
killer applications. As the market evolved and viewed as one virtual global consolidated informa-
matured, standards came into effect and it is this tion base. The database software to support con-
standardisation which drove the consolidation solidated contextual information bases requires
and the creation of market dominance by a few a new generation of object orientated databases.
key players. Whether it is a user on a mobile device or a wire-
Within the emerging mobile business market, less napkin holder, these consolidated contextual
the de-facto standard for hardware, both server information bases will service millions of devices.
and end device, is still to be defined. The operat- Context based searching – Contextual based
ing system, for both the server and end device, is searching is key to the operation of real time
also in need of standardisation. Market dominance decision making. The leading search capabilities
comes with being the de-facto standard. The op- present currently are not only data type specific,
portunities that come from standardisation in the e.g. documents or transaction data, but because of
mobile market are more extensive because of the the lack of context and work-flow, the searching is
convergence of the mobile phone with the laptop not contextual. Development of contextual based
801
searching provides any given business with an be realised. Whilst a unified approach may cur-
unprecedented business opportunity and control rently be absent from the industry, the business
over mobile business globally. opportunity is too great for the collaboration not
Artificial intelligence – Contextual based to occur.
searching, the management of information rela- Whilst the demands of mobility shall be the
tionships and linking to work-flow are necessary major driver in realising real time decision mak-
for real time decision making but the challenge ing on any device anywhere anytime, it is the
is the sheer volume of data that currently exists, telecommunications network that underpins real
plus that which will evolve going forward. It is time decision making. Major telecommunications
simply not possible for humans to classify and companies like Verizon and British Telecom are
context this volume of information by audience; deploying fibre optic networks to support not only
nor meet the dynamic nature of information and an integrated communications environment but
the evolving needs. Artificial intelligence shall also the whole scale future transmission of data.
be used. Whilst the mobile dominates the last step to the
Contextual based searching, the management end user, the bulk of data traffic movement shall
of information relationships (metadata) and work- remain the fibre optic backbones. The future shall
flow shall all merge to form a layer of artificial see an operating system layered on the routers and
intelligence (Figure 2). The existing use of pattern fibre optics of the telecommunication network to
matching and predictive capabilities in artificial form one global virtual mainframe to support real
intelligence shall be expanded upon to include time decision making.
perceptive and awareness capabilities. Artificial Whilst the emergence of artificial intelligence
intelligence shall initially come to the fore in er- is the critical layer in the solution that supports
ror and exception handling for routine transaction real time decision making (Figure 2), at least one
processing because the majority of information more layer is to still to be developed, that of voice.
accessed in business is as part of routine transac- Gone will be the days of data entry and manage-
tion processing and it is standardised transaction ment through a keyboard, voice shall be the key
processing that underpins business operations mechanism. Other bio-recognition solutions are
(Moonis 2006). The potential applications of also expected to be developed.
artificial intelligence within mobile are almost The ICT industry has seen standardisation of
limitless and with artificial intelligence present the operating system and of application. Standar-
on all hand held devices, all fixed devices (nap- disation has occurred at the desktop and is being
kin holders, fridges, PCs and TVs), the market seen at the enterprise level. Whilst standardisation
opportunity is unprecedented. of the mobile device is still to be realised, the real
opportunity is standardisation of the marketplace.
Business operates in market places, e.g. banks and
FUtUrE DIrEctION standards to operate globally, and the opportunity
to standardise the marketplace ICT awaits (the
Even though real time decision making is still emergence of marketplace computing).
emerging and there is much work to be done to see Access to information shall become a consumer
its full realisation, some future trends are already right in the knowledge era. In the knowledge era,
starting to be identified. Real time decision mak- information shall underpin society and like power
ing is an emerging trillion dollar industry that will and water, the Internet and knowledge will be a
evolve over the next 10-years and collaboration utility. Of all the utilities (gas, water or electricity),
between key players is how the opportunity will and of all the infrastructures (roads, ports, rail or
802
communications), the knowledge utility will be Alag, H. S. (2006). Business Process Mobility.
the most demanding and the most valuable. In Unhelkar B. (Ed.), Handbook of Research
cONcLUsION Global
Armstrong, M. (2006). A handbook of management
Living in the knowledge era provides many op-
techniques: a comprehensive guide to achieving
portunities and rewards to the individual, to busi-
managerial excellence and improved decision
ness and to society. For humanity to realise its true
making, Rev. 3rd ed. London: Kogan Page
potential in the knowledge era, users need to be
freed from the need for advanced information skills Balthazard, P. A., & Cook, R. A. (2004). Orga-
and empowered by real time decision making on nizational Culture and Knowledge Management
any device anywhere anytime. Information in con- Success: Assessing the Behaviour-Performance
text, sourced from virtual consolidated contextual Continuum. Paper presented at the Proceeding
information bases and integrated with work-flow of the 37th Hawaii International Conference on
for delivery through a common interface across System Sciences, Hawaii, USA.
devices is what is required. Artificial intelligence
Ekionea, J. B., & Abou-Zeid, E. (2005). Knowl-
is how contextual based searching is achieved
edge Management and Sustained Competitive
and how the relationships between information
Advantage: A Resource-Based Analysis. Paper
managed to give context.
presented at the IRMA Conference, SanDiego,
Real time decision making is a trillion dollar
USA.
business opportunity that is set to evolve over
the next 10-years and to become the de-facto Ghanbary, A. (2006). Evaluation of mobile
industry standard. Whilst the demands of mobil- technologies in the context of their applications,
ity shall drive the evolution of real time decision limitations and transformation. In Unhelkar B.
making, it is the benefits derived in routine busi- (Ed.), Chapter 42 of book: Handbook of Research
ness transaction processing that will be the initial in Mobile Business: Technical, Methodological
incentive for realisation. Beyond this, however, and Social Perspectives. Hershey, PA, USA: IGI
lies the moral responsibility for realising real time Global
decision making for the betterment of humanity
Ghanbary, A., & Arunatileka, D. (2006). Enhanc-
because knowledge is freedom and knowledge is
ing Customer Relationship Management through
the liberator from poverty and tyranny.
Mobile Personnel Knowledge Management
(MPKM). Proceedings of IBIMA International
Conference. IBIMA 2006. Bonn, Germany. 19-
rEFErENcEs
21 June.
Adair, J. E. (2007). Decision making & problem Gupta, J. N. D. (2006). Intelligent Decision-
solving strategies 2nd edition, Philadelphia: making Support Systems Foundations, Applica-
Kogan Page tions and Challenges. London: Springer-Verlag
London Limited
Kaliszewski, I. (2006). Soft computing for complex
multiple criteria decision making. New York,
NY: Springer
803
Lucas, H. C. (2005) Information technology: Yang, C. C., & Wang, F. L. (2006). Information
strategic decision making for managers. Hobo- Delivery for Mobile Business: Architecture for
ken, NJ: Wiley Accessing Large Documents through Mobile
Devices. In Unhelkar B. (Ed.), Chapter 18 of
Macmanus, D. J., Snyder, C. A., & Wilson, L. T.
book: Handbook of Research in Mobile Business:
(2005). The Knowledge Management Imperative.
Technical, Methodological and Social Perspec-
Paper presented at the IRMA Conference 2005,
tives. Hershey, PA, USA: IGI Global.
San Diego, USA.
Moonis, A. (2006) Advances in Applied Artificial
Intelligence. Proceedings of 19th International
Conference on Industrial, Engineering and Other
Applications of Applied Intelligent Systems, IEA/ Activity Objects: A series of objects invoked
AIE 2006, Annecy, France, June 27-30. Berlin from a standard iconic interface that provide busi-
Heidelberg: Springer-Verlag GmbH. ness functionality because they contain the neces-
Raisinghani, M. S. (2006). M-Business: A Global sary content, images, business logic, processing
Perspective. In Unhelkar B. (Ed.), Chapter 31 of rules, work-flow and presentation rules.
book: Handbook of Research in Mobile Business: Contextual Search: Searching of information
Technical, Methodological and Social Perspec- in context so that useful results are obtained.
tives. Hershey, PA, USA: IGI Global. FedEx Model: A model for the operation of
a unified messaging environment based on the
Raton, B. (2006). Autonomous mobile robots: proven principles to move messages (parcels)
sensing, control, decision-making, and applica- around the world by leading logistics companies.
tions. FL: CRC/Taylor & Francis Information Bases: The next generation of
Sherringham, K. (2005). Cookbook for Market databases but instead of storing data, information
Dominance and Shareholder Value: Standardising is stored in context.
the Roles of Knowledge Workers. Athena Press: Information Relationships: The associa-
London. (p. 90). tions between elements of information to provide
context and convert information to knowledge.
Sherringham, K. (2008). Catching the Mobility Knowledge Utility: Like power and water,
Wave Information Age April-May 2008 (p. 5). knowledge shall become a utility infrastructure
Sherringham, K., & Unhelkar, B. (2008a). Ele- that underpins humanity.
ments for the Mobile Enablement of Business. Marketplace Computing: Standardised com-
In (Unhelkar et al. 2008) Handbook of Research puting (tightly integrated hardware and software)
in Mobile Business: Technical, Methodological operating at the marketplace level to allow busi-
and Social Perspectives – 2nd Edition, IGI Global. nesses to interact effectively in a marketplace.
Business currently standardises at the enterprise
Sherringham, K., & Unhelkar, B. (2008b). Busi- level but to operate effectively, standardisation
ness Driven Enterprise Architecture and Applica- shall be at the marketplace level.
tions to Support Mobile Business. In (Unhelkar Real Time Decision Making: The provision
et al. 2008) Handbook of Research in Mobile of information in context and integrated with
Business: Technical, Methodological and Social work-flow in real time to any device anywhere
Perspectives. 2nd Edition, IGI Global anytime is needed so that decisions can be made.
This work was previously published in Handbook of Research in Mobile Business, Second Edition: Technical, Meth-
odological and Social Perspectives, edited by Bhuvan Unhelkar, pp. 173-181, copyright 2009 by Information Science
804
805
Chapter 3.16
Business Driven Enterprise
Architecture and Applications
to Support Mobile Business
Keith Sherringham
IMS Corp, Australia
Bhuvan Unhelkar
AbstrAct approach to solutions architecture, information

management, and business integration. At the
Information Communication Technology (ICT) enterprise level, the desktop, the mobile device
needs to provide the knowledge worker with an and at the emerging marketplace level, the evolv-
integrated support system of information manage- ing need for real-time decision making on any
ment and work-flow. This challenge, however, is device, anywhere, anytime, to support mobile
further exacerbated in mobile business wherein the business is providing a framework for aligning ICT
knowledge work is not identified with a particular to business. Further details are presented in this
location. Information systems need to be analyzed chapter together with some of the challenges and
and modeled, keeping the location-independence opportunities to be seen within mobile business.
of the users in mind. A Model Driven Architecture
(MDA) approach, aligned with Object-Orientated
Design principles, and driven dynamically as the INtrODUctION
user interacts, has immense potential to deliver
solutions for the systems used by the knowledge Enterprise architecture, application development
worker. An MDA approach provides a unified and requirements gathering have all faced a com-
mon problem, that of the business environment
DOI: 10.4018/978-1-60566-156-8.ch021 being highly dynamic and continuously evolving.
Business Driven Enterprise Architecture and Applications to Support Mobile Business
An application that worked is often quickly in need rOLE OF KNOWLEDGE

of revision and an existing infrastructure readily MANAGEMENt IN
looses its performance advantage because busi- MObILE bUsINEss
ness needs are continually changing. Although the
demands of mobile business are adding another Through the application of proven business princi-
level of complexity to application development ples, business has standardised catering, cleaning,
and enterprise architecture, the mobile enablement farming, minerals extraction and manufacturing.
of business (Sherringham and Unhelkar 2008a) The last great challenge is the standardisation of
provides a convergence of events to realign In- knowledge workers to lower costs and assure
formation Communication Technology (ICT) as guaranteed service deliver (Sherringham 2005).
the assembly line for knowledge workers. This need for standardisation and the resolution of
Further recognition of ICT as a utility infra- information management and work-flow becomes
structure and all of the utility principles underpin- more pressing when the needs of mobile business
ning design, operation and management of ICT are considered (Sherringham 2008).
can also be realised in the mobile enablement of This situation portrayed in Figure 1 often
business. The significance of a business focused occurs in organisations, where a Customer con-
approach, driven by how the customer interacts, tacts a Service Representative who is faced with
will also be championed during the alignment of querying multiple disparate backend systems to
ICT to meet mobile business (Lan and Unhelkar find the required information to respond to the
2005). Using the demands of mobility, this chapter Customer’s request. The Service Representative
discusses the alignment of enterprise architecture may not find what they want, so they have a dis-
and application development to meet current and cussion with a co-worker who tries to do the same
future needs and how the resulting need for real thing and who may bring in another co-worker.
time decision making will shape some key trends In the mean-time, the Customer gets frustrated
in the ICT industry. and approaches another Service Representative
who goes through the same process. Add to this
the duplication between Internet and Intranet,
Figure 1. Hidden costs of knowledge management present in the enterprise
806
disparate Web sites and the sending of e-mails on their business processes and users (Cummins,
that are not coherently managed and an in built 2002). Through defining the ICT necessary to sup-
hidden cost with a failure to guarantee service port knowledge workers as an assembly line for
delivery is seen. knowledge workers and by addressing the issue of
ncumbent within the desktop environment and integration of information with work-flow (driven
within many enterprise architectures is the isola- by how the customer interacts), the necessary
tion of data in disparate silos with a resulting elements of enterprise architecture can be readily
duplication of effort. A scarcity of context for the defined and the necessary integration required
information and a lacking of integration with determined. This assembly line approach lever-
work-flow further increases hidden costs because ages the modelling capabilities of Model Driven
of the time spent trying to find information. The Architecture to develop platform independent
demand by customers for mobile business ser- models and solutions (McGovern et al. 2004).
vices and because of the constraints imposed by In addition to assembly techniques, there are
mobile devices, a redefinition of enterprise archi- several other principles that have been have been
tectures and an optimisation of the desktop envi- standard engineering practices for many years
ronment shall result.a which can be brought to ICT, applications and
The small screen size inherent in current mobile enterprise architecture to support mobile busi-
devices means that if mobile business services are ness and align information management with
to be provided and accepted by the user, all of work-flow:
the information management currently required
will need to have occurred before delivery to the • Market maturity: The maturity of
mobile device. Mobile business will drive the the market in which mobile business is
implementation of real time decision making. occurring
Instead of users searching and sifting through • Business dynamics: How a business re-
information, the right information is presented at sponds to the forces of markets, customers,
the right time in the right way to allow decisions suppliers and legislation.
to be made, e.g. our favourite restaurants bid in • Business maturity: The maturity of a
real time to achieve our patronage on any device business in the application of ICT.
anywhere anytime (Sherringham and Unhelkar • Utility infrastructure: ICT as a utility in-
2008b). frastructure for mobile business.
The demand for real time decision making
from mobile business is expected to be one of the Market maturity – As mobile business offer-
main drivers for the provision of mobile business ings are brought to market, businesses will operate
services, resolution of information management in an emerging market (size, share or offering).
and for the realignment of ICT to support busi- In such emerging markets there are few standards
ness needs. and the market is highly dynamic. Solutions need
to be rapidly developed and quickly changed to
support growth and product diversification. As
ALIGNMENt DrIVErs FOr Ict markets mature, product diversification is re-
WItHIN MObILE bUsINEss quired, specialist needs arise and standards start
to develop. Change becomes less prevalent and
Having the correct enterprise architecture is an en- the focus moves to assured delivery and scalable
terprise’s key business system, information / data, growth.
application, technology strategy and it impacts
807
In highly mature markets, standards dominate, and reconstruct events impact upon the solution
e.g. ATMs or air-craft. Government compliance design considerations.
is stringent and only a few players can effectively Customers are highly dynamic and are often
compete. In mature markets, utility infrastructure very demanding. New customers are added regu-
is the order of the day. A different level of enter- larly to systems and products are quickly shipped
prise architecture is required to operate in all of in response to demand. Rapid and frequent updates
these markets and to supporting evolving mobile are required to ensure currency of information to
business. service customers. Mobile business solutions that
Business dynamics – Even within a market support customers often require sustained and
and its sector and segment, business needs are frequent network connectivity with user authen-
dynamic and are not uniform. Business is driven by tication and transaction validation.
market forces, government legislation, customer Standing orders and long term contracts are
demand and costs (Figure 2) and although a lot of often in place with suppliers and much of the
commonality of function exists across business, supply process is automated. The ability to place
different types of business and different areas of a routine order from a mobile device and to track
business have differing needs. delivery requires connectivity to the network but
Markets are often highly dynamic and where frequent updates are not required.
business is heavily impacted by market trends, Business maturity – Within an organisation,
the need for dynamic real time information and different areas of business are at different levels
fast updates prevail. Mobile business offerings in their respective markets, in their mobile busi-
that support dynamic markets are often high ness enablement and in their ability to apply ICT
volume in nature with a strong focus on support- to business. This diversity is a powerful tool for
ing real time updates. business growth, enterprise architecture align-
Legislative changes are often slow but regularly ment and in application development because it
have a significant impact, e.g. Sarbanes Oxley Act provides an upgrade path for application sophisti-
(Bowersox et al. 2007). Less dynamic and more cation, business maturity and mobile enablement.
considered solutions with extensive audit capa- By looking at the next level of performance up
bilities are required to support legislative needs. from current operations, the goal of achieving
The ability to record mobile business transactions that level of performance and operation can be
set and achieved without the need to reinvent the
wheel. Successive levels of performance can be
Figure 2. Business factors impacting enterprise progressively realised (Figure 3).
architecture and solution design in mobile business Utility infrastructure – With ICT being an
assembly line for knowledge workers and its
critical role within business and mobile business,
ICT plays the role of a utility infrastructure with
the following principles included within its design:
• Redundancy: Surplus capacity is includ-

ed and protected and available to readily
scale.
• Fail-over: Include self-initiation and self-
configuration should fail-over occur.
808
Figure 3. Progressive approach to business ma-

UsEr INtErActION AND rEAL
turity and application sophistication
tIME DEcIsION MAKING tO ALIGN
Ict FOr MObILE bUsINEss
With a strong business focus inherent to the design

and application of ICT and a recognition of the
drivers impacting design, a methodology for de-
livering a scalable Service Orientated Architecture
(Soley et al. 2000) that meets current and future
mobile business needs can be established Figure 4.
The process starts with a common interface
that crosses platforms and devices and which
incorporates work-flow to provide context. Such
an interface can be readily designed and applied
• Load bearing capacity: Capability to bear by users at all levels of business. Using the vi-
load throughout all parts of the solution. sual elements of the interface to drive requirements
• Multiple layers of safeguard: Assumes and process definition, non-technical resources
failure will occur, single points of failure can conceptualise the required business function-
are avoided and are not aligned. ality and with clarity of vision comes a well defined
• Simple: Solutions are kept simple and are scope and a clear expectation.
highly standardised and modularised. The user interface of Apple’s I-phone marks an
evolution in the type of interface that is required for
The following design considerations are also supporting mobile business and real time decision
catered for within utility infrastructure and will making on any device anywhere any time. The
be required to support mobile business: interface is clean, simple and uses self-explanatory
icons that when invoked, provide the required
• Accommodates change: Change is the functionality. The extension of the Apple I-phone
norm and are designed to accommodate approach is to build in the required work-flow to
through automatic configuration.
• Achieve scalability: If it can not be auto-
mated, it is not scalable. Figure 4. ICT to support mobile business solutions
• Best of breed: Best of breed is brought to-
gether to provide an assembly line for the
processing of jobs.
• Form an emergent behaviour:
Standardised components do what they
do best and the resulting emergent behav-
iour delivers an industrial strength utility
solution.
All of these principles impact upon the enter-

prise architecture, the design of applications and
the ability to provide mobile business services.
809
conduct business into the interface. Rather than the appropriate activity object that delivers the
having icons launching software applications required business functionality, e.g. those for a
(the desktop) or control elements (adjust volume sales process (Figure 5A). As its name implies and
on an I-phone), they launch objects of business to be discussed subsequently, an activity object
functionality, e.g. pay an account. is an object of business functionality that when
An iconic interface has other advantages in- invoked, implements the task required off it. An
cluding the use of icons is intuitive and spans lan- activity object contains images, data, processing
guages; the interface can be readily customised to rules and business logic necessary to complete
reflect branding and personal preferences; and the the purpose it has been created for.
interface presents a common environment across Invoking one activity object would implement
different devices and operating systems. One other the other activity objects required for the process.
advantage that becomes more significant to mobile Using the sales process example, objects for
business is that an end user defined and process prospecting, product details, account details,
driven iconic interface would also eliminate the contact details, events management and financials
need for specific applications because it is about could all be included. Activating the account
presenting the required elements of functional- object (Figure 5B) would initiate a series of
ity only irrespective of where it resides within a other objects, e.g. credit management, service
software application suite. management, account update, company and con-
An interface consisting of a series of icons that tact details, account creation and update, and
invoke business functionality can be created in guarantees and warranties. An object hierarchy
real time to reflect specific needs i.e. the interface exists and this extends from the interface through
automatically refreshes to reflect the changing the hierarchy into the event specific details and
activities being undertaken by a user. The interface finally, the underlying code.
would also include the supporting elements to This activity object approach allows common
complete an activity, i.e. seamless interface into objects of functionality to be established, e.g.
searching and messaging, whilst connecting to banking. A hierarchy of activity objects exists to
other tasks without the need for cumbersome menu reflect standardised processes and these activ-
driven navigation. The interface embeds business ity objects are drawn together and presented as
logic and supports standardised processes. required. These activity objects can be included
Whether the icon is on a desktop PC or a in multiple processes and accessed from within
mobile device, clicking an icon would invoke and across enterprises on any device.
Figure 5. Definition of activity objects: A) Sales object, B) object hierarchy, C) object structure
810
Any activity object (Figure 5C) is made of • Activity object databases: Database for
the following: the configurable rules, images, etc. for the
activity objects that flow through into the
• Content: The content that an activity ob- interface.
ject needs to use and process. Content shall • Exceptions databases: Details on how to
be for both the activity objects themselves handle exceptions for a given activity ob-
and their functionality. ject in a given circumstance.
• Images: The images that an activity object • Messaging databases: Handling the ren-
needs to use and process. Images are for dering of information to multiple devices
both the activity object itself and for its irrespective of location and time. This in-
functionality. cludes device configuration and manage-
• Business logic: This is the actual business ment information.
logic an activity object uses for completion
of its processing task. This simple approach of defining activity
• Processing rules: These are the rules that objects from the user interface into the enterprise
define how an activity object operates, i.e. architecture has many advantages including:
what information, images and business
logic an activity object needs to process. • Solutions are driven by the end user as the
• Work-flow: This is the work-flow in which customer interacts – gain their buy in and
an activity object resides, i.e. the context support.
of the activity object. This allows users to • The required information and businesses
work their way through a process to the re- processes are readily determined.
quired outcome. • The solutions required to support operation
• Presentation rules: The rules that go to are defined.
create or make an activity object, i.e. what • The required enterprise architecture and
information, images and business logic an applications are determined.
activity object needs to function and how • Simply rendering activity objects to an end
the activity object functions. device with all of the required elements
ready for implementation has many ad-
Supporting the activity objets is a series of vantages to the implementation of mobile
virtual contextual information bases that allow business.
the activity objects to function.
The use of activity objects has the following
• Information databases: Information that impacts:
provides the required knowledge within an
activity object. • Seamless iconic interface: The interface
• Information relationship databases: is a series of icons managed within one
Manage informational relationships that coherent framework which invoke func-
transforms information to knowledge. tionality. The interface encapsulates the
• Rule processing databases: Providing required work-flow which integrates and
the rules for work-flow, business logic and refreshes as required. Such an interface
processing of the information, i.e. a recipe works on any device (desktop or mobile)
compiler. providing ready business functionality, in-
tuitive use and ease of training.
811
• Business driven functionality: The activ- • Resolution of information: An iconic

ity objects and their hierarchy can be read- interface traps work-flow, processes are
ily defined and reflect business operations. defined and the steps required to deliver
The activity objects drive outcomes and outcomes are also clearly specified. By
can be integrated to support many business definition the information required at each
operations. A single set of activity objects step is also determined. This approach
can be determined and used on both mobile defines the information required and the
devices and the desktop. context necessary for its application, i.e.
• Definition of work-flow: Business pro- it resolves the information management
cess and the information required at each problem.
step are defined as the customer interacts • Information sharing: Activity objects are
to realise the outcome. Contrast this with ideal for sharing information between oth-
the current practice of designing processes er objects or for interfacing with other ap-
around feature driven software. This func- plications and mobile devices because they
tional approach is well suited to mobile contain the required information, business
devices because of their small screen size. rule and processing rules necessary to
• Clear requirements: Many development function. In addition, activity objects can
projects suffer because requirements are address the error handling and exception
not clearly defined. By defining the re- handling. These self contained objects can
quired processes and objects necessary to be readily used and shared between mobile
deliver outcomes, requirements are clearly devices.
defined. In contrast with many ICT proj- • Resolution of the ICT assembly line:
ects, mobile business services can now be Whether it is insurance premiums, foreign
brought to market on time and to budget. exchange trades or answering customer
• Application development: Application queries for product, business is about rou-
development changes from feature driven tine transaction processing and having the
and a series of specialised applications to a right information presented at the right
series of activity objects that are linked to- time in the right way, i.e. ICT is the assem-
gether within a common framework as re- bly line for knowledge workers. The activ-
quired. Developers no longer code applica- ity object approach defines the knowledge
tions, they code object functionality which worker assembly line across the desktop
are brought together to deliver outcomes as and mobile devices.
required by the user. • Definition of Service Orientated
• Standardised activity objects: Elements Architecture: From a definition of the
of business functionality are common knowledge worker assembly line comes
across many areas of business operation a resolution of the ICT required to sup-
and types of business. When using activ- port the assembly line, both hardware and
ity objects, banking functionality is no software. How ICT is to be deployed and
longer tied to accounting and financial ap- operated is also determined, i.e. a Service
plications; elements of banking functional- Orientated Architecture is defined that is
ity become available as needed. This use driven as the customer interacts.
of activity objects allows for standardised
elements to be developed and delivered
across business, markets and devices.
812
MObILIty, ActIVIty ObJEcts AND tected around the FedEx model, where the
MODEL DrIVEN ArcHItEctUrE quality of hand-off, the message and the
delivery and storage of the message are all
The Model Driven Architecture approach and the separated and utility infrastructure used.
use of activity objects comes to the fore when it A business model driven architecture ap-
comes to mobile computing and the demand for proach services the needs of a unified mes-
mobile services (Lee et al. 2004). Many incum- saging environment.
bent solutions with their inefficiency in accessing • Error and exception handling: Effective
information and an absence of the unification of and efficient transaction processing is all
information with work-flow, become almost inef- about how the errors and exceptions are
fective in servicing mobile devices. Cumbersome managed. Whilst addressing the quality of
interfaces that do not deliver outcomes can not be hand-off issues is part of the solution, hav-
used on mobile devices because of their screen ing dedicated processes to mange errors
size. Unlike the desktop PC, the storing of data and exceptions is also required. Specific
locally on an end device is no longer an option. objects to handle errors and exceptions can
Complex business applications rich in features be defined and called in sequence until the
that people rarely use are not easily transferred issues are resolved.
to mobile devices (Paavilainen 2001). • Hand-off: Integration of information, in-
Simplicity, with clear interfaces that deliver tegration between systems, integration of
outcomes is what works on the mobile device. The context and work-flow all rely on an effec-
need for effective and efficient mobile interfaces tive quality of hand-off between different
shall in turn impact upon the incumbents in the elements. Self contained objects with all of
desktop environment and create both a common the required elements provide an effective
interface across devices and an optimised sup- solution for ensuring a quality of hand-off.
porting infrastructure. • Security: The topic of security is complex
Other areas where the application of mobility and is the subject of much concern and ex-
leverages model driven architecture include: tensive discussion (Nand 2006). An object
approach helps to facilitate the security be-
• Information Relationships: To be effec- cause the necessary role-based access and
tive, information needs to be delivered in functionality is included within the object.
context for decision making. Context is Additional security features and capabili-
given by managing the relationships be- ties can be included as elements within the
tween information through a metadata objects.
framework that integrates with work-flow.
• Work-flow: Activity objects include their Mobility and the demand for mobile services
own work-flow and business logic but their is what will drive the development of object ori-
strength comes in the combining of stan- entated solutions that shall manifest on both the
dard activity objects together to provide a desktop and in the enterprise. Delivering objects
consolidated work-flow. of functionality to a mobile device, as required
• Messaging environment: Mobility re- with all of the necessary information, business
quires a unified messaging environment logic and work-flow has several advantages:
combining voice, data, text, images and
video. To guarantee service delivery the • Across platform – A common interface
messaging environment needs to be archi- with standard user driven functionality can
813
be delivered across devices and platforms. dardised objects can be used defined and
This provides the end user with what they applied across many processes and areas
need irrespective of which device they use of business, application development be-
and location in which they operate, i.e. “my comes much more responsive to dynamic
desk where I want it and how I want it”. business needs.
• Consolidated information-bases – With
the creation of virtual consolidated infor-
mation-bases, information can be single FUtUrE DIrEctIONs
sourced. Information is no longer trapped
on the end device and all of the archiving Real time decision making and activity objects
and backup are conducted at the server. have many uses within both mobility and in the
• Smart end devices – The benefits of dis- wider enterprise business application of ICT. One
tributed computing power are realised of the evolving needs is in the application of gam-
without all of the complexities inherent to ing solutions to business training, education and
the desktop environment. The end device simulation. Gaming is one of the fastest growing
is a smart device. The benefits of lower services of mobile computing and whilst gaming
cost, ease of use and better asset manage- offers many market opportunities in its own right,
ment are transparent. it is the wider business use from mobile devices
• Information management simplification that are to be realised. Flight simulators and their
– The inefficiency of the desktop environ- use has been widely applied to train and skill staff
ment and the need for advanced informa- and to provide experience in handling difficult
tion management skills to manage ver- situations. The next level is to assist in business
sions, locations, formats and applications decision making through simulation and what-if
is not sustainable on the mobile device. scenarios. Strong visual presentation and a rich
Information presented at the right time in immersive and interactive experience to provide
the right way to any device anywhere any- scenarios and identify issues and outcomes is a
time is required. The optimisation neces- major area where model driven architecture and
sary for mobility shall also drive changes activity objects shall be used. The simulation and
in the desktop environment. modelling results shall be accessible from mobile
• Defines ICT required – A business focused devices as well as the desktop.
model driven architecture approach helps The provision of Software as a Service (SaaS)
determine the business need, is driven is in its infancy and the mobile opportunities
by the business process, the information for SaaS are still to be realised. Much of SaaS
required is identified and integrated to is still very application specific and there have
work-flow. At this point the ICT solution been concerns about the quality and applicability
necessary to support a business and how it of code supplied with SaaS. The activity object
should be deployed is transparent. approach is the natural extension of SaaS; with
• Responsive to business need – Business only those activity objects required being rendered
is very dynamic and one of the existing as a service.
challenges is trying to define requirements The development of mobile computing will
against a rapidly changing business envi- see a proliferation of solutions and services
ronment. An object approach allows those and an incredible diversity of offerings. As the
objects required to be combined in real market matures, standards shall come into play
time to meet business needs. Since stan- and consolidation shall be seen. Standardisation
814
lower costs and guarantees service delivery and Lee V., Schneider H., & Schell R. (2004). Mobile
in turn this creates market dominance. Being the Applications: Architecture, Design, and Develop-
de-facto market standard is how a business gains ment. Hewlett-Packard Development Company
market dominance. The software of the future is L.P., publishing by Pearson Education as Prentice
not feature rich applications but tightly integrated Hall Professional Technical Reference.
activity objects drawn together as needed to deliver
McGovern, J., Ambler, S. W., Stevens, M. E.,
outcomes. Being the de-facto standard for activity
Linn, J., Sharan, V., & Jo, E. K. (2004). Foreword
objects shall lead to standardisation of the desktop,
by O. Sims. In A Practical Guide to Enterprise
at the enterprise, at the mobile device level and
Architecture. Pearson Education, Inc.
in turn, at the marketplace level (the emergence
of marketplace computing). Nand, S. (2006). Developing a Theory of Portable
Public Key Infrastructure (PORTABLEPKI) for
Mobile Business Security. In Unhelkar B. (Ed.),
cONcLUsION Chapter 27 of book: Handbook of Research in
Mobile Business: Technical, Methodological
Mobile business and the demand for real time and Social Perspectives., Hershey, PA, USA:
decision making is a powerful driver for aligning IGI Global.
ICT and applications to deliver the ICT infrastruc-
Paavilainen J. (2001). Mobile business strategies:
ture necessary for mobile business. The use of a
understanding the technologies and opportunities.
standard iconic interface and activity objects that
Wireless Press; Addison-Wesley in partnership
invoke business functionality provides an effective
with IT Press.
solution for delivering mobile business services,
whilst resolving the information management and Sherringham, K. (2005). Cookbook for Market
work-flow issue and aligning enterprise architec- Dominance and Shareholder Value: Standardis-
ture and application development. ing the Roles of Knowledge Workers. London:
Athena Press. (p. 90).
Sherringham, K. (2008, Apirl/May). Catching the
rEFErENcEs
Mobility Wave Information Age. (p. 5).
Bowersox, D. J., Closs, D. J., & Cooper, M. B. Sherringham, K., & Unhelkar, B. (2008a). Ele-
(2007). Supply Chain Logistics Management, 2nd ments for the Mobile Enablement of Business. In
edition. McGraw-Hill Companies, Inc.: Irwin (Unhelkar et al. 2008) Handbook of Research in
Cummins, F. A. (2002). Enterprise Integration: Mobile Business: Technical, Methodological and
An Architecture for Enterprise Application and Social Perspectives. (pp. xxx –yyy).
Systems Integration. Canada: Willey Computing Sherringham, K., & Unhelkar, B. (2008b). Real
Publishing, John Wiley & Sons, Inc. Time Decision Making and Mobile Technologies.
Lan, Y., & Unhelkar, B. (2005). Global Enterprise In (Unhelkar et al. 2008) Handbook of Research
Transitions: managing the process. Hershey, PA: in Mobile Business: Technical, Methodological
IGI Global. and Social Perspectives. (pp. xxx –yyy).
Soley, R. & OMG Staff Strategy Group (2000).
Model Driven Architecture. Object Management
Group. White Paper.
815
KEy tErMs AND DEFINItIONs Marketplace Computing: Standardised com-

puting (tightly integrated hardware and software)
Activity Objects: A series of objects invoked operating at the marketplace level to allow busi-
from a standard iconic interface that provide businesses to interact effectively in a marketplace.
ness functionality because they contain the neces- Business currently standardises at the enterprise
sary content, images, business logic, processing level but to operate effectively, standardisation
rules, work-flow and presentation rules. shall be at the marketplace level.
FedEx Model: A model for the operation of Real Time Decision Making: The provision
a unified messaging environment based on the of information in context and integrated with
proven principles to move messages (parcels) work-flow in real time to any device anywhere
around the world by leading logistics companies. anytime is needed so that decisions can be made.
Information Bases: The next generation of
databases but instead of storing data, information
is stored in context. ENDNOtE
Information Relationships: The associa-
tions between elements of information to provide a
The introduction of the automatic telling
context and convert information to knowledge. machine (ATM) led many banks to redevelop
Knowledge Worker Assembly Line: Knowl- their enterprise architectures and those that
edge workers take information and value-add to could not respond effectively were at a com-
it to provide services. ICT needs to provide the petitive disadvantage and in some selling
right information at the right time in the right way off their retail operations.
for knowledge workers to effectively operate, i.e.
ICT is the assembly line for knowledge workers.
816
817
Chapter 3.17
Mobile Technologies
Extending ERP Systems
Dirk Werth
Institute for Information Systems at German Research Centre for Artificial Intelligence, Germany
Paul Makuch
Institute for Information Systems at German Research Centre for Artificial Intelligence, Germany
Nowadays the majority of enterprises use En- Enterprise Resource Planning (ERP) systems have
terprise Resource Planning (ERP) software to become the IT backbone of most enterprises. Sev-
improve their business processes. Simultaneously, eral publications, articles and surveys mention that
mobile technologies which can be used within ERP almost 70-80% of Fortune-1000 enterprises use
have gained further importance. This is because ERP systems to improve their business processes.
ERP, together with mobile technologies, offers ERP systems have changed the way enterprises
a wide spectrum of synergies and both have a conduct their business as many functionalities,
significant impact on enterprise efficiency. The which only a few years ago had be done manually,
improvement possibilities in ERP due to mobility now are automatically provided by the system.
range from sales activities, over logistic processes, Similar to the Internet, technologies have
up to effects on the human resource management. largely grown in the last years. E.g. mobile phones
have become a standard communication device in
most countries. In the near future mobility and flex-
DOI: 10.4018/978-1-60566-156-8.ch041
Mobile Technologies Extending ERP Systems
ibility will be a key issue that will enable organiza- • Branch neutrality: ERP software is nor-
tions to withstand competition in an environment mally not aligned to a specific branch
characterized by increasing cost pressures. Using • Operating efficiency: the special emphasis
mobile technologies for commercial purposes is placed on efficiency, not on technology
is one option that can certainly make business • Modularity: There are enclosed areas of
processes more efficient. This chapter discusses activity within the software, called modules
mobile business has opened new opportunities in • Integration: All business activities as an
ERP systems. Furthermore, we also discusses vari- aggregate are continuously supported
ous other business processes that are influenced by • Standard software: ERP systems are not
mobile technologies (e.g. buying train tickets via designed for individual purpose. In fact
the mobile phone). However, it should be noted they are sold on an anonymous market,
that the influence of mobile technologies is not but of course they can be customized, i.e.
limited to consumer interactions. Also, established adapted to fit customer needs.
applications can be enriched by mobile technology
resulting in new or improved functionalities. This ERP systems differ from each other in their
chapter explores the impact of mobile technology complexity, range of functions and procurement
on ERP systems and demonstrate some use cases costs. First it depends on the branch the enterprise
where such technology can significantly improve operates using ERP technology. With an ascending
ERP functionality. number of suppliers or products, more modern
warehouse systems or new distributions chan-
nels, the complexity of such a system increases.
ErP systEMs Second, the size of the enterprise including their
whole network matters. Small and medium sized
Enterprise Resource Planning Software offers a businesses don’t need the same range of functions
spectrum of activities which support enterprises like a worldwide operating multinational company.
to organize important business processes by pro- This certainly has an influence on the price of the
viding multimodular software applications. ERP software solutions, e.g. small sized businesses can
systems have evolved in the middle of the 1990s install standard versions, whereas concerns need
from manufacturing resource planning (MRPII) specially developed additional modules. Third,
systems. Such systems aim to plan and steer the the number of users working with the ERP system
output generation within an enterprise. They com- plays an important role. The more accounts work
prise of all logistic activities, from the purchase simultaneously, the more powerful hardware is
planning and execution through the manufacturing needed to guarantee an unproblematic process.
planning, steering and supervision to the sales and Last, the technological base used to realize an ERP
after-sales activities. MRPII systems mainly cover system, especially the database and the program-
the logistical view of the enterprise. Extending ming language, is a key factor determining the
MRPII systems by human resource management complexity and the range of functions.
and by financial management has resulted in ERP In the future, ERP systems will be increas-
systems, that aim to cover all activities and busi- ingly standardized. Therefore the flexibility and
ness processes within an enterprise. Nowadays mobility of user interfaces will play a more and
every business transaction can be monitored, more important role in order to generate additional
analyzed and evaluated. advantages.
The performance properties of ERP systems
are:
818
MObILE tEcHNOLOGy addition, the costs for mobile devices like PDAs
are continuously decreasing and far away from
Mobile technology is newer than ERP systems. comparable notebook prices. In terms of realizing
The first technological achievements covered the saving potentials the localizability aspect probably
mobile speech transmission using analog mobile plays the most important role. For example RFID
phones. At the end of the 1990s, the technology tags allow enterprises to improve their logistic
has broadened by two streams: On the one side activities (see point “Logistic improvements”).
the digitalization of mobile technology, on the In the next sections we will discuss potential
other side the inclusion of text and data services. advantages by presenting a use case to each im-
Besides, the mostly used service is the short mas- provement field.
sage service (SMS), originally developed for usage
in the global systems for mobile communications
(GSM). Today, several kinds of mobile devices sALEs IMPrOVEMENt,
are available on the market. It ranges from simple cOst rEDUctION
GSM mobile phones, over ultramodern personal
digital assistants (PDA) connected via the univer- Sales can be increased by using mobile technology
sal mobile telecommunications standard (UMTS), in order to extend the functionality of ERP systems.
up to radio frequency identification (RFID) tags In order to get a better idea of the improvement
which can simplify warehouse processes. potentials we take a look at the example of the
By using portable terminals and mobile data “traveling salesman”. His main fields of activity
transfer technology, users establish a connec- are customer acquisition, providing information
tion to wireless firm-owned network services. to customer, product sales and their ordering and
They are locally and temporally independent after-sale activities. These tasks are normally sup-
and always available. As a result they are able to ported by the ERP system, e.g. new customers
make transactions from almost every place on the have to be set up with their individual customer
earth. Additionally, portable terminals are easier number. All customer orders, including their
to operate and have a shorter boot time than lo- content, volume and value are registered and can
cally installed user interfaces due to the fact that be tracked up on the basis of a voucher. With a
decentralized devices normally only include the mobile device this information can be entered
essential range of functions. As far as security or accessed wherever the salesman, respectively
is concerned the software or the hardware nor- the customer is located. Furthermore, constantly
mally includes personal identification processes, updated price lists can be presented and products
e.g. a subscriber identity module (SIM) card or can be ordered online, including such special
password protection. This is necessary to ensure services like delivery time determination. This
that third parties are not able to enter the network allows both to increase customer satisfaction and
and see or manipulate enterprise information. As business efficiency. On the one side the enterprise
each network user has its own mobile device and is able to respond fluidly to changing conditions
the corresponding account, personalization pos- in customer demand as the salesman is always
sibilities are nearly unlimited, i.e. it is possible to directly linked to the development and purchase
define different views and accesses to the central department via mobile technology. Thus produc-
database. Personalization allows users to work tion can be flexibly adapted and it is not necessary
more effective because everyone is allowed to to estimate potential sales figures, i.e. the risk
individually determine its preferred properties. of wasted production capabilities as a result of
This also leads to more cost-effectiveness. In non-saleable goods is significantly reduced. On
819
the other side the complexity of the ERP system • After sales activities: Even after the sell-
is decoupled. Only the functions needed for ing process, the RFID tags remain on the
salesman’s activities are supported by the mobile products and can be used for automatic re-
device, unnecessary and complex features of the plenishment, reclamation or exchange pro-
system are removed, respectively not available. cedures and after-sales disposal activities.
By embedding automatic synchronize and update
functions the employee spends a minimum of its Due to relatively high costs of the needed
working time on administrating the ERP system. transponder technology RFID was only used in
big business logistics, e.g. in container handling
facilities. As costs of the obligatory hardware are
LOGIstIcs IMPrOVEMENt decreasing the usage of the technology becomes
more and more efficient for other purposes. Point-
With regard to cost efficiency ERP supported of-Sales will use RFID in order to accelerate sale
enterprise warehouse systems gain in importance. activities or to reduce consignment and personnel
In the near future the new generation of the radio costs. E.g. a supermarket can save costs by provid-
frequency identification (RFID) standard will help ing fully automatic cash desks. It is not necessary
to save expenses. The RFID concept is based on to scan each product a consumer wants to buy,
contactless data transmission by electromagnetic not even a visual contact has to be established,
alternating fields [Hertel J., Zentes J., Schramm- as several RFID data carriers can be ascertained
Klein, H. (2005)]. Special RFID tags serve as within one single read operation. All goods within
data volumes and allow the reading, processing the customer’s shopping carriage are identified by
and changing of chip contained information. The driving through a scanner unit.
main application possibilities are quiet varied
[ECR-D-A-CH (2003); Füßler (2004)]:
HUMAN rEsOUrcE IMPrOVEMENt
• Production: After production goods are
individually equipped with an RDIF tag In this section we discuss improvement potentials
which allows identifying their position at for human resources (HR) by looking at the ac-
every step of the supply chain. counting of travel expenses. ERP systems also
• Stocks monitoring: RFID technology support automatic note of expenses. If an employee
allows to trace the receipts of goods, the comes back from a business travel and wants its
warehouse process itself, and outgoing travelling expenses to be reimbursed, the system
goods. As periodically recurring inventory only needs the payment vouchers, the employee
processes always retain a lot of employees, number and the release signal for clearing the
mobile devices in correspondence with the payment. With mobile technology this operation
RFID technology can help to make count- can be accelerated once again. Contemporary to
ing operations more efficient, i.e. easier the prepayment of the employee, the vouchers can
and faster. All stocks can be counted by be digitally submitted to the office, the system
scanning the RFID tags with a mobile validates the sums and pays the bill, e.g. paying bus
scanner unit, e.g. a modern PDA extended and train ticket via mobile phone. Since April 2007
by a radio frequency receiver. Especially there is a pilot project with twelve participating
with regard to homogeneous stocks, docu- German cities, providing such a service to their
mentation operations can be accelerated. citizens. The system is nationally standardized
and was developed by member firms and groups
820
of the Association of German Transport Compa- rEFErENcEs

nies (VDV), Siemens IT Solutions and Services,
DVB LogPay and the Frauenhofer-Institut IVI ECR D-A-CH. (2003). RFID – Optimierung der
Dresden. After completing a one-time registra- Value Chain. Köln.
tion and selecting a preferred payment method, Füßler, A. (2004). Auswirkungen der RFID-Tech-
users receive a text message containing a Java nologie auf die Gestaltung der Versorgungskette.
application element that is used for ordering the In J. Zentes, H. Biesiada, H. Schramm-Klein
tickets. The mobile phone screen allows the user (Hrsg.), Performance-Leadership im Handel.
to enter the type and value of the ticket, whereby Frankfurt a.M., (pp. 137-155).
single tickets and day passes are available. After
the payment process the mobile phone owner Hertel, J., Zentes, J., & Schramm-Klein, H.
receives an on-screen confirmation serving as a (2005). Supply Chain Management und Waren-
receipt, e.g. for ticket inspection [Soft32 (2007)]. wirtschaftssysteme im Handel. Heidelberg:
This could be extended by directly sending a Springer (pp.207-210).
voucher to the company’s ERP system. Obvi-
Soft32 (2007). http://news.soft32.com/bus-
ously this development reduces personnel cost
and-train-tickets-via-mobile-phone-in-munster-
and saves administration time because no paper
germany_5232.html.
documentation is needed.
cONcLUsION KEy tErMs AND DEFINItIONs
This chapter explored new ways of enriching Business Process: A target-oriented, logical
standardized ERP systems with modern mobile sequence of activities which can be performed
technology. Mobile components or devices enable by multiple collaborating organisational units by
enterprises to make their business processes more using information and communication technolo-
efficient. Via wireless networks employees are gies. This system of functions makes a substantial
permanently linked to the ERP system, are able to contribution to the generation of added value.
work online and can use the saved time for more Enterprise Resource Planning (ERP)
productive activities. Systems: Integrated packages of standardized
The cases we presented above give a general software applications supporting the resource
overview of possible new business opportunities planning of an enterprise. Financial, logistical and
if ERP systems are used in connection with mobile human resource related business processes can be
devices. They will have a significant influence improved by using an ERP system.
on future activities within an enterprise. With Mobile Business: Describes the initiation and
increasing technological performance of mobile the entire support, execution and maintenance of
devices the structure of an ERP system will be business transaction between business partners by
changing from a centralized main system into a the use of wireless electronic network communi-
network consisting of independently operating and cation technology and mobile devices.
interlinked mobile devices. The discussion in this Mobile Business Processes: Integrate mobile
chapter can be further enhanced by subjecting it solutions into classic business processes. Mobile
to research validation that is currently outside the work leads to new collaborative opportunities,
scope of this chapter. improves the enterprise workflow and enables the
transaction of digital business processes.
821
Mobile ERP: Solutions extend traditional ERP Sensory ERP: A concept for next generation
systems by location-independently collecting and ERP systems. It enables the ERP system to auto-
exchanging data via mobile devices and wireless matically acquire data and supervise enterprise
transfer mechanisms. Standardized interfaces states and events by using sensors (e.g. RFID
allow a direct and steady connection to the ERP tags and gates, GPS tracker, etc.). Interfacing
hardware and lead to more flexible and efficient between the physical world and the ERP system
business processes within an enterprise. is no longer performed through human workers,
Radio Frequency Identification (RFID) but this data is collected by sensors that directly
System: Allows contactless data transmission by assess the physical states and that are part of the
electromagnetic alternating fields and is often used real world itself. By this, the error rate signifi-
for automatic identification and data acquisition. cantly decreases and business processes become
more efficient.
822
823
Chapter 3.18
Convergence in Mobile
Internet with Service
Oriented Architecture and
Its Value to Business
Marco Garito
Digital Business, Italy
AbstrAct Architecture (SOA): the first two technologies are

dealt with in parallel to provide an overall view of
The word “convergence” refers to the combina- advantages and disadvantages, while SOA will be
tion of fixed and mobile communication, a situ- part of a distinct discussion and analysis. Eventu-
ation where a private or business user can take ally, some practical examples of these discussed
advantage of being constantly connected and be technologies are provided.
able to retrieve applications and data by swapping
device, with the limitations that a mobile device
may have such as smaller screen and keyboard, INtrODUctION
reduced storage capability, and limited power
provided by batteries. Convergence can also in- RFID and Bar code are two emerging mobile
clude imagining how mobile technology can be technologies able to provide competitive strategic
a component of everyday items and how data, advantage to business when properly deployed
applications, and services can be delivered via the and implemented. As extension of current fixed
network infrastructure. This chapter aims to cover network infrastructure, the coordination with
RFID technology, Bar code and Service-Oriented existing business processes, department and
organizational structure are an essential part of
DOI: 10.4018/978-1-60566-156-8.ch054 rewarding implementation. The development of
Convergence in Mobile Internet with Service Oriented Architecture and Its Value to Business
SOA environment can further enhance the capabil- RFID in different situations, according to their
ity and possible outcome of RFID and Bar code. technical features
The chapter outlines advantages and disadvantages The advantages of RFID can be classified as
for both of them, providing examples of how they follows (Lahiri, 2006; Garfinkel, Rosenberg 2006;
co-exist and how they can create value. Ascential 2005):
rFID and bar code • RFID tag can be read without any physical
contact between the tag and the reader
Radio frequency identification (RFID) is one of • The data of RFID can be rewritten sev-
the most interesting technologies today: its use eral times with no diminishing quality or
impacts a large number of protagonists in private integrity
and business environment but it also raises simple • A line of sight is not required for an RFID
and dramatic issues in legal, social and political reader to read a tag
affairs. RFID have histories back to 1930 and 1940, • The range can vary from few centimetres
when the British Army, during WW 2 pioneered to some metres
RFID to identify their own aircraft returning home • Storage capability of a tag is unlimited
from bombing Europe. Early radar systems could • A reader can read different tags within its
spot an incoming airplane but not its type (Lahiri, reach for a limited time
2006; Garfinkel, Rosenberg 2006; Ascential 2005) • A tag can be structured to perform unlim-
RFID uses radio waves to detect physical items ited duties
(both living and inanimate) and therefore the range • The data quality is 100% guaranteed
of identifiable object includes everything and
everywhere: RFID is an example of automatic RFID has its limitation that can be summarized
identification technology through which an item as it follows
is automatically detected and classified. Bar code,
biometric, voice identification and optical char- • RFID do not work well or not work at all
acter recognition systems are other example of with RF-opaque items or RF-absorbent
automatic identification. items
The RFID environment consists of a set of • Surrounding conditions may affect
mandatory and optional components: Mandatory performance
parts are: tag, reader, antenna controller. Sensor, • There is a limit for how many tags can be
actuator, host and software system and com- read within a time slot
munication infrastructure are the optional parts. • Hardware set up may limit performance
The table below describes the types and usage of • The technology is still immature
Figure 1. Type and usage of RFID
824
Bar code is a scheme representing textual of data capacity, up to 3750 number of characters;
information; the symbols are generally vertical Three dimensional which is a linear bar code
lines, spaces, squares and dots: the bar code is integrated in a surface (Lahiri, 2006; Garfinkel,
probably the newest technology as the first patent Rosenberg 2006; Ascential 2005)
was issued in 1949 and the first application of bar Bar codes are read by scanners which flash a
code technology was a rail car tracking system, light through the bar code area: during this process,
implemented in 1960 (Lahiri, 2006; Garfinkel, the scanner measures the intensity of the light
Rosenberg 2006; Ascential 2005). The method reflected by the white and dark area of the bar
encoding letters and numbers using these elements code: the dark area absorbs light, the white area
is defined as symbology, which has the following reflect back the light. The light pattern is captured
characteristics: and translated by a photodiode or photocell into
an electric signal which is again converted into
• Symbology with better encoding technique digital data, represented as ASCII characters.
leads to error free and efficient encoding: This same data has been incorporated within the
• Better character density can represent more bar code at the origin (Lahiri, 2006; Garfinkel,
information per unit physical area Rosenberg 2006; Ascential 2005).
• Better error checking capability enables The advantages are (Lahiri, 2006; Garfinkel,
data reading even in those cases where the Rosenberg 2006; Ascential 2005):
some components are damaged or missing
• Rapid and accurate data collection;
It is possible to have three different categories • Increased operation efficiency;
of symbols: linear, with vertical lines with differ- • Reduced operation costs;
ent spaces with the white spaces separating two • The shortcomings are:
adjacent lines, with an overall maximum number of • Bar code can be easily damaged
50 characters; two dimensional with high storage
Figure 2. Comparison table between RFID and bar code
825
• Reader efficiency can be affected by envi- rFID and bar code Application; An
ronment conditions Example In travel Industry
• Presence of obstacles does not allow the
scanner to read the bar code RFID and Bar Codes coexist though with mobile
• Speeded items do not allow the scanner devices. In a joint initiative Finnair and Nokia
working properly have implemented a new initiative to manage
ground-based staff: work assignment are directly
The following Table 2 summarizes advan- transmitted via mobile so employees can directly
tages and disadvantages in RFID and Bar code tackle their job tasks and when they have finished,
all by listing specific and relevant characteristics, their mobile device can read the RFID tags lo-
based on the considerations made above cated at each key point locations and this data is
However both RFID and Bar code are not routed to the central management database. Again
immune of common disadvantages and the table Finnair and TDC mobile are piloting a test to use
below provides an overview mobile device to check in: the initiative is called
The situation described above enables to “Mobile bar code boarding pass”. After buying
conclude that a replacement or takeover of RFID the ticket, passengers check in by mobile, on the
technology against Bar code is unlikely to hap- internet or at the check-in desk and they receive
pen: RFID is still an immature and developing a message with a 2-dimension bar code on their
technology compared to a widely consolidated mobile phone; once at the airport, passengers
and used Bar code. It is possible to see many retrieve the message and scan it for checking
areas of improvement for both of them based on luggage, security screening and when boarding
the analysis done so far and at the moment the the aircraft: no further paper-based boarding card
logical conclusion is that RFID and Bar code can is needed (Toro, 2007)
co-exist and many of the applications for RFID
can be located out of the reach of current Bar code rFID Applications Examples
scenario where business needs require something
else or other than Bar Code (Lahiri, 2006; Garfin- Mobil introduced a “Speedpass” System in
kel, Rosenberg 2006; Ascential 2005). 1997 to fast track the payment to their petrol
stations network: the system was later extended
to convenience stores of Mobil. In 2001, the
company augmented their Speedpass services to
McDonald’s, in 2004 Mobil entered an agreement
with Stop & Shops stores to test whether or not
Figure 3. Main disadvantages for RFID and bar code
826
Mobil’s customers could buy their grocery and get a cash back and even utilize fidelity vouchers
food.: most likely, car drivers on the move do not in view of current or future purchases.
have enough time, to dedicate to shopping task This is also an extremely powerful way to
and the speed concept underneath demonstrates capture data for marketing research and analy-
how mobile technology can enhance traditional sis because while scanning the product, the
shopping experience in everyday life (Lahiri 2006, information is automatically transmitted to the
Garfinkel and Rosenberg 2006). central database which can thus calculate trends
Another example of RFID use can be retrieved and attractiveness of the same product, the other
in hospitals where three types of RFID can be products bought in combination with it and the
identified: those to track people and items around success of a promotion campaign with real time
the premises, those to safeguard use of medical information transmitted directly to the back office
equipment and those to assist medical staff in of the store chain to organize replenishment in a
their daily job. Different situations but each of timely manner. Therefore, self service check-out
them can interact providing a unique view of kiosks is not just a way to reduce the number of
hospital’s environment. RFID can provide the cashiers and reduce the space.
needed granularity to properly asses moments, The availability of complex and aggregated
people or objects, even at home, when the need data about products in retail industry, within a
is to monitor the wellbeing of elderly people or bar code also play an important role in health and
children needing special care and attention (Lahiri safety in food and many regulations around the
2006, Garfinkel, Rosenberg 2006).. world prescribe very specific and rigid rules for
Still an example in retail industry, RFID and fresh and perishable food products. Each store
tag technology play an important role in asset can retrieve on daily basis the use by date food
protection in not food products: expensive items products and pull out each item from the regular
and items that can be easily hidden by shoplift- shelf space or organize a dedicate space where
ers are protected by visible or invisible tags that these same items can be sold at a reduced price.
must be removed at the check out points: in this This happens in many stores around London area:
case, RFID and Bar code technology can coex- as the risk for the retail shop is to throw away
ists because they cover two different needs. Once food and therefore asset, it is a better off to try to
the tag has been removed (so the customer can sell as much as possible, even at a reduced price,
pass through the exit grid without no alarm be- relying on the fact that customers may be more
ing activated), the same item can be scanned for tempted to buy items when their price is lower
payment process: eventually, the recording of than usual for immediate consumption at home.
the transaction is transferred to the back office There are some though some concerns about
enabling the marketing analysis described privacy and more broadly pervasive computing
because these are emerging technologies far from
bar code Applications Examples being stable yet: a tag reader can detect and collect
information from any belongings, document or
The largest deployment of Bar code technology shop items bought elsewhere (the ringing alarm
happened in retail industry: the major retail chains at the exit gate grids many people experience is
(Wal-Mart in US, Tesco, Asda and M&S in UK an example). The main deployment of tagging and
for example, but list could continue) have recently RFID is in retailer and supply chain environment
adopted their stores with self check-out points or therefore privacy is not an issue in such a circum-
kiosk, enabling thus customer to scan their shop- stance; however when this technology is applied
ping products, pay by cash or credit/debit card, and attached to consumer good at item level.
827
It is now time to introduce the Service Oriented to respond to competitive market and rapidly de-
Architecture concept to analyze how wireless velop new solutions and at the same time change
technologies can be deployed in a dynamic and business models by gathering real time informa-
user oriented way tion through sensitive applications and services
(Sonic 2006; Cisco 2005; Plumtree 2005, Symons
service Oriented Architecture (sOA) 2005). The table 4 if properly understood and
implemented, opens the door to the semantic web,
Service oriented architecture can be defined as where content and applications are exchanged,
a method to conceive, implement and distribute captured and analyzed real time, enabling thus to
business functions, data or applications, based adopt and deliver sensitive answers to business
on geography or across enterprise, enabling the needs. RFID and Bar code do this by capturing
reconfiguration of new business processed when data from the end point and transferring it back
necessary; there are some key points which must to the source, at the opposite end, to eventually
be taken into account (Sonic 2006; Cisco 2004 and change the characteristics of the original data itself
2005; Plumtree 2005, Sprott 2004, Symons 2005): It is now necessary to move onto the next
step and possibly describe the IT and Business
• SOA is based on WWW standards architecture Today’s enterprises require a new
• The services or, more precisely, the content IT strategy, one that will improve their ability
of SOA allows business flexibility to respond to competitive pressures and market
• SOA can incorporate best practice to create demands.
design aimed to develop and enhance busi-
ness processes From sOA to service Oriented
• SOA covers existing system (Sonic 2006; Network Architecture
Cisco 2004 and 2005; Plumtree 2005,
Sprott 2004, Symons 2005) The emerging solution takes advantage of a more
flexible, adaptive, and feature-rich IT architecture:
The combination of these characteristics Service-Oriented Network Architecture. It helps
demonstrates and confirm the fit between the enterprises evolve their existing infrastructure
RFID/Bar codes described below and the real- into an Intelligent Information Network (IIN)
ization and distribution of services, applications that supports new IT strategies, including service-
or content through SOA: the difficulties and the oriented architecture (SOA), Web services, and
challenges brought in by new technology develop- virtualization and mobile. By integrating advanced
ment dictates to implement a model to facilitate capabilities enabled by intelligent networks,
the evolution of business process and supporting enterprises reduce complexity and management
technology (Sonic 2006; Cisco 2005; Plumtree costs, enhance system resiliency and flexibility,
2005, Ascential 2005). The table below provides and improve usage and efficiency of networked
a practical example about the moment in business assets. It allows enterprises to use their network as
development terms, in which mobile technologies a strategic asset that properly aligns IT resources
can be adopted. with Business priorities. The result is lower total
The key for a successful adoption and dis- cost of ownership (TCO) and increased revenue,
tribution of mobile technology based on SOA which over time enable organizations to shift an
infrastructure is an evolving and continuous increasing proportion of their IT budgets toward
alignment between Business and Technology and strategic investment and business innovation
the capability of IT to properly support Business
828
Figure 4. Adoption of mobile technology
(Sonic 2006; Cisco 2004 and 2005; Plumtree • Interactive services layer: this layer opti-
2005, Sprott 2004, Symons 2005). mizes communications between applica-
Service oriented networks architectures are tions and services in the application layer
based on a three-layer design by taking advantage of intelligent network
functions such as embedded security, iden-
• Application layer: this layer includes all tity, and quality-of-service (QoS) features.
software used by end users within the en-
terprise for business purposes (such as en- The 7 OSI (Open System Interconnection) lay-
terprise resource planning and customer ers, described in Figure 5, provide the ground for
relationship management) and software understanding where the spoken mobile technolo-
used for collaboration (for example, uni- gies can be successfully implemented (Simeneau
fied messaging and conferencing). 2005, White 2001). Since tags store information,
• Networked infrastructure layer: this layer which may include location data, not all intel-
interconnects devices at critical points in ligence need be held in corporate networks and
the network (campus, data center, network enterprise systems. Exchange of information may
edge, metropolitan-area network [MAN], be restricted to tags and readers may be processed
WAN, branch offices, and tele-worker lo- by a local server via a LAN, or aggregated and
cations) and facilitates transport of services passed on to a distribution centre.
and applications throughout the enterprise. Many organizations are starting with RFID
pilots within the confines of their own environ-
829
Figure 5. OSI 7 layers

creation of Value
the creation of value using RFID (as well as any

other mobile technology) can be summed up by
analyzing the Figure 6.
The circled areas describe the main develop-
ment steps from an internally targeted phase
(internal deployment) to a more complex extra-
company connected enterprise encompassing
many businesses: this evolutionary process implies
ment, as localized internal deployments. The
two main issues:
reason for this is simple: it restricts the scope to
a manageable limit, allowing learning to occur in
• A growing sophistication of solutions and
a controlled manner. But most organizations
intensive use of the network infrastructure
implement RFID to facilitate supply chain effi-
(vertical ax)
ciencies, so extending RFID beyond the organiza-
• An increasing density of the device which
tion is inevitable. Even so, there are some small
probably means a wider adoption of open
steps that an organization can take in venturing
source software to make easier the coding
outside their own environment (Sonic 2006;
and connection between heterogeneous de-
Cisco 2005; Plumtree 2005, Brown, Wiggers 2005,
vices and systems, particularly when the
Sprott 2005).
need is to link enterprises (horizontal ax)
IT is still and enabler of efficiency and ef-

fectiveness (Sonic 2006; Cisco 2005; Plumtree
2005): as long as the development environment
becomes more and more demanding, Business
Figure 6. Creation of Value by using RFID
830
needs to step in and take the lead to properly sup- Robinson 2000) and it is possible to understand
port the roll out of the deployment by facilitating how SOA environment can additionally enhance
the required change in process and organizational the capability with flexibility, speed of implemen-
structure and capturing inputs from the market tation, standardization of protocols and business
and to translate into competitive advantage (La- practices, reducing the uncertainty of a relatively
hiri, 2006; Garfinkel, Rosenberg 2006; Ascential new wireless technology. From an end user point
2005) This means having a clear understanding of view, this scenario can be considered as an ex-
of priorities and issues of RFID deployment and ample of MIMO (Multiple Inputs Multiple Output)
the table below describes them all based on the where companies are able to capture and assess
deployment phase, with the left side being the relevant business information about processed
starting point and markets for a wider and heterogeneous set
Over the years, computer systems and networks of sources at the very same time (Lahiri, 2006;
enabled the tracking of product across the supply Garfinkel, Rosenberg 2006; Ascential 2005, Cisco
chain to make sure that items departing from point 2004 and 2005, Sprott 2004)
A could get point B: in a retail environment, this
is a crucial step to replenish shelves on time and Mobile business, sOA
based on customers’ demand. The acid test is given and semantic Web
by when the shopper decides which product to
select off the shelf and when the shopper decide Mobile technologies introduce advantages that
to buy it again after being satisfied by the product cannot be obtained with fixed connectivity: these
itself (Lahiri 2006, Garfinkel, Rosenberg 2006, include localization and personalization. Previous
Cisco 2004 Brown and Wiggers 2005). paragraphs demonstrate this is valid for RFID
The analysis carried out on previous para- and Bar codes because these allow the delivery
graphs allow us to state the case for SOA as of customized and customizable information and
suitable vehicle for a relatively secure and fast data to user both on local or remote role (Coyle
deployment of the spoken mobile technologies: 2004; Sheshagiri, Sadeh, Gandon 2004)
SOA and service oriented network architecture The Semantic Web is an initiative supported
environment is a flexible enough infrastructure to by W3C aimed to support semantic meaning and
host mobile devices which can now interact with context for internet resources: in such an environ-
other existing applications and systems, both in ment, web services are an enabler for applications
back end or front end side of the company’s IT which can communicate with other automatically
and Business structure. via the internet (both fixed and wireless) using
The ever growing complexity and need of ac- standard internet protocol. It has been demon-
curate and timely information about process and strated and explained that such an opportunity
market data requires the availability of easy to use also exists for RFID and Bar code and SOA can be
and portable devices. Mobile technology can be the delivery platform for these protocols. (Coyle
successfully implemented as extension of Sup- 2004; Sheshagiri, Sadeh,Gandon 2004)
ply Chain Management, Customer Relationship All together these technologies open a com-
Management, business-to-employees for safety pletely new scenario for mobile computing, espe-
and security monitoring and the workflow of real cially because device capability and sophistication
time data and information may be transferred to is increasing and new enhanced devices reach
Enterprise Resource Planning and Business Intel- the market at a fast pace. The wireless network
ligence and more broadly to the whole organization therefore extend the richness and reach of the
and externally to Business Partners (Kalakota, traditional web, which means that there is a need
831
Figure 7. Understanding priorities in RFID
to add meaning to the data that’s is generated and the market at the Rocks area, a concert at Opera
delivered; the broader scope of semantic web House and the Mardi’ Gras Parade. With the
is hence to support the mapping of current and “Exploration” step, it is possible, for example to
future system, protecting the universality of the view some pictures or the places, the map to get to
web with the localized scenario. Once again, a each of them and the timetable for the events and
scenario that fits with RFID and Bar code/ (Coyle eventually restart the loop to shortlist and schedule
2004; Sheshagiri, Sadeh,Gandon 2004) the visit according to user’s own preferences.
Location Based Services and disseminated
Mobile semantic Web @ Work: applications across multiple infrastructures and
Mspace converging fixed/mobile hubs and devices is
another confirmation of the suitability of SOA
MSpace Mobile is a Semantic Web application and mobile semantic web as proper environment
enabling people explore the world around them by enabling and delivering enriching user experience:
leveraging contexts that are meaningful to them in a business-to-business standpoint, more relevant
in time, space and subject (Mulholland, Collins, for this paper, the same comprehensive view of
Zdrahal 2005, Wilson, Russell, Smith, Owens data and information can be achieved by integrat-
and Schaefel 2005). People unfamiliar with any ing RFID and Bar codes and replacing the de-
city, Sydney for example, may find their physical
location to be the main context around which they
Figure 8. Workflow for Mspace
wish all other requests to circulate (Mulholland,
Collins and Zdrahal 2005, Wilson, Russell, Smith,
Owens and Schaefel 2005).
Query like “which are main attractions at
Circular Quay and Sydney CBD I’d like to see
during the weekend within a walking distance from
Sydney Tower” can be gathered in the “Selection”
field of the diagram below, which contains the
main parameters: Sydney, CBD and Circular Quay
area, Sydney tower, weekend events”
The next step is provided by the “Organization”
process which collects the available events on
display during any given weekend: for example,
832
scribed tourist scenario with a proper business Business overview of service oriented service
environment. network architecture (2005). Cisco Systems. http://
www.cisco.com/en/US/netsol/ns477/network-
ing_solutions_white_paper0900aecd803efff3.
cONcLUsION shtml
Cisco Internet Business Solution Group. (2004).
It has been demonstrated how mobile technology
“2010” The retail roadmap for chief executives.
can successfully benefit the business when IT
Cisco Systems US. http://newsroom.cisco.com/
and Business align toward a clear understanding
dlls/2004/ts_011204.html
of targets and when companies are able to capi-
talize on market input. It has been proven, also, Cisco Systems. (2005). Business overview of
how innovative and “off the shelf” mind set can service oriented network architecture.http://www.
create real changing experience for customers cisco.com/sona
and companies. It has also been demonstrated
Coyle, F. (2004). Mobile computing, web services
how a service oriented (network) architecture is
and the Semantic Web: Opportunities for m-
one of the best hubs of business requirements and
commerce. Computer Science department, School
technology features to deliver applications and
of engineering, Southern Methodist University,
contents through mobile devices. This technology
Dallas, Texas.
can deliver a unique view of processes, data and
end standpoint in several environments Dubey, A., & Wagle, D. (2007). Delivering
As the technology is still in its early stage, its software as a service. The McKinsey Quarterly.
potential is not well understood in each aspect and http://www.mckinseyquarterly.com/article_ab-
privacy is one of the most important. Once mobile stract_visitor.aspx?ar=2006&L2=4&L3=43&sr
applications are layered on network infrastructure id=17&gp=0
with shared programs and features, they are part
Garfinkel, S., & Rosenberg, B. (2006). RFID. Ap-
of the internet and exposed to its risk and threats.
plications, security and privacy. In S. Lahiri RFID
Sourcebook. New Jersey: Pearson Education.
rEFErENcEs Kalakota, R., & Robinson, M. (2000). E-Business
2.0: Roadmap for Success. Addison Wesley.
A new service oriented architecture (SOA) matu-
rity model (2006). Sonic Software Corporation. Mulholland, P., Collins, T., & Zdrahal, Z. (2005).
http://www.sonicsoftware.com/solutions/ser- Bletchley Park Text: Using mobile and semantic
vice_oriented_architecture/soa_maturity_model/ web technologies to support the post-visit use of
index.ssp online museum resources. UK: Knowledge Media
Institute, The Open University.
Brown, D., & Wiggers, E. (2005). Planning for
proliferation. The impact of RFID on the network. Retail and consumer goods – RFID deployment
http://newsroom.cisco.com/dlls/2005/Whitepa- (2005). Ascential Company – white paper, US.
per_031105.pdf http://knowledgestorm.fastcompany.com/fastco/
search/keyword/RFID+TAG/RFID+TAG
833
Sheshagiri, M., Sadeh, N. M., & Gandon, F. (2004). KEy tErMs AND DEFINItIONs
Using Semantic Web services for context aware
mobile awareness. Mobile Commerce laboratory, Bar Code (also barcode): A machine-readable
School of Computer Science, Carnegie Mellon representation of information (usually dark ink on
University. a light background to create high and low reflec-
tance which is converted to 1s and 0s).
Simeneau, P. (2005). The OSI model: Understand- Open Systems Interconnection Basic Ref-
ing the seven layers of computer network. Global erence Model (OSI Reference Model or OSI
Knowledge LLS. Model): An abstract description for layered
Sprott, D. (2004). Service oriented architecture: An communications and computer network protocol
introduction for managers. CBDI Forum, Ireland. design.
http://www.cbdiforum.com/report_summary. Radio-Frequency Identification (RFID):
php3?topic_id=20&report=709&order=member_ An automatic identification method, relying on
type&start_rec=0 storing and remotely retrieving data using devices
called RFID tags or transponders.
Strategic decision on SOA (2005). Plumtree Semantic Web: An evolving extension of
Software Inc. LLP while paper US http://www. the World Wide Web in which the semantics of
plumtree.com information and services on the web is defined,
Symons, C. (2005). IT strategy maps. A tool making it possible for the web to understand and
for strategic alignment. Forrester Research US. satisfy the requests of people and machines to use
http://www.forrester.com/Research/Document/ the web content.
Excerpt/0,7211,38215,00.html Service-Oriented Architecture (SOA):
A software architecture where functionality is
Toro, R. (2007). Bon voyage. W3 IBM com grouped around business processes and packaged
corporate. as interoperable services.
White, W. S. (2001). Enabling eBusiness integrat- Service Oriented Network Architecture:
ing technologies, architectures and applications. Emerging technology based on network infra-
Wiley. structure particular set up which is similar to
Service oriented architecture but based on a dif-
Wilson, M., Russell, A., Smith, D. A., Owens, ferent approach.
A., & Schaefel, M. C. (2005). mSpace Mobile: A
Mobile Application for the Semantic Web. IAM
Research Group School of Electronics and Com-
puter Science University of Southampton.
834
835
Chapter 3.19
Enterprise Specific BPM
Languages and Tools
Steen Brahe
Danske Bank, Denmark
Many enterprises use their own domain concepts Business Process Management (BPM) is currently
when they model business processes. They may also receiving much focus from the industry. Top man-
use technology in specialized ways when they imple- agement demands to understand and control their
ment the business processes in a Business Process business processes and agility to adjust them when
Management (BPM) system. In contrast, BPM tools market conditions change. This can be achieved
often provide a standard business process modeling through Process Aware Information Systems
language, a standard implementation technology and (Dumas et al.,2005). A BPM system (Jablonski
a fixed transformation that may generate the imple- and Bussler,1996; Leymann and Roller, 2000) is
mentation from the model. This makes the tools one example of such a system. It allows execution
inflexible and difficult to use. This chapter presents and automation of a business process that can be
another approach. It applies the basic model driven described explicitly as an executable workflow.
development principles of direct representation and Although the hype about BPM and process
automation to BPM tools through a tool experiment automation is high, previous work has shown that
in Danske Bank. We develop BPM tools that capture it is relatively complex to understand, model and
Danske Banks specific modeling concepts and use implement a business process as an executable
of technology and which automate the generation of workflow (Brahe, 2007). First the process must
code. An empirical evaluation reveals remarkable be understood, second it must be formalized and
improvements in development productivity and modeled at a highly conceptual and logical level,
code quality. We conclude that BPM tools should and third the process design must be transferred to
provide flexibility to allow customization to the technology. Many software vendors have complete
specific needs of an enterprise. BPM tool suites for modeling and implementing
business processes. Such tools are mostly based
DOI: 10.4018/978-1-60566-669-3.ch002 on a predefined process modeling language like
Enterprise Specific BPM Languages and Tools
the Business Process Modeling Notation (BPMN) should be able to model its business processes
(White, 2006) for capturing the business process directly in enterprise specific concepts, decide on
at the conceptual level and one technology like a target platform and write transformations that
the Business Process Execution language (BPEL) encapsulate its specific use of technology, and
(BPEL, 2003) for implementing the process. These that automate the generation of code.
tools also assume a fixed development process This leads us to the research question which
where only two models exist, i.e. the conceptual we will answer through this chapter: Does an en-
business process and the implementation. terprise specific BPM tool improve the efficiency
Using such tools causes two challenges for and quality of modeling and implementing busi-
an enterprise that has specific requirements to ness processes, how difficult is it to create, and
its development process, uses its own modeling is it worth the effort?.
concepts and uses technology in specialized ways; We use a design research approach to answer
First, a standardized modeling notation does not this question; We will implement above hypothesis
allow users to use domain concepts and may con- though an experiment where we develop BPM
tain too many modeling constructs which makes languages, tools and transformations for a specific
the tool difficult to use. The models may also be enterprise. Successively, these will be empirically
difficult to understand and use as a communica- evaluated to show the validity of the hypothesis.
tion media. Second, transformation of a model We use Danske Bank, the second largest financial
into implementation must be done manually as institute in northern Europe, as a case study. In
the enterprise may use a variety of technologies lack of sufficient industrial standards, Danske
to implement the process and not only e.g. BPEL Bank has defined its own development process
as many state-of-the-art tools support today. and uses a number of different tools to support it.
Even if one technology as e.g. BPEL is used, the This has caused several challenges as described
enterprise may be using its own implementation by Brahe (2007).
patterns which cannot be generated because the A prototype tool was developed to show that
transformations are hard-coded into the BPM tools. it provide value to develop BPM tools fitted for
The approach behind current BPM tools is the needs of a specific enterprise. The prototype
similar to the extinct Computer Aided Software illustrates that it is possible to do model driven
Engineering (CASE) tools from the 90es. They development of a business process with nearly
also often used a standard modeling language, 100% code generation. The prototype is fitted
one implementation technology and a standard- specially for Danske Banks development process
ized transformation. Their limited flexibility in and consists of three different Domain Specific
supporting enterprise specific standards was one Languages (DSLs) (Mernik et al., 2005) and cor-
of the reasons why they were never accepted responding editors that are used to model a business
(Windsor, 1986; Flynn et al., 1995). process and related information. Furthermore, the
This chapter takes another approach than state- tool provides transformations between the DSLs
of-the-art BPM tools. In order to avoid the CASE and a transformation to BPEL. These transforma-
trap we must come up with an approach that allows tions capture implementation patterns specific for
an enterprise to use its own modeling notations and Danske Banks modeling standards and use of the
specific use of technology. Our hypothesis is that implementation technology. Manual changes can
this can be achieved through applying the basic be introduced into the generated BPEL code by a
model driven development (Stahl et al., 2006) persistence utility feature.
principles of direct representation and automation We use a fictitious Project called Customer
(Booch et al., 2004) to BPM tools; An enterprise Quick Loan throughout the chapter. First, we il-
836
lustrate the current development process in Danske Support for executing and automating business
Bank and the observed challenges of using current processes is achieved through different process
BPM tools. Second, we show how the prototype execution engines. One of them is batch execution
tool eliminates these challenges. We conclude that of process implementations in PL1 and COBOL.
BPM tools customized to a specific enterprise Another one is a BPM system from IBM, where
potentially have a huge effect on the efficiency of the business processes are implemented in BPEL.
a project team and will result in implementations The BPM system has been extended in areas
with fewer errors. where business requirements were not fulfilled.
However, we also conclude that developing For example, the enterprise has created its own
BPM tools from scratch requires high expertise Human Task Manager to handle and distribute hu-
and much effort and is a strategic decision that man tasks that are part of an executable workflow
many enterprises will not take. What we need is and its own task portal where process participants
a tool based framework that allows the enterprise claim and execute human tasks. Furthermore, it
to customize languages, transformations and tools has specific uses of BPEL fault handlers and has
to their specific needs instead of develop them defined specific strategies of capturing business
from scratch. and technical faults during process execution.
The rest of the chapter is organized as fol- Danske Bank has defined its own development
lows. Section 2 introduces Danske Bank and process as no standardized development process
related work. Section 3 introduces the fictitious as e.g. the Rational Unified Process (Kroll and
Customer Quick Loan project. Section 4 discusses Kruchten, 2003) was sufficient to fulfill their
challenges regarding the development process requirements. Business and IT solutions are de-
and used modeling tools. Section 5 abstracts the veloped as one for any business problem. This is
development process into metamodels and algo- in contrast to most development methodologies
rithms for transforming models into code. Based which focus on producing software solutions.
on this abstraction, the developed BPM tools are It is based on service oriented principles (Erl,
described in section 6. Section 7 describes an em- 2005) where business requirements are mapped
pirical evaluation of the tool. Section 8 discusses into required business services and processes.
the results and section 9 summarizes the chapter All important stakeholders are represented at a
and outlines future work. project team to ensure that different issues are
addressed. This includes defining efficient busi-
ness processes and specifying and developing IT
bAcKGrOUND systems that may support them. A project team
includes business process participants, business
This section introduces Danske Bank and presents analysts, solution architects, system developers
related work. and test specialists. Most requirements and design
decisions are captured in models. They are used
Danske bank for documentation and communicative purposes
and as blueprints for the implementation. The
Danske Bank has grown to become the largest development process includes specialized model-
financial group in Denmark - and one of the ing notations and creation of different modeling
largest in northern Europe. It comprises a variety artifacts.
of financial services such as banking, mortgage
credit, insurance, pension, capital management,
leasing and real estate agency.
837
related Work composition languages e.g. Bézivin et al. (2004);

Bordbar and Staikopoulos (2004a,b); Skogan et
Only limited work has previously been reported on al. (2004); Koehler et al. (2003, 2005). Common
customizing business process modeling notations for all approaches is the use of a fixed modeling
and corresponding tools to a specific domain and notation and a fixed transformation. In contrast,
enterprise. In general, most modeling languages Brahe and Bordbar (2006) present a transforma-
like Petri nets (Murata,1989), Event-driven Pro- tion framework that builds upon the use of domain
cess Chains (EPC) and the Business Process specific business process modeling languages and
Modeling Notation (BPMN) (White, 2006) only customized transformations. They also introduce
have one notation which all domains have to fol- a prototype implementation that allows definition
low. An exception is UML activity diagrams that and execution of customized and pattern-based
can be extended by a profile for a specific domain. transformations for a domain specific modeling
UMLActivity diagrams are used by both academia language.
and industry for its extensibility and available Fowler (2005) talks about Language Work-
tool support in form of general UML modeling benches as tools that allow definition and usage
tools. Dumas and Hofstede (2001) argue based of Domain Specific Languages, editors and trans-
on workflow patterns (van der Aalst et al., 2003) formation between languages. Several of such
that the expressiveness of activity diagrams as a workbenches exist such as MetaEdit+ (Tolvanen
workflow language is large and Guntama et al. & Rossi, 2003), GME (Ledeczi et al., 2001), Mi-
(2003) have extended activity diagrams to enable crosoft DSL tools, and many others. The Eclipse
flexible business workflow modeling. There are projects used to build Danske Bank Workbench
also various UML profiles for business process can also be considered as a language workbench.
modeling, e.g. List and Korherr (2005) who The research presented in this chapter follows
consider both the business process flow as well cutting edge trends in language workbenches;
as the business process context. models should be constructed in domain or en-
Brahe and Østerbye (2006) use UML activity terprise specific concepts and transformed into
diagrams as the semantic base for creating domain an implementation. The experiment presented in
specific modeling languages for business process this chapter has been documented in details in a
modeling based on UML profiles. They suggest technical report (Brahe (2008)).
that many enterprises need their own modeling
notations and present a prototype tools that allows
metamodeling of domain-specific workflow- EXAMPLE: cUstOMEr
based languages and automatically generation of QUIcK LOAN
domain-specific tool support in form of editors.
Jablonski and Götz (2007) have a similar approach. The fictitious project “Customer Quick Loan”
They present a flexible and extensible metamodel will be used for illustrative purposes throughout
and the concept of perspective oriented business the rest of the chapter.
process visualization that allows multiple visual Changes in consumer patterns have required
presentations of a business process model. immediate action for introducing a new type of
Model Driven Development (Stahl et al., 2006) customer loans. The new loans can be requested
and the Model Driven Architecture (Frankel, from email and mobile phones with possible im-
2003) have been used extensively in transform- mediately approval and transfer of the requested
ing business process models to implementations, amount to the customers account. A project team
particularly from UML activity diagrams to service is established which includes a loan specialist,
838
a business analyst, a solution architect, system event is classified as external if it is invoked by

developers and a project manager. They name an actor as e.g. a customer for another department
the project Customer Quick Loan. in the enterprise. It is classified as timedependent
In the following we will see how the project if the event is occurring at a certain point in
team follows Danske Banks development process time. The ApplyForLoan event is external as it
to model business events, design the solution, is invoked by a customer while the Payoff loan
specify the physical design and implement it as is time dependent as it is invoked once a month.
an executable workflow.
solution Process Flow
business Events
When a business event takes place it will involve
The business analyst defines all possible business execution of some business logic. For each
events that may occur for a given business case. business event, the business analyst together
A business event is defined as an occurrence that with a solution architect and possible a system
influences a business area and which initiates a well developer will model such business logic as a
defined business process in this area. The events Solution Process Flow (SPF), which is a tech-
are described in a model called an event map. nology independent or logical business process
For the Customer Quick Loan the primary model. Each task in the Solution Process Flow
events are ApplyForLoan, which occurs when a must either be of type Automatic, which means
customer requests a loan, and PayoffLoan, which handled automatically by a service invocation,
occurs each month after a loan has been created. Manual as e.g. moving papers from a table to
The event map is defined as a UML class diagram an archive or UserExperiance as e.g. creating
in Rational Software Modeler (Swithinbank et al., a loan using an application user interface. The
2005) (Figure 1). Additional information about the IBM Websphere Business Modeler is used to
events is specified in MS Word documents. An define these models.
Figure 1. Business events for the Customer Quick Loan project modelled as a UML class diagram in
Rational Software Modeler.
839
The Solution Process Flow for the primary busi- COBOL for batch execution, and finally some
ness event, ApplyForLoan, is illustrated in Figure may not be implemented by IT at all. The project
2. It consists of four logical tasks; the automatic team decides to automate the execution of the Ap-
ApproveLoan task will make a risk calculation of plyForLoan process by implementing it in BPEL.
the customer. If the risk is high, the loan request is Two kind of physical specifications now have to
rejected; a process participant will be notified by be made: BPEL process design, also called Control
the Reject task of type UserExperience, and will Flow Behavior, and a Workflow Specification,
have to send a rejection message to the customer which contains additional information required
using an application interface. If the risk is low, to implement the Solution Process Flow and all
the loan, or possible several loans applied for at its tasks in the BPM system.
once, will be created by the automatic Create-
Loans tasks, and a confirmation will be sent to control Flow behavior
the customer by the Confirm task.
The project team has examined the local service The Solution Process Flow is the starting point
repository for existing services and has found for the Control Flow Behavior, a model of the
that two existing service operations called in a physical implementation in BPEL which is also
sequence will fulfill the requirement for a Confirm created using Websphere Business Modeler. Three
task in the Solution Process Flow. Therefore, the physical design decisions make them different.
Confirm task is further broken down and modeled This is described in the following three sections.
in a separate sub process as illustrated in Figure
3. First, a service operation is invoked to create Separate vs. Inlined Subprocess
the content of the confirm message, and second, a
service operation is invoked to send the message It must be decided if the Solution Process Flow
by SMS, email or letter. should be implemented as one BPEL process,
or if it should be broken down into several. Ex-
Physical Design tracting parts of the process into sub processes
has advantages: More than one developer can
Some business processes may be implemented simultaneously work on the construction, it is
in BPEL, others may be implemented in PL1 or easier to make a change and deploy a small sub
Figure 2. Solution Process Flow for the ApplyForLoan business event modeled in Websphere Business
Modeler.
840
Figure 3. The Confirm task modeled as a sub

complex data transformations inside in a BPEL
process.
process that are externalized as separate service
invocations, it could be synchronization of data
between different systems that make up the ex-
tended BPM system in Danske Bank, or it could
be a specific service invocation that updates the
business state for the specific process instance, a
feature of Danske Banks extended BPM system.
An additional service invocation has to be
inserted in the Control Flow Behavior for the Ap-
process compared to change and deploy the main plyForLoan process after the AssessRisk service
process, and a sub process can be reused by other invocation. It updates the business state of the pro-
processes. Though, extensively use of sub process instance to either “Approved” or “Rejected”.
cesses has the disadvantages of maintaining and This information can be viewed by employees in
operating several processes instead of one main the Danske Bank through the Human Task Portal.
process. This causes an overhead and introduces The additional functionality is not required if the
complexity regarding change management. The Solution Process Flow was implemented using
developer decides to implement the Confirm sub another technology like COBOL (see Figure 4).
process as an inlined sub process in the ApplyFor-
Loan BPEL process. This is illustrated in Figure Implementation Patterns
4. The developer could also have implemented the
sub process as a separate BPEL process. Each task in the Solution Process Flow has to be
mapped to a task in the Control Flow Behavior.
Technology Dependent Functionality A task can be implemented by different imple-
mentation patters. In this context implementa-
Functionality needed for the implementation in a tion patterns are patterns, or code templates and
specific technology should not be modeled in the rule used by Danske Bank to implement tasks of
Solution Process Flow. Using BPEL, this could be different types. An Automatic task type can be
Figure 4. Control Flow Behavior of the ApplyForLoan SPF.
841
implemented by three different patterns; Servi- data values in three to five different languages
ceOperation, MultipleInstances and Bundle. Tasks must be described to be presented to the process
of type Manual and UserExperience are always participants in the task portal, and rules about
implemented using a HumanTaskManual or a Hu- earliest start of the task and a possible deadline
manTaskGUI pattern. The patterns are explained and several others also has to be specified. The
a later section. When modeling the Control Flow HumanTaskGUI task further has a link to an
Behavior, these pattern names are used to clas- existing application interface where the process
sify all tasks in the same way as the Automatic, participant has to handle the task. It must also
Manual and UserExperience classifiers were used be specified which data values from within the
in the SPF model. Table 1 lists how tasks from running BPEL process instance the link should
the ApplyForLoan Solution Process Flow have transfer to the business system. For the process
been mapped to the Control Flow Behaviour. itself, additional information also has to be
The implementation pattern to be used in the specified. This includes input data for the pro-
physical design is determined from the task type cess, allocation rules, and description in several
in the SPF and the description of the task in the languages for presentation in the task portal.
corresponding System Use Case. All information for one task is specified in
a MS Word document and is called a Workflow
Workflow specific Information Task Specification. A Word template is available
for each task type which describes required in-
Much information has to be specified to implement formation. Six task specification documents are
the Control Flow Behavior in the BPM system. For created for the ApplyForLoan process.
a ServiceOperation task this includes information
about which service operations to invoke, input bPEL construction
and output data structures, exception handling and
escalation of errors, if the task must be restarted All required information and design decisions
in case of failures during service invocation etc. are now available, and the BPEL process can be
A task of type HumanTaskManual or Human- constructed after the Control Flow Behavior and
TaskGUI is a task handled by humans. Process Workflow Specification have been completed.
participants will be able to list, claim and execute A system developer now maps the Control
such a task from a task portal. For both type of Flow Behavior into a BPEL process. From the
tasks following information is needed; groups Workflow Specification he is able to specify input/
allowed to claim and execute a task are defined output data, set attributes about the process as
as Allocation Rules, labels, descriptions and e.g. when it is valid from, if it is a long running
process etc. Also other systems as e.g. Danske
Banks proprietary Human Task Manager can be
Table 1. Solution Process Flow tasks mapped into populated with allocation rules specified in this
Control Flow Behavior. document. Each task in the Control Flow Behavior
is mapped to the BPEL implementation based on
Task Implementation pattern
the developer’s knowledge of BPEL implemen-
AssessRisk ServiceOperation
tation patterns in Danske Bank. Each task type
CreateLoans Bundle
introduced in previous section has a certain BPEL
CreateContent ServiceOperation
template and an algorithm for how to implement
SendMessage ServiceOperation
it. The pattern names and corresponding BPEL
Reject HumanTaskGUI
templates are illustrated in Table 2.
842
The Service Operation pattern invokes a service The service operation invoked in the loop may
operation and incorporates specific way of using initiate another process or thread that runs con-
logging and fault handling. All service operations currently. For some business scenarios the main
in Danske Bank throw a Technical Fault, which is business process is not allowed to continue before
caught by the fault handler for the Invoke node. all initiated processes behind these service invoca-
The fault handler forces the invoke node into tions have finished. Danske Bank has extended the
stopped state. BPM system with infrastructure functionality that
The MultipleInstances pattern is a loop con- allows such a mechanism. In the BPEL process it
taining a service invocation as implemented by is called the Bundle pattern and is implemented
the Service Operation pattern. It is similar to the as the MultipleInstances pattern followed by an
workflow pattern “Multiple instances without event. At runtime after invoking the service op-
priori runtime knowledge” (van der Aalst et al., eration a number of times, the main process will
2003). wait until all the initiated concurrently running
processes have finished. The BPM infrastructure
extension will be notified about the state change
Table 2. Danske Bank specific BPEL implementa- and will fire the event that will cause the BPEL
tion patterns. The dots are replaced with infor- process to continue executing.
mation from the Control Flow Behavior and the The HumanTaskManual and GUI patterns
Workflow Specification documents. are implemented by invoking a specific service
operation exposed by Danske Banks Human Task
Pattern name BPEL template Manager followed by an event node. The transla-
ServiceOperation <assign name=. .>. .</assign> tion of a task and its related information is purely
<invoke name=..>
<catch faultName=TechFault . .> manual, even though it is the same patterns that
.. are implemented multiple times.
</catch>
</invoke> Above descriptions only show a subset of the
MultipleInstances <assign name=. .>. .</assign> implementation steps that the developer has to go
<while name=. .> through when implementing the tasks from the
<assign name=. .>
<invoke name=..> Control Flow Behavior. Common for all patterns is
<catch faultName=TechFault> that data mappings also have to be specified before
..
</catch>
invoking a service operation. Control flow logic
</invoke> also has to be specified by the developer. This is
</while> described at the edges that connect the tasks in
Bundle <assign name=. .>. .</ assign> the Control Flow Behavior model.
<while name=. .>
<assign name=. .>
<invoke name=..>
<catch faultName=TechFault>
.. A NEED FOr cUstOMIZED tOOLs
</catch>
</ invoke>
</while> The development team faces several challenges
<receive name=..>. .</receive> in using the described BPM tools and develop-
HumanTask <assign name=. .>. .</assign> ment process:
<invoke name=SetTaskInQueue>
<catch faultName=TechFault>
Difficult to use domain concepts. Danske
.. Bank has defined its own concepts for modeling
</catch>
business processes, but it is not possible to create
</ invoke>
models by directly using these concepts. Tools
843
have been twisted and tweaked to force them to The core of the problem is that the com-
behave as desired. The usability is low and it is mercial tools used presume one development
hard to use the models for communicative means. process defined by the software vendor, a fixed
Difficult to comprehend information. A set of modeling languages and a specific way to
number of different tools are used to describe use the implementation technology. This is in
and specify how a business process should be deep contrast to the requirements from Danske
implemented. The developer and the architect Bank who found the standard development pro-
therefore need to look into several different tools cess and standard notations insufficient for their
and models to find relevant information. needs. They need to build their own development
Missing traceability and consistency. It is process into the tools, to use their own modeling
difficult to find relevant models because trace- notations and artifacts and to define their own
ability between models is handled by textual use of technology.
descriptions. Furthermore, a model created in one
tool cannot refer to models created in other tools.
Consistency between models must therefore be AbstrActED DEVELOPMENt
handled manually. PrOcEss
Imprecise data definitions. The data defini-
tions can only be interpreted by humans. For In order to develop tools that address above chal-
instance, the name and version of the service lenges, we need a formal description of required
operation is specified in the MS Word Workflow information and transformational algorithms. In
Specification document. this section we therefore use the Customer Quick
Because of above challenges, transformation Loan example to abstract the development pro-
of models and information from specification cess into metamodels and transformations. First,
and physical design into physical artifacts as we give an overview of the current development
e.g. BPEL have to be done purely manually. The process and describe requirements to a model
system developer needs to open models in Ratio- driven development process. Second, we intro-
nal Software Modeler and Websphere Business duce the abstracted development process, which
Modeler and retrieve information manually, and uses the metamodels and transformations that we
he/she must open many MS Word documents to will develop in this section. Last, we define these
get detailed information about design decisions. metamodels and describe algorithms of how to
Although model driven development is the goal carry out the transformations. The metamodels
of the development process, the result is mere a and transformation algorithms form the basis for
document driven development process. the prototype tool described in next section, which
For the simple example of the ApplyForLoan has been developed specific for Danske Bank.
business process, the number of models and docu-
ments get high even for a simple example with only current Development Process
four tasks. One RSM model, two WBM models and
about 10 word documents make up the specification. Figure 5 gives an overview of the development pro-
It is quite difficult to comprehend the large amount cess described in previous section, and illustrates
of distributed information required for constructing the created artifacts as well as design decisions.
the BPEL code. Further, the construction process The artifacts are depictured with rounded boxes
is inefficient and error-prone as much of the infor- to indicate they are not precisely modeled, and the
mation from the specification has to be manually clouds indicate decisions that are not documented
reentered into the physical artifacts. but instead put directly into models or code. Much
844
of the information required through the develop- and then generate the implementation code. In our
ment process is described as plain text. A human example this means transformation of a Solution
must read and interpret it to be able to construct Process Flow into a Control Flow Behavior from
the implementation. The cloud between the Solu- which the BPEL implementation and related
tion Process Flow and the Control Flow Behavior documents can be generated. In general, three
illustrates that decisions about how the physical basic requirements must be fulfilled to enable
design are taken by the architect or developer; an efficient model driven development process:
First, for each sub process modeled in the SPF it
must be decided if it should be implemented as Information and design decisions must be speci-
an inlined flow or as a separate process. Second, fied precisely in models.
additional functionality must be specified. By Transformation between models must be for-
defining the Control Flow Behavior from scratch, mally described.
but inspired by the Solution Process Flow, the Information added to generated models or code
possibility to have tool based consistency check must survive future transformations.
between them is lost. The Control Flow Behavior
model needs to be manually updated each time the Creating precise models requires availability
Solution Process Flow changes. The cloud between of languages or metamodels that support modeling
the Control Flow Behavior model and the BPEL standards and which allow modeling of all required
code indicates decisions taken by the developer information in a precise manner. As Danske Bank
about BPEL specific information as e.g. the name has created its own notations and use technology
of the project where the code is being developed, in specific ways, they need to be able to express
default package name, target namespace to use in this in their models.
the BPEL process, if generated WSDL files are
kept in separate directories, etc. Abstracted Development Process
requirements to a Model Driven Figure 6 illustrates the model driven develop-

Development Process ment process that we will describe through the
rest of this section. It uses metamodels, called
One of the main ideas behind model driven devel- Eventmap, SPF and WFSpec, for modeling event
opment is to have tools that can transform platform maps, Solution Process Flows and Workflow
independent models to platform specific models, Specifications.
Figure 5. Current development process with main modeling artifacts and decision points. The clouds
indicate that decisions are not documented, and the rounded boxes indicate that no metamodels are
used.
845
Figure 6. New development process with metamodels and transformations. Information is specified
precisely by using the Eventmap, SPF, WFSpec, BPELCodeGen and ModelInjections metamodels.
It further uses a ModelInjection metamodel, The Control Flow Behavior model has dis-
and a BPELCodeGen metamodel. They are used to appeared as it is indirectly generated from the
capture decisions currently taken in the “clouds”. Solution Process Flow, the WFSpec model and
The metamodels form the basis for algorithms that theModelInjection model. The development pro-
can generate models and code. The BPELCode- cess illustrated in Figure 6 has been implemented
Gen metamodel is used to describe specific BPEL in the prototype tool described in next section
implementation decisions, while the WFSpec that uses the metamodels to capture information
metamodel and the ModelInjection metamodel precisely and transformations to automate the
are used to describe the three differences between generation of the BPEL implementation.
the Solution Process Flow and the Control Flow
Behavior described in a previous section. Metamodels
Decisions about how to implement sub pro-
cesses modeled as part of the Solution Process The five metamodels introduced above will now be
Flow is captured by the WFSpec metamodel. It described. They have been developed by analyzing
also specifies workflow specific information for the current development process. This includes
each task in the Solution Process Flow. discussions with development teams, enterprise
Additional technical functionality is modeled architects and examination of educational material.
as separate process fragments, also called model
injections as it is to be injected at a specific point in Eventmap Metamodel
the Solution Process Flow to generate the Control
Flow Behavior. Process fragments are modeled The Eventmap metamodel, depictured in Figure 7,
using the Solution Process Flow metamodel. The expresses how events can be modeled and related.
relation between a process fragment and where to The metamodel has incorporated all information
inject it is captured by a ModelInjection metamodel. that previously was described as plain text in
The implementation patterns to be used for MS Word documents. Inheritance has been used
implementing tasks in the Solution Process Flow to define the External and Timedependent event
are documented by the WFSpec metamodel, for types and requirement for different information.
instance that an Automatic task is implemented by An abstract Event metaclass contains attributes for
the ServiceOperation or the Bundle implementa- information common for both types of events while
tion pattern. the TimedependentEvent and the ExternalEvent
846
Figure 7. Event map (EventMap) metamodel. An event can either be external or timedependent and
consists of a number of scenarios.
subclasses contain specific attributes, which were in MS Word documents. A WFSpec model re-
previously described in MS Word documents. fers directly to a Solution Process Flow model
instead of referring to a Control Flow Behavior
Solution Process Flow Metamodel model, as this is not explicitly modeled after the
introduction of the Model Injection concept in
A Solution Process Flow is constructed for previous section. Much information is required by
each event modeled in the event map. The SPF the WFSpec metamodel, therefore only selected
metamodel is illustrated in Figure 8. It is a simple parts of it are described here. The SPF4WFM
flow based metamodel that reminds much of a metaclass is the main element. It refers to a
UML activity diagram. The difference is the use Solution Process Flow model and has several
of the domain specific task types, i.e. Automatic, attributes specifying information required for
Manual and UserExperience, and the domain implementing the BPEL process, e.g. a deadline
specific edges, i.e. ProcessConnection, Dialog- rule, process lifetime information, allocation
Connection and ProcessDialogConnection. rules about process responsibility, department
owner, process type etc. Many of these attributes
Workflow Specification Metamodel are specific for Danske Bank as a BPEL process
implemented in the BPM system is a part of
The Workflow Specification (WFSpec) metamod- a larger proprietary case system that extends
el, illustrated in Figure 9 is a formalization of the commercial BPM system with additional
the Workflow Specification previously defined functionality.
847
Figure 8. Solution Process Flow (SPF) metamodel. Tasks are modeled by the Automatic, Manual, User-
Experience and SubProcess tasks types and connected in a control flow by using edges of type Process,
Dialog or ProcessDialog.
The SPF4WFM metaclass contains a number of Addtional Metamodels

Task- Specification elements. A TaskSpecification
can either be an AutomaticSpecification, Manu- Two more metamodels have been defined to cap-
alSpecification, UserExperienceSpecification or ture additional information in the development
a SubProcessSpecification. An Element of one of process. The BPEL code generation (BPELCo-
these metaclasses refer to a task of type Automatic, deGen) metamodel is used to store decisions of
Manual, UserExperience and SubProcess respec- how to implement a physical design in BPEL. It
tively. A TaskSpecification specifies required refers to a Solution Process Flow and specifies
additional information for the implementation in target namespace to use for the BPEL process,
BPEL and which implementation pattern to use. name of the project that should contain the BPEL
Previously, information about the implementation process, base package name to define the BPEL
pattern was stored directly in the Control Flow process in, if WSDL files should be located in
Behavior while additional information was stored the separate folders, and the name of the base
in MS Word documents. WDSL folder name. In previous section it was
described how the Control Flow Behavior could
be generated based on the Solution Process Flow
model and model injections. The ModelInjection
848
Figure 9. Workflow specification (WFSpec) metamodel. The SPF4WFM metaclass refers to a Solution
Process Flow and contain a number of TaskSpecifications. A task specification refers to a task in a Solu-
tion Process Flow and can be of type Manual, UX, Automatic or SubProcess
849
metamodel keeps track of all process fragments has been modeled and the ModelInjection model
to inject and where to inject them in a Solution created, and BPEL specific attributes have been
Process Flow model. defined in the BPELCodeGen model. Algorithm 3
merges information from these three models with
transformations information from the Solution Process Flow model
and generates BPEL code and related artifacts.
Now, when all information required during the Only control flow logic and data mappings have
development process can be stored precisely in not been generated.
models, we are able to explicitly define the trans-
formation algorithms illustrated in Figure 6 that
depictured the abstracted development process. tOOL DEVELOPMENt
Here, we only give a short description of the
algorithms. Detailed descriptions in pseudo code In this section we describe a tool called Danske
can be found in Brahe (2008). Bank Workbench (DBW) that was developed as
part of this research. It implements the metamod-
From Event to Solution Process Flow els and transformations described in last section.
Hence, it directly supports Danske Banks devel-
One Solution Process Flow model has to be cre- opment methodology, modeling concepts and use
ated for each event in the eventmap. Algorithm of technology.
1 in Figure 6 creates an empty Solution Process
Flow model. It is named and stored according to tool Architecture
Danske Banks standards. The analyst and the ar-
chitect model the behavior of the business process Danske Bank Workbench is built on Eclipse plat-
inside the generated model. form (Eclipse, 2008) and various Eclipse open
source projects. The Eclipse Modeling Framework
From SPF to Workflow Specification (EMF) (Budinsky et al., 2003) has been used for
defining the abstract syntax, or metamodels of the
Algorithm 2 in Figure 6 generates the Workflow DSLs, that have been implemented. The concrete
Specification (WFSpec) model based on the syntax of the DSLs and editor support have been
Solution Process Flow model. It is named and implemented by using the Graphical Modeling
stored according to Danske Banks standards. A Framework (GMF, 2008), while openArchitec-
TaskSpecification class is generated and added to tureWare (oAW) (oAW, 2007) has been used to
the WFSpec model for each task in the Solution implement the semantics of the DSLs as model-
Process Flow. The WFSpec model contains all to-model and model-to-text transformations.
required information, but all attributes contain Danske Bank Workbench consists of several
default values. Successively the architect therefore independent tools for developing the different
fills it with correct information. artifacts in the development process. These are
depictured in Figure 10, which also illustrates
Generation of the BPEL Implementation dependencies to other Eclipse projects. The
names of the projects conform to the names of
The BPEL implementation can be generated by the metamodels previously described.
Algorithm 3 in Figure 6. All required information
has been specified; information about each task is
stored in the WFSpec model, process fragments
850
Figure 10. Overview of Danske Bank Workbench and its dependencies of other Eclipse projects.
Metamodels and Editors the initial node. When a model injection or a sub
process is detected, the corresponding process
All metamodels have been modeled in Rational fragment or sub process is be interpreted and BPEL
Software Architect as UML class diagrams. Each code generated, which must then be merged into
of these were exported as an XMI representation the partly generated BPEL code. It requires much
of UML and imported into Eclipse by using the book keeping handling the associations between
EMF model creation wizard which comes as a models as four different models, i.e. SPF, WFSpec,
part of the EMF project. The GMF editors were ModelInjection and BPELCodeGen are used as
created based on the EMF metamodels. Figure input to the transformation. A number of utility
11, Figure 12 and Figure 13 illustrate the GMF functions have been written in Xtend and in Java
based Eventmap, SPF and Workflow Specification to support this.
(WFSpec) editors in action.
tool Utilities
transformations
Several tool utilities have been developed to
The three transformation algorithms described enhance usability of Danske Bank Workbench
previously have been implemented in oAW. Algo- and to smoothen the use of the different tools.
rithm 1 and 2 has been implemented as model-to- The users of the tools are guided from one step
model transformations using the Xtend language. in the development process to the next by using
Algorithm 3, which is supposed to generate BPEL these utilities.
code, has been implemented as a model-to-text
transformation using the Xpand language. The Transformation Execution
implementation is quite complex. It is imple-
mented as a graph transformation that recursively One kind of tool utility is the generation of “the
runs through the SPF control flow starting with next” development artifact in the development pro-
851
Figure 11. Event map editor with events for the CustomerQuickLoan project. External and Timedepen-
dent events can be modeled directly from the tool palette and required information can be specified as
properties.
cess, i.e. execution of transformation workflows The executed action looks up the defined service
that implement Algorithm 1, 2 and 3. They are operation in (a mock up of) Danske Banks cen-
implemented as actions that appear on the context tralized service repository, retrieve definitions of
menu when the user right-clicks on an event in an data structures and updates the WFSpec model
Eventmap model, at the canvas for an SPF model with these.
and at the canvas for a WFSpec model.
Persistence of Manually
Service Repository Data Extract Changed BPEL Code
The architect has to find definitions of input and Generated BPEL code needs to be updated with
output data structures for service operations and data mapping and control flow logic. A small
put them into the WFSpec model. The user right- persistence framework has been developed that
clicks on a task specification for an Automatic allows the developer to persist logic from within
task and chooses “Retrieve Repository Data”. an assign node or a control link in a separate file.
852
Figure 12. Solution Process Flow editor. The ApplyForLoan process is modeled. Task and connection
types are available from the tool palette. The concrete syntax is customized for tasks as well as edges.
Figure 13. WFSpec editor with task specifications for the ApplyForLoan process. Information can be
modeled precisely for Automatic, Manual and UserExperience tasks.
853
The developer simply right-clicks on the assign (Figure 16b). An empty Solution Process Flow is
node or control link and chooses “Persist element”. now generated in a subfolder named “SPF” and
The action creates a separate file where the assign is given the same name as the business event. It
or control flow logic is persisted. Next time the is then modeled by either the business analyst or
transformation that implements Algorithm 3 is the solution architect. Tasks may now be modeled
executed, the changed BPEL code is overridden, directly as Automatic, Manual or UserExperience
but successively, the persisted changes are copied (as defined by the SPF metamodel) by dragging
into the newly generated BPEL code. them directly onto the canvas from the tooling
palette. Connections of type Process, Dialog or
customer Quick Loan retooled ProcessDialog are also directly available.
Danske Bank Workbench will now be illustrated Workflow Specification

by applying it at the example. Figure 14 illustrates
a workflow of the development process with the The architect right-clicks at the Solution Pro-
artifacts that are created and the transformations cess Flow when it is complete (Figure 16c) and
between them. The letter tags in the figure refer chooses “Generate WFSpec model” (Figure 16d).
to screen dumps of tool utilities and artifacts A WFSpec model is now generated under the
developed for the ApplyForLoan process. After Implementation folder and a subfolder named
Figure 15 they are depictured in Figure 16, Figure after the Solution Process Flow. It contains task
17, and Figure 18 which can be found in the end specifications for all tasks and has been populated
of the chapter. with default data. The task specifications and
all objects inside them conform directly to the
Eventmap WFSpec metamodel.
Now, information has to be entered into the
First, a business analyst creates a new Eventmap. specification. For example, the architect defines
All business events are now modeled as either ex- that the CreateLoans task must be implemented
ternal or timedependent, and scenarios are added to by the Bundle pattern; he or she selects the Opera-
each event (Figure 16a). The editor provides direct tion object in the CreateLoans task specification
support for these concepts from the tool palette. and in the properties view changes the type from
The analyst simply drags and drops events and ServiceOperation to Bundle (Figure 16i).
scenarios to the canvas. The property view reflects
properties for the selected event type, where e.g. Subprocess
priority can be selected as low, medium or high
and business possibilities can be described. Event The Confirm task is modeled as a SubProcess task
types and properties directly reflect the defined type in the Solution Process Flow. The architect
Eventmap metamodel. chooses that it must be implemented as an inlined
flow in the BPEL process by selecting the Confirm
Solution Process Flow task specification in the WFSpec model and sets
the Type property to InlinedFlow (Figure 16j).
After finishing the event map, the business analyst The subprocess to which the Confirm task refer
has to create a Solution Process Flow for each is generated by right clicking on it and choose
event. The analyst simply right-clicks on the “Generate SubProcess”. An empty sub process
event, for instance the ApplyForLoan event, in is created and opened automatically. It is named
the Eventmap editor and chooses “Generate SPF” accordingly to the name of the Confirm task and
854
Figure 14. Workflow for using tools through the customized development process. Letter tags refer to
screen dumps in Figure 16. A thick arrow indicates a tool utility while a thin arrow indi
Figure 15. File structure of the Customer Quick Loan project containing all generated files.
855
Figure 16. Using Danske Bank Workbench
856
Figure 17. Using Danske Bank Workbench (cont’d)
857
Figure 18. Using Danske Bank Workbench (cont’d)
optionally put in a sub directory if the path property Model Injection

at the Confirm task has a value. The subprocess
is now modeled as a sequence of two automatic The architect and the developer recognize that an
activities (Figure 16g), and its Workflow Speci- additional task is needed in the physical imple-
fication can be generated (Figure 16h). mentation. The task should set the business state
858
of the process instance to either “Approved” or Subsequently, the BPEL code generator can use
“Rejected” depending of the outcome of the As- these data structures to create a correct WSDL
sessRisk task. document for the service operation. Without
They right click on the control link in the Solu- this import utility, the developer had to find data
tion Process Flow that connects the AssessRisk definitions and create XSD schemas manually.
task with the CreateLoans, and the Reject task
and choose “Create Injection”. An empty model Code Generation
is created under the Injections folder and is auto-
matically opened in an SPF editor. The developer The developer sets parameters for the code genera-
models the process fragment as one automatic tion in a BPELCodeGen model before executing
activity (Figure 16l) and generates the Workflow the BPEL code generation. Previously, these de-
Specification for it (Figure 16m). sign decisions were not documented. Figure 16r
The book keeping of Model injections are illustrates that the developer has selected default
handled by the ModelInjection model. This model values for the code generation; The BPEL code
is illustrated in Figure 16q. It contains one injection will be generated in the same project as where the
that has two important properties; the injection models are, and WSDL files will be located in the
point in the Solution Process Flow, which is the same directory as the BPEL file. The developer
ID of the control link, and the process fragment generates the BPEL code by right-clicking on the
to inject (SPF to Inject) at the injection point. WFSpec model and chooses “Generate BPEL”
Before having the WFSpec metamodel and (Figure 16n).
editor, all the design decisions were modeled in Figure 15 shows the Customer Quick Loan
the Control Flow Behavior without any reuse of project and all files generated through the de-
the Solution Process Flow, and required additional velopment process and Figure 16p shows the
information was defined in textural documents. generated BPEL code opened in the Eclipse open
Now, the project team has modeled three processes; source BPEL editor. Only the event map has
one SPF, one subprocess and one process frag- been created manually. The rest of the artifacts
ment. They are all modeled in the same language have been created by tool utilities supporting the
and have each a corresponding WFSpec model. enterprise specific development process. Hence,
the file structure follows specified standards, and
Synchronizing Data the traceability between models can be ensured.
Without Danske Bank Workbench these artifacts
All automatic task specifications must be synchro- and all the information bookkeeping are handled
nized with the centralized Service Repository to manually by the project team.
obtain correct input and output data definitions.
Figure 16o shows the selected CreateContent task
specification in the WFSpec model for the Confirm EVALUAtION
subprocess. The modeler has right-clicked on it
and selected “Retrieve Repository Data”. The op- The tool was evaluated through an empirical test
eration name for the CreateContent specification which involved five people employed at Danske
has been set to CREATECONTENT. The same Bank. They have all worked as workflow develop-
operation name exists in the service repository, ers. Two of them have experiences from working
which can also be seen in the figure. The action as - or closely together with - a business analyst,
now retrieves data definitions from the repository and one of them is a solution architect. They used
and populates the WFSpec model with these. Danske Bank Workbench to model an event map,
859
a Solution Process Flow and a Workflow Specifi- work. This is reflected in question 6, 7, 9, 10 and
cation and generate the BPEL code. The business 11. They all got an average score at 5 out of 6,
scenario was the same as presented in this chapter. which is equal “Better”. One of the developers
They got a one-page description of how to use suggested that data and control link logic should
the tool. From the description they used about also be modeled to allow 100% code generation.
30-40 minutes to complete the exercise. They Some of the developers were quite skeptical about
filled out questionnaires with 12 questions and the quality of the generated code as they suspected
were interviewed about their experiences with the that manually written code would perform better
tool. Each question asked about the experience than generated one (question 8). However, they
of using the tool compared to current practice in thought that the number of errors would be lower
Danske Bank. The questions and their ratings can in generated code compared to manually written
be found in Table 3. code (question 12).
The developers found that the tool would im- The evaluation has several limitations; The
prove their productivity significantly and it would number of participants could be increased and in-
be easier to work with. Especially, they were happy clude a more diverse population of people, the case
with the Danske Bank specific modeling capabili- could be extended to a realistic business scenario,
ties. It was much easier and intuitive to work with and the questions could be accomplished with
domain specific modeling. Further, it was easier to measurements of the number of errors in-, and the
comprehend the workflow specific information that efficiency of the code. Further, a control group of
had to be specified for the Solution Process Flow developers could use the current method and tools
(question 1-5). Some of the developers suggested in Danske Bank at the business scenario, and the
that validation rules would improve the develop- results from the two groups could be compared.
ment process as a modeler would be caught if However, making a realistic evaluation of a
required information was not specified. prototype tool in an industrial setting is extremely
In general, they found that model driven difficult. It is hard to get permission to use time
development would help improving their daily from the right people, the tool is not mature enough
Table 3. Questions and answers for the empirical evaluation. The ratings were: 1 is “Much worse”, 2
is “worse”, 3 is “a little worse”, 4 is “a little better”, 5 is “better” and 6 is “much better”
Question Mean Value

1 How is the Danske Bank specific syntax to work with compared to Websphere Business Modeler? 5.5
2 How is it to work with the WFSpec editor compared to MS Word? 5.6
3 Is the information easier to comprehend and access? 5.2
4 How is it to comprehend the number of modeling artefacts and locate where they are? 5.4
5 Are the tool utilities helpful and support the developement process? 5.4
6 Is the code generation to prefer over manual translation? 5.0
7 Do you believe in model driven development as the right direction to go in? 5.0
8 How is the quality of generated code compared to manually written code? 3.8
9 Do you prefer to model and generate the solution instead of manually implement it? 5.0
10 Does the tool eliminate tedious work? 5.0
11 Will the tool influence on the development productivity? 5.0
12 Will the tool decrease the number of errors in implemented code? 4.8
860
for large scale testing, and the tool requires more Restrictions on the SPF. The prototype only
development to model a realistic business scenario. supports a one-to-one relationship between a
Despite these limitations, the evaluation of the Solution Process Flow and a BPEL implemen-
tool shows that it provides significant improve- tation. In reality there are often cases where an
ments over the current use of commercial tools SPF might be divided to several BPEL processes
in Danske Bank. or several SPF’s may be merged into one BPEL
implementation.
Implementing above items in Danske Bank
DIscUssION Workbench is a demanding task; First, it requires
further analysis of requirements in Danske Bank.
We have used Danske Bank Workbench for model- Second, it requires design and implementation
ing and implementing the ApplyForLoan business of several advanced tool utilities. Especially, the
process. The exemplification of the tool and the last item may show up to be very hard to specify
empirical evaluation has shown that the develop- and implement.
ment process becomes more efficient as the different
experts are supported directly in their work. They are
able to use familiar domain concepts directly in the sUMMAry AND FUtUrE WOrK
modeling tools, they are guided to provide correct
information, and execution of the transformation In the introduction of the chapter we postulated
algorithms has been automated. We have shown that general purpose business process modeling
that it is possible to define and utilize a number of and implementation tool suites are not feasible
DSLs and tools to effectively support an enterprise for many enterprises. Using the case study of
specific development process for business process Danske Bank and an example we showed that a
modeling and implementation. development team faces many challenges when
Danske Bank Workbench is a prototype, and they use standard modeling languages and tools
therefore it has a number of limitations and points but have to use enterprise specific modeling nota-
for improvements; tions, follow an enterprise specific development
Consistency checking. We have not defined process and use technology in specialized ways.
methods, nor implemented tools to check consis- We abstracted the development process into
tency between different models. metamodels and transformational algorithms and
Validation and modeling constraints. Vali- developed a tool called Danske Bank Workbench,
dation rules and constraints on how a model can fitted specially for Danske Bank. The tool imple-
be constructed should be specified by the team mented the model driven development principles
responsible for defining the metamodels. These of direct representation and automation as it al-
rules and constraints should be handled by the lowed creating models directly in Danske Bank
modeling tool to avoid creation of invalid models. specific concepts and it automated the generation
Controlflow. Several controlflow structures of lower level models and code.
cannot be handled by the transformation such as We saw through the example that it is possible
cyclic behavior and loop constructs. to achieve an efficient model driven development
Data mapping. It might improve the proto- process where a project team collaborate to create
type and the development process to abstract the different modeling abstractions of a business pro-
definition of data mappings to either the WFSpec cess with tool based transformations and ensured
model or to a generated Java class which would synchronicity between the different modeling
be responsible for the data mapping. abstractions. Using the tool, information only
861
has to be defined once, and it is easy to compre- tool-based frameworks that feature extensions of
hend. Knowledge of implementation patterns is predefined BPM languages, editors and visualiza-
reused by automated transformations. Several tool tions to a specific enterprise. It would require less
utilities support the development process which investment and it would be easer for an enterprise
makes Danske Bank Workbench very efficient without experienced tool developers to customize
to work with. An empirical evaluation of the tool BPM tools instead of develop them from scratch.
confirmed this. Hence, we have confirmed the
hypothesis that was set up in the introduction,
which stated that applying the basic model driven rEFErENcEs
development principles of direct representation
and automation to BPM tools would solve many Bézivin, J., Hammoudi, S., Lopes, D., & Jouault,
of the experienced challenges. F. (2004). An experiment in mapping Web services
Danske Bank Workbench was not difficult to implementation platforms (Tech. Rep. LINA,
to build as many language workbenches exist University of Nantes).
for building metamodels, editors and transfor- Booch, G., Brown, A., Iyengar, S., Rumbaugh,
mations (though it did require deep insight in J., & Selic, B. (2004). An MDA Manifesto. Busi-
various Eclipse technologies and MDD concepts). ness Process Trends - MDA Journal. Retrieved
However, it has several limitations, and it only from http://www.bptrends.com/publication-
addresses a small subset of business processes that files/05-04%20COL%20IBM%20Manifesto%20
may be modeled. It requires much more effort to -%20Frankel%20-3.pdf
make it a production ready tool that can be used by
the organization. Despite a promising prototype, Bordbar, B., & Staikopoulos, A. (2004a). Mod-
our guess is that only a very limited number of elling and transforming the behavioural aspects
enterprises will go the way and implement their of web services. In Third Workshop in Software
own tools. While it may be economical beneficial Model Engineering (WiSME2004) at UML, Lis-
to develop ones own tools, there might be politi- bon, Portugal.
cal reasons not to do so. To answer the research Bordbar, B., & Staikopoulos, A. (2004b). on
question set up in the introduction we can now say, behavioural model transformation in Web ser-
“Defining and developing a model driven devices. In Conceptual Modelling for Advanced
velopment tool to support an enterprise specific Application Domain (eCOMO) (pp. 667-678).
business process development process seems Shanghai, China.
promising. It will heighten the productivity of
development teams and probably cause fewer BPEL. (2003). Business process execution lan-
errors in implementations. However, it requires guage for Web services (BPEL4WS). Version
a high degree of expertise in model driven devel- 1.1.Retrieved from http://www-128.ibm.com/
opment methodology and technology to develop developerworks/library/specification/wsbpel
such a tool. It will probably be unachievable for
Brahe, S. (2007). BPM on top of SOA: Experi-
most enterprises”
ences from the financial industry. In G. Alonso,
Although language workbenches provide
P. Dadam, & M. Rosemann (Eds.), BPM2007
huge support in development of model driven
(LNCS 4714, pp. 96-111).
development tools, it should be much easier to
customize ones own BPM languages and tools. For
future research we therefore suggest to work on
862
Brahe, S. (2008). An experiment on creating Frankel, D. S. (2003). Model driven architecture:

enterprise specific BPM languages and tools Applying MDA to enterprise computing. OMG
(Tech. Rep. ITU-TR-2008-102). IT University Press.
of Copenhagen.
GMF. (2008). Graphical Modeling Framework
Brahe, S., & Bordbar, B. (2006). A pattern-based project. Retrieved from http://www.eclipse.org/
approach to business process modeling and imple- gmf.
mentation in Web services. In D. Georgakopoulos
Guntama, E., Chang, E., Jayaratna, N., & Pudhota,
(Ed.), ICSOC 2006 (LNCS 4652, pp. 161-172).
L. (2003). Extension of activity diagrams for
Brahe, S., & Østerbye, K. (2006). Business process flexible business workflow modeling. Interna-
modeling: Defining domain specific modeling tional Journal of Computer Systems Science &
languages by use of UML profiles. In A. Rensink Engineering, 18(3), 137–152.
& J. Warmer (Eds.), ECMDA-FA 2006 (LNCS
Jablonski, S., & Bussler, C. (1996). Workflow
4066, pp. 241-255).
management - Modeling concepts, architecture
Budinsky, F., Steinberg, D., Merks, E., Ellersick, and implementation. London: Intl. Thomson
R., & Grose, T. J. (2003). Eclipse Modeling Frame- Computer Press.
work: A Developer’s Guide. Addison Wesley.
Jablonski, S., & Götz, M. (2007). Perspective
Dumas, M., & Hofstede, A. H. M. (2001). UML oriented business process visualization. In 3rd In-
activity diagrams as a workflow specification ternational Workshop on Business Process Design
language. In UML 2001 (LNCS 2185, pp. 76-90). (BPD) in conjunction with the 5th International
Conference on Business Process Management
Dumas, M., van der Aalst, W., & Hofstede, A.
(BPM 2007). Brisbane, Australia.
(2005). Process-aware information systems:
bridging people and software through process Koehler, J., Hauser, R., Kapoor, S., Wu, F. Y., &
technology. John Wiley & Sons, Inc. Kumaran, S. (2003). A Model-driven transfor-
mation method. In 7th International Enterprise
Eclipse (2008). The Eclipse project. Retrieved
Distributed Object Computing Conference (EDOC
from http://www.eclipse.org
2003) (pp. 186-197).
Erl, T. (2005). Service oriented architecture:
Koehler, J., Hauser, R., Sendall, S., & Wahler, M.
Concepts, technology and design. Prentice Hall.
(2005). Declarative techniques for model-driven
Flynn, D., Vagner, J., & Vecchio, O. D. (1995). business process integration. IBM Systems Jour-
Is CASE technology improving quality and nal, 44(1), 47–65.
productivity in software development? Lo-
Kroll, P., & Kruchten, P. (2003). The rational uni-
gistics Information Management, 8(2), 8–21.
fied process made easy. In A Practitioner’s Guide
doi:10.1108/09576059510084966
to the RUP. Addison Wesley.
Fowler, M. (2005). Language workbenches:
The killer-app for domain specific languages?
Retrieved from http://martinfowler.com/articles/
languageWorkbench.html.
863
Ledeczi, A., Maroti, M., Bakay, A., Karsai, G., Tolvanen, J.-P., & Rossi, M. (2003). MetaEdit+:
Garrett, J., Thomason, C., et al. (2001). The Defining and using domain-specific modeling
generic modeling environment. In Workshop on languages and code generators. In OOPSLA ’03:
Intelligent Signal Processing. Budapest, Hungary. Companion of the 18th annual ACM SIGPLAN
Retrieved from http://www.isis.vanderbilt.edu/ conference on Object-oriented programming,
Projects/gme/GME2000 Overview.pdf. systems, languages, and applications (pp. 92-93).
New York: ACM.
Leymann, F., & Roller, D. (2000). Production
workflow: Concepts and techniques. Prentice Hall. van der Aalst, W. M. P., Hofstede, A. H. M.,
Kiepuszewski, B., & Barros, A. P. (2003). Work-
List, B., & Korherr, B. (2005). A UML 2 profile
flow patterns. Distributed and Parallel Databases,
for business process modelling. In Perspectives
14(1), 5–51. doi:10.1023/A:1022883727209
in Conceptual Modeling, ER 2005 Workshops
(LNCS 3770, pp. 85-96). White, S. (2006). Business process modeling
notation (Version 1.0). Available at http://www.
MDAGuide. (2003). MDA Guide Version 1.0.1.
bpmn.org/Documents/OMG-02-01.pdf.
Retrieved from http://www.omg.org/docs/
omg/03-06-01.pdf. Windsor, J. (1986). Are automated tools changing
systems analysis and design? Journal of Systems
Mernik, M., Heering, J., & Sloane, A. M. (2005).
Management, 37(11), 28–33.
When and how to develop domain-specific lan-
guages. ACM Computing Surveys, 37(4), 316–344.
doi:10.1145/1118890.1118892
Murata, T. (1989). Petri nets: Properties, analysis
and applications. Proceedings of the IEEE, 77(4), BPM Tools: A collection of modeling and
541–580. doi:10.1109/5.24143 implementation tools specialized for modeling a
oAW. openArchitectureWare. Retrieved from business process and implement it as a workflow
http://www.openarchitectureware.org. in a workflow language.
Control Flow Behavior: A physical model of
Skogan, D., Grønmo, R., & Solheim, I. (2004). a business process. It specifies how the process
Web service composition in UML. In Eighth should be implemented in BPEL.
IEEE International Enterprise Distributed Object Domain Specific Language: (DSL) A special-
Computing Conference (EDOC’04) (pp. 47-57). ized programming or a modeling language that
Stahl, T., Völter, M., Bettin, J., Haase, A., & allows expressing solutions directly in concepts
Helsen, S. (2006). Model-driven software devel- of a problem domain.
opment: technology, engineering, management. Eventmap: A model of all business events that
Wiley. may occur in a given business context.
Model Driven Development: A development
Swithinbank, P., Chessell, M., Gardner, T., Grif- paradigm that focuses on using models in software
fin, C., Man, J., Wylie, H., & Yusuf, L. (2005). development. Models are used for analysis, simu-
Patterns: model-driven development using IBM lation, verification and code generation
rational software architect. IBM Redbooks. Avail- Model Transformation: A model transfor-
able at http://www.redbooks.ibm.com/abstracts/ mation takes one or several source models and
sg247105.html?Open. generates one or several target models, or tex-
864
tural documents. It is based on a transformation Workflow Specification: A document that

definition that specifies how to map elements in describes additional information required to imple-
the source DSLs to elements in the target DSLs ment a Solution Process Flow model in Danske
Solution Process Flow: A logical or conceptual Banks extended BPM system.
model of a business process. It specifies business
logic for one business event.
This work was previously published in Handbook of Research on Complex Dynamic Process Management: Techniques for
Adaptability in Turbulent Environments, edited by Minhong Wang and Zhaohao Sun, pp. 23-56, copyright 2010 by Information
Science Reference (an imprint of IGI Global).
865
866
Chapter 3.20
Semantic Business Process
Mining of SAP Transactions
Jon Espen Ingvaldsen
The Norwegian University of Science and Technology, Norway
Jon Atle Gulla

The Norwegian University of Science and Technology, Norway
AbstrAct executed transaction sequences with predefined

process hierarchies.
This chapter introduces semantic business process
mining of SAP transaction logs. SAP systems are
promising domains for semantic process mining INtrODUctION
as they contain transaction logs that are linked to
large amounts of structured data. A challenge with To describe the current situation in dynamic busi-
process mining these transaction logs is that the ness process environments we need tools that can
core of SAP systems was not originally designed assist rapid modeling. Process mining tools meet this
from the business process management perspective. requirement by extracting descriptive models from
The business process layer was added later without event logs in the underlying IT-systems to construct
full rearrangement of the system. As a result, sys- the business process descriptions from actual data.
tem logs produced by SAP are not process-based, SAP systems are promising domains for process
but transaction-based. This means that the system mining. SAP is the most widely used Enterprise
does not produce traces of process instances that Resource Planning (ERP) system with a total market
are needed for process mining. In this chapter, we share of 27 percent worldwide in 2006 (Pang, 2007).
show how data available in SAP systems can enrich Even though there may be blue print models defined
process instance logs with ontologically structured for how the systems should support organizational
concepts, and evaluate techniques for mapping business processes, there are often gaps between
how the systems are planned to be used and how
DOI: 10.4018/978-1-60566-669-3.ch017 the employees actually carry out the operations.
Semantic Business Process Mining of SAP Transactions
The magnitude of data sources in a running ERP Ontologies tend to grow huge and complex, and
system is large, and within SAP there are several both domain expertise and ontology modeling
event and transaction logs that can be analyzed expertise are needed in ontology engineering
with process mining. (Gulla, 2006). In the underlying databases of SAP
In this process mining work, we use transac- systems there are lots of structured data that can
tion data that describe document dependencies be extracted to form and populate ontologies. In
between executed transactions. A transaction in semantic business process mining of SAP trans-
a SAP system can be viewed a small application. actions, we can exploit available data structures
An example of a transaction is “ME51 – Create to limit the extent of ontology engineering work.
Purchase Requisition”. As the name indicates, One particular challenge with process mining
this transaction enables a user to create a purchase of SAP transactions is the many-to-many relation-
requisition. “ME51” is the unique identifier for ship between transactions and defined business
this transaction, called the transaction code. Such a processes. Figure 1 shows an example from
transaction would produce a purchase requisition, the business process hierarchy in SAP. In SAP
which further can be referred to by a purchase order systems, business processes are defined at four
created in another transaction, like “ME21 – Cre- levels, “Enterprise Area”, “Scenario”, “Group”
ate Purchase Order”. By tracing such document and “Business Process”. At the second lowest
dependencies, we are able to extract transaction level, Figure 1 shows two business processes,
sequences that can be explored and analyzed with “Subsequent debit for empties and returnable
use of process mining. packaging” and “Sales activity processing (stan-
Data in the underlying databases of SAP sys- dard)”. As shown in the hierarchy, both of these
tems contain business processes can involve the transaction
“V+01: Create Sales Call”. The transaction logs
• Transactional data – Daily operations, such in SAP systems contain no information about busi-
as sales orders and invoices. ness process context. If we do process mining on
• Master data – Business entities such as transaction logs where “V+01: Create Sales Call”
customers, vendors and users. occurs, there is no available data that explicitly
• Ontological data – Metadata for interpreta- states whether this transaction was carried out in
tion and structuring of instances. the context of “Subsequent debit for empties and
returnable packaging”, “Sales activity processing
The transactional data are the basis building (standard)” or another business process.
blocks for process mining analysis and describe Transaction sequences themselves can be used
events that are carried out. In the transactional as input to process mining algorithm to extract
data we typically find execution timestamps and flow models and performance indicators. How-
relations to involved master data sources. The ever, if we could map the executed transactions
ontological data in SAP databases can be used to precisely to concepts in the defined business
extract descriptions of the transactions and related process hierarchies, we would be able to extract
entities. For instance, in the SAP database there business process models with aggregated levels,
are table structures that contain full text descrip- and relate performance indicators to higher level
tions of transactions and business processes and process definitions.
their internal relationships. In this chapter we will show how transaction
Construction and maintenance of ontologies is sequences extracted from SAP systems can be
work-intensive and has so far been a bottleneck enriched with relations to ontological concepts
to realization of many semantical technologies. and we will evaluate three techniques for mapping
867
Figure 1. Example of business process hierarchies
executed transactions with the standard business Contemporary information systems (e.g.,
process hierarchies in SAP. WFM, ERP, CRM, SCM, and B2B systems)
record business events in so-called event logs.
Process mining takes these logs to discover
bAcKGrOUND process, control, data, organizational, and social
structures (van der Aalst, 2007). Within the BPM
In the past year, Semantic Web technology has life-cycle, process mining can be applied to gather
gained a substantial interest from Business Pro- knowledge about the past and find potentials for
cess Management (BPM) research. Traditional change and optimization.
process mining has successfully been shown to Approaches for semantic process mining have
extract flow models that describe how activities been proposed by incorporating ontologies, refer-
and organizational units depend on each other in ences from elements in event logs to ontological
dynamic business process environment. Semantic concepts and ontological reasoners. Ontologies
business process mining (SBPM) takes advantages define the set of shared concepts necessary for
of the rich knowledge expressed in ontologies the analysis, and formalize their relationships
and associated process instance data and extracts and properties. The references associate mean-
semantic models that enable reasoning in a wider ings to syntactical labels in event by pointing to
context than traditional process mining. SBPM defined ontology concepts. The reasoner supports
has been proposed as an extension of BPM with reasoning over the ontologies in order to derive
semantic web and semantic web service technolo- new knowledge, e.g., subsumption, equivalence,
gies in order to increase and enhance the level of etc. (Alves de Medeiros, 2008).
automation that can be achieved within the BPM Alves de Medeiros (2007) point out two im-
life-cycle (Alves de Medeiros, 2007; Ma 2007). portant potentials for leveraging process mining
868
with a conceptual layer: • Organizational structures: Employees,

roles, departments, controlling areas, etc.
1. To make use of the ontological annotations • Geographical structures
in logs and models to develop more robust • Material and product groupings
process mining techniques that analyze on • Business process compositions
conceptual levels.
2. Use process mining techniques to discover In some cases these structures can be utilized
or enhance ontologies based on the data in directly as complete domain ontologies, while in
the event logs. other cases they require manual processing before
they can be used as complete domain ontologies
Pedrinaci (2008) argue that business process or as input to populate general or incomplete
analysis activities require semantic information ontologies.
that spans business, process and IT levels and Extraction of long transaction sequences
is easily retrieved from event logs. They have from SAP systems is a task that involves large
developed the Events Ontology and the Process amounts of data and multiple database tables.
Mining Ontology that aim to support the analysis EVS ModelBuilder is a tool that is designed to
of enacted processes at different levels of abstrac- support extract entries from SAP systems that are
tion spanning from fine grain technical details to suitable for process mining. The tool is created
coarse grain aspects at the business level. in cooperation between Businesscape AS and
ProM is an open source process mining frame- the Information Systems group at the Norwegian
work that is built up of plug-ins that targets differ- University of Science and Technology. In EVS
ent process mining analysis (van der Aalst, 2007). ModelBuilder, the user can describe on a type-
Already, there are developed plug-ins and input level how business object and events are related.
format for ProM that targets ontological reasoning. Based on these descriptions, the program carry out
The Semantically Annotated Mining eXtensible necessary database queries, merge data sources
Markup Language (SA-MXML) format is a and exports transaction logs that can be processed
semantic annotated version of the MXML input and explored by analysis tools like ProM. More
format used by the ProM framework. In short, the details about how this tool can support the pre-
SA-MXML format opens for linking elements processing phase of process mining projects are
in logs to concepts in ontologies. The Semantic described in (Ingvaldsen, 2007).
LTL Checker is a ProM plug-in that allows for
semantic verification of properties in SA-MXML POPULAtING ONtOLOGIEs FrOM
logs (Alves de Medeiros, 2008). AVAILAbLE DAtA strUctUrEs
We will use a simple example to show the mag-

EXtrActION OF sEMANtIcALLy nitude of structured context information available
ENrIcHED trANsActION in the transaction logs in SAP systems. Figure 2
sEQUENcEs shows three SAP tables that can be involved in
a semantic process mining project. The EKKO
SAP systems contain data structures (typically table contains data describing purchase order
hierarchical) that can be exported to form and headers. Entries in this table contain references
populate ontologies. Such data structures include: to a vendor and the user that created the purchase
order. Both the user and the vendor are described
869
Figure 2. Three SAP tables that are used to describe events related to the creation of purchase orders.
more in detail in separate database tables. USR03 human knowledge is used to structure and place
is one of the tables that describe SAP users in different concepts into hierarchical levels, but the
detail. It contains the full name of the user and population of ontological concepts are done by
describes which department the user belongs to. use of data available in the SAP tables.
LFA1 include vendor details such as their location, In real systems, the three tables in figure 2
specified by country, district and city. involve many more attributes that describe the
Based on data found in the EKKO table, we entity properties. To limit the extent of this ex-
are able to extract events for creations of purchase ample, we have only involved a subset of avail-
orders. Data in the two other tables can be used to able attributes. By including other attributes and
populate two distinct ontologies; one describing related tables into this simple process mining
the breakdown of the company into departments example, we can enrich the event further with
and employees, and one describing how coun- context details and link the event to other events
tries, districts, cities and vendors relate to each like creation of purchase requisition and receipt
other. Figure 3 shows how our example data can of order confirmations and goods.
describe events in detail and relate them to struc- In SA-MXML, the event information and on-
tured ontological data. In both of the ontologies, tologies are is stored in separate files. The event
Figure 3. An example of a “Create purchase order” event that is related to two ontological concepts
870
log files describe the process instances that are possibilities within its transactions, where detailed
carried out, the work items they consist of, rela- functions are carried out. Examples of business
tions to aggregated process definitions, involved scenarios, process groups and processes for the
users, and execution timestamps. These elements sales enterprise area are shown in figure 1.
can further point to ontological concepts by use In this paper we will focus on the sales enter-
of Uniform Resource Identifiers (URIs) . prise area for evaluating techniques for mapping
entries in transaction logs to defined business
MAPPING EXEcUtED processes. The sales enterprise area deals with the
trANsActIONs tO DEFINED tasks of performance utilization and thus organizes
bUsINEss PrOcEssEs the business relationships in the market. The task of
this enterprise area is to provide customers with the
Much process mining work assumes that event goods produced in the enterprise, or with financial
logs extracted from the IT systems contain rela- or other services offered by the enterprise. This
tions between the executed events and higher level includes the planning and control of distribution
business process definitions. Such information is channels, from advertisement, inquiry and quota-
very helpful if we want to do delta analysis. That tion processing, sales order processing, delivery
is, comparing the actual processes with predefined processing, invoicing, and down to checking of
process models. By comparing the predefined incoming payments (Keller, 1998).
process models with models discovered by use A process can be carried out by completing
of process mining, discrepancies between them one or more transactions. Figure 4a shows how
can be detected and used to improve the processes many processes transactions are involved in within
(van der Aalst, 2005). the enterprise area “sales”. Most transactions are
By linking entries in SAP transaction logs to only involved in one single process, and in aver-
aggregated business process definitions we can age a transaction is involved in 1.55 processes.
lift the models and performance indicators that we This means that for very many of the transactions
discover from process mining from a somewhat in a transaction log we can identify the correct
system technical level to higher business levels. process easily.
Predefined business process hierarchies are On the other hand, there are a significant num-
available and serialized in database tables in SAP ber of transactions that occur in many processes.
systems. As shown in Figure 1, this hierarchical For the enterprise area “sales” some transactions
structure consists of four levels. An enterprise area occur in up to eight processes. These transactions
depicts a business structure, which represent a can be seen as the most common transactions, and
homogeneous unit in the sense of process-oriented these are the transactions we find most frequently
structuring. Examples of enterprise process areas in the transaction logs. For such transactions, it
are enterprise planning, production and sales. is much more challenging to identify the correct
Business scenarios are assigned to a particular process for a given execution context.
enterprise process area, and describe on an ab- Figure 4b shows that the number of transactions
stract level the logical business flow across dif- a process can involve vary significantly. In total
ferent application areas involving the processes. the sales enterprise area consists of 59 standard
Process groups contain individual processes that processes, and in average a process involves
are bundled such that they can be visualized 10.92 transactions. Most processes contain only
more easily. Processes describe the smallest self- a handful of transactions, but there are processes
contained business sequences and represent the that can involve close to a hundred transactions.
871
Figure 4. a) Distribution of process occurrences per transaction within sales. b) Number of transaction
per process
Mapping Approaches Simple lookup, search and indexing of processes,

and graph operations.
A transaction sequence is a list of executed and Simple lookup is the naïve approach of
related transactions that are ordered by their ex- searching for processes that contains the transac-
ecution timestamp. They are related in the sense tion execution we want to enrich with business
that they produce and consume the same set of process context. In cases where several processes
resources. I.e., an execution of the transaction contain the given transaction the one with the least
“ME51, Create Purchase Requisition” produces number of transactions is selected as the correct
a purchase requisition that can be consumed business process context. If we use simple lookup
(referred to) by the transaction “ME21, Create to identify the business process context for the
Purchase Order”. transaction “VL01: Create Delivery”, we would
Figure 5 shows a transaction sequence with search for all processes where this transaction oc-
seven entries and a hierarchy with defined busi- curs. Such a query would result in the following
ness processes. From the transaction logs in SAP process candidates from the sales enterprise are:
systems, we can extract transaction sequences as “Batch search strategy processing (standard)”,
shown at the bottom of the figure, but the mapping “Delivery for returnable packaging subsequent
to business process definition is not explicitly debit”,” Delivery for returns”, and ” Delivery
stated in the data. processing”. Here, the first process candidate
Transaction sequences typically span multiple would be as this contains the least number of
process boundaries, and therefore each transaction transactions.
in the sequence must be mapped to the correct Search and indexing of processes is a slightly
business process context. more sophisticated approach where the defined
Although the relations between transaction processes are indexed by their involved transac-
executions and defined business processes are not tions in a search index. Then, instead of just using
explicitly stated in the transaction logs there are a single transaction to form a search query, we
approaches for identifying likely mappings. We also include the other entries of the transaction
will propose and evaluate three such approaches: sequence. The idea behind this approach is that by
872
Figure 5. A transaction sequence where each entry is mapped to a defined business process.
incorporating neighbor transactions of a transac- transaction codes that are included to describe
tion we have more information about the execu- the specific execution context more in detail. As
tion context and are more likely of identifying a the last two transactions are only present in the
unique and correct business process definition. process “Sales activity processing (standard)” the
Figure 1, shows an example of two processes that other process, “Subsequent debit for empties and
both include the transaction “V+01: Create Sales returnable packaging”, would not be considered as
Call”. If we have a transaction sequence where this a candidate process for this execution of “V+01:
transaction, “VC02: Change Sales Activity” and Create Sales Call”.
“VC03: Display Sales Activity” occurs together, Graph operations are an alternative approach
we can include all the three transactions to form that views the set of candidate solutions as a
a query and retrieve a business process context. graph with nodes and edges. The problem that
In a Lucene1 search environment, such a query motivated this approach is that several standard
could be expressed as +”V+01” “VC02” “VC03”. SAP processes contains exactly the same set of
The plus sign states that the following transaction potential transactions. For instance, “Master trans-
code is required to exist in the result set entries, fer for contact documents”, “Message transfer for
and this plus sign is used in front of the transaction billing documents”, “Message transfer for sales
code that we want to map with a business process documents”, and ”Message transfer for supplies”
definition. The other query terms express optional are all processes that involve exactly the same set
873
of transactions: “VL14: Mail control decentral- Evaluation

ized shipping”, “VL20: Display Communication
Document”, “VL70: Output From Picking Lists”, To evaluate the three approaches we constructed
“VPAK: Packing list”, and “VT70: Output for ten transaction sequences spanning alternative
Shipments”. Neither simple lookup or search and routes through the following sales processes: Sales
indexing of processes would be able to distin- Quotation Processing, Sales Order Processing,
guish and select one such process as they contain Delivery Processing, Goods Issue Processing, and
exactly the same set of transactions. However, Billing Processing. The sequence between these
in longer transaction sequence typically spans processes and the transactions they can involve
several processes, and a process, like “Master are shown in figure 6. This set of possible transac-
transfer for contact documents”, is often just one tions is used together with process definitions in
of several processes. This process is more likely (Keller, 1998) to construct transaction sequences
to be the correct process if the other transactions for testing that are realistic and annotated with
in the same transaction sequence are within the correct business process context.
same process group, process scenario or at least In total 12 test transaction sequences was con-
enterprise area. For each entry in a transaction structed. To make them realistic with respect to
sequence, this approach uses simple lookup to typical process mining logs, the constructed trans-
retrieve sets of solution candidates. Then, Dijkstras action sequences have a lot of variation in length,
shortest path algorithm is used find the shortest and some are incomplete with respect to describe
distance through the business process hierarchy end-to-end processes. In process mining, extracted
between the set of process candidates and the transaction logs typically contain transactions
transactions in the actual transaction sequence. carried out within a certain time frame. Entries
For a given transaction, the business process that are close to the start or end of are frequently
with the lowest average distance to all entries in incomplete as only parts of their history overlaps
the actual transaction sequence is selected as the with the selected time frame. Cancelled processes
right candidate. lead also to incomplete transaction sequences.
Figure 6. Typical sales processes and their transactions. The process sequences are read from left to
right.
874
Figure 7. Test set of ten transaction sequences where each entry is labeled with a correct business
process.
The test transaction sequences and their business indexing of processes identify the correct mapping
process annotations are shown in Figure 7. in 45% of the test cases. For simple lookup and
Figure 8 shows how to which extent each ap- graph operations the average scores are 42% and
proach is able to identify the correct mapping for 31% respectively. In four of the test cases they
the twelve test sequences. In average search and have the same score.
Figure 8. Percentage scores for how many correct business process mappings the three approaches are
able to identify. The scores are shown as bars for each of the 12 tests.
875
The graph operations approach has the highest FUtUrE trENDs

deviation in its overall performance. In several of
the test sequences this approach identifies none In the last years there has been a consolidation
of the mappings correctly. A reason for this might between ERP and Business Intelligence (BI) sys-
be that this approach can fall into local areas of tems. The largest ERP vendors, SAP and Oracle,
the business process hierarchy where the short- have acquired the two BI vendors Business Ob-
est path criterion is optimized, but the identified jects and Hyperion and integrate their solutions
mappings are distant from the correct solution. into their customer offerings. Now, ERP does not
Search and process indexing outperforms the only provide a shared data source for various or-
simple lookup approach in two of the test cases. In ganizational units, but also a valuable data source
both of these the search and process indexing finds that companies can utilize to extract knowledge
the correct mapping because the query contains and competitive advantages.
context information that enable the system to limit Software vendors that focus on supporting the
the solution space and eliminate false candidates. Business Process Management (BPM) lifecycle,
All of the three approaches could be modified like IDS Scheer, have already shipped products
and tuned to improve the performance. In the that borrow elements from process mining. The
evaluations all approaches describe the context ARIS Process Performance Manager enables
by including a varying number of transaction companies to relate performance indicators to real
codes. These descriptions could also be extended business flows. As ERP solutions also are moving
with information like involved documents, users, into directions of Service Oriented Architectures
vendors, products, geographical locations, etc. (SOA) and (BPM), we believe that future busi-
As we get a better and more complete picture of ness applications in the ERP and BI area will
context around the execution of a transaction we focus more on analysis of business flows. This
can limit our solution space further and increase will create need for information systems that
the probability of identifying correct mappings. create event logs that contains a lot of structured
However, such systems would also require manual context information and relations to higher level
labeling of training sets or search indices where process definitions.
such context information is related to the defined In this paper we have focused on showing
business process hierarchies. how elements in the event logs can be linked to
Another alternative for improving the mapping ontological concepts. In addition, event logs can
between transaction executions in the transaction also be enriched with numerical and date attributes.
logs with defined business processes is to com- For a purchase process, events can be enriched
bine the method. Search and process indexing with order amounts, values, expected delivery
or simple lookup could be chosen to the main dates, and so on. With such information available
strategy for suggesting the right business process, in the event logs we can not only describe how
and in those cases where the business process dynamic business processes are executed, how the
candidates contain the same set of transactions loads are distributed and where most of the time
the graph operations approach could be applied is consumed, but also to which extent the actual
to make a suggestion. processes are meeting expectations.
As shown throughout this paper the amount
of information that can be related to event logs
is enormous, and there are large potentials for
merging elements from process mining, data
876
mining and ontological reasoning. Process min- concepts. As the reusability of such enterprise
ing is used to find out how people and systems specific ontologies are low, tools that utilize
work. Ontological reasoning provides answers available data to assist or automate parts of the
to test hypotheses, and data mining can extract ontology engineering process is of great value.
descriptive and predictive models that support In this paper we have shown how available
the whole picture. data structures in SAP systems can be utilized to
If the core of ERP systems was originally populate ontologies and construct transaction logs
designed for business process management and for semantic business process mining. We have
monitoring of process instances we would not in particular focused on the challenge of making
need techniques to identify mappings between use of the business process hierarchy definitions
historical transaction executions and defined available in SAP systems. Three approaches for
business processes. Traces of process instance mapping entries in transaction sequences with pre-
information in transaction logs would create defined business process hierarchies are evaluated,
valuable opportunities for using process mining and the results from these evaluations shows that
to describe the real business flow and measuring it is difficult to completely automate the process
the deviations against business process definitions of identify such mappings. By including a more
and procedures. However, as large software ven- complete information picture around the execution
dors needs to be backwards compatible with older of transactions, we can limit the solution spaces
systems and customers, it is difficult to do major and increase the probability of identifying correct
modification of the kernel design. Until traces of business process mappings.
process instance information are provided in the Correct mappings between transaction log
event log structures, a process miner needs tools entries and defined business processes, enable
that can assist the mapping of transaction execu- process mining techniques to construct models
tions and defined business processes. that are lifted from a somewhat system technical
transaction flow focus up to aggregated levels that
describe higher level business terms. This makes
cONcLUsION process mining models valuable both for IT and
business people.
Transaction logs in SAP systems contains sub-
stantial amount of context information that can
be utilized to create references from the execu- rEFErENcEs
tion instances to ontological concepts. Also, SAP
databases contain ontological data that describe Alves de Medeiros, A. K., Pedrinaci, C., van der
relations between concepts involved in transaction Aalst, W. M. P., Domingue, J., Song, M., Rozinat,
executions. This availability of context informa- A., et al. (2007). An outlook on semantic business
tion and ontologically structured concepts reduce process mining and monitoring. In OTM 2007
manual ontology engineering work and make SAP Workshops (pp. 1244-1255). New York: Springer.
systems a promising arena for semantic process Alves de Medeiros, A. K., van der Aalst, W. M.
mining. P., & Pedrinaci, C. (2008). Semantic process min-
General ontologies, like the Process Mining ing tools: Core building blocks. Paper presented
Ontology, are important for semantic interpreta- at the 16th European Conference on Information
tion of common process mining terms. However, Systems, Galway, Ireland.
customized domain ontologies are also important
for being able to reason over enterprise specific
877
Gulla, J. A., Borch, H. O., & Ingvaldsen, J. E. van der Aalst, W. M. P., & Weijters, A. J. M. M.
(2006). Unsupervised keyphrase extraction for (2005). Process mining. In M. Dumas, W.M.P. van
search ontologies. In Natural Language Process- der Aalst & A.H.M ter Hofstede (Ed.), Process
ing and Information Systems (pp. 25-36). New aware information systems (pp. 235-255). Wiley
York: Springer. Interscience
Ingvaldsen, J. E., & Gulla, J. A. (2007). Prepro-
cessing support for large scale process mining of
SAP transactions. In Business process manage- KEy tErMs AND DEFINItIONs
ment workshops (pp. 30-41). New York: Springer.
Process Mining: Research area that aims at
Keller, G., & Teufel, T. (1998). SAP R/3 process creating tools for discovering process, control,
oriented implementation. Reading, MA: Addison- data, organizational and social structures from
Wesley. event logs.
Ma, Z., Wetzstein, B., Heymans, S., & Anicic, D. SAP Systems: ERP solutions delivered by
(2007). Semantic business process repository. In SAP for large organizations.
Proceedings of the International Workshop on Semantic Business Process Management
Semantic Business Process Management (SBPM (SBPM): An extension of BPM with semantic web
2007), CEUR Proceedings. and semantic web service technologies in order
to increase and enhance the level of automation
Pang, C., Eschinger, C., Dharmasthira, Y., & Mo- that can be achieved within the BPM life-cycle
toyoshi, K. (2007). Market share: ERP software, Transaction: A small application within SAP
worldwide, 2006. Gartner report. systems that have a unique transaction code.
Pedrinaci, C., & Domingue, J. (2007).Towards an Transaction Log: Data source in a transaction
ontology for process monitoring and mining. In based information system that describes histori-
Proceedings of the Workshop on Semantic Busi- cal events.
ness Process and Product Lifecycle Management Transaction Sequence: An ordered chain of
(SBPM-2007), CEUR-WS events that describe transaction carried out, how
they depend on each, when they were executed,
van der Aalst, W. M. P., Reijers, H. A., Weijters, and relations to involved entities like users, ven-
A. J. M. M., van Dongen, B. F., Alves de Me- dors, products, etc.
deiros, A. K., Song, M., & Verbeek, H. M. W.
(2007). Business process mining: An industrial
application. Information Systems, 32(1), 713–732. ENDNOtE
doi:10.1016/j.is.2006.05.003
van der Aalst, W. M. P., van Dongen, B. F., Gün-
1
Lucene is an open source information
ther, C. W., Mans, R. S., Alves de Medeiros, A. K., retrieval library, supported by the Apache
Rozinat, A., et al. Verbeek, H. M. W., & Weijters, Software Foundation. See: http://lucene.
A. J. M. M. (2007). ProM 4.0: Comprehensive apache.org/
support for real process analysis. In Petri Nets
and Other Models of Concurrency – ICATPN
2007 (pp. 484-494). New York: Springer.
Adaptability in Turbulent Environments, edited by Minhong Wang and Zhaohao Sun, pp. 416-429, copyright 2010 by Informa-
tion Science Reference (an imprint of IGI Global).
878
879
Chapter 3.21
Mining Association Rules
from XML Documents
Laura Irina Rusu
La Trobe University, Australia
Wenny Rahayu
La Trobe University, Australia
David Taniar
Monash University, Australia
Abstract be used in various ways, for example, to decide

on a business issue or to make predictions about
This chapter presents some of the existing mining future e-customer behaviour in a Web application.
techniques for extracting association rules out of One type of knowledge that can be discovered in
XML documents in the context of rapid changes in a collection of XML documents relates to asso-
the Web knowledge discovery area. The initiative ciation rules between parts of the document, and
of this study was driven by the fast emergence of this chapter presents some of the top techniques
XML (eXtensible Markup Language) as a standard for extracting them.
language for representing semistructured data and
as a new standard of exchanging information be-
tween different applications. The data exchanged Introduction
as XML documents become richer and richer every
day, so the necessity to not only store these large The amount of data stored in XML (eXtensible
volumes of XML data for later use, but to mine Markup Language) format or changed between
them as well to discover interesting information fferent types of applications has been growing
has became obvious. The hidden knowledge can during the last few years, and more companies
are considering XML now as a possible solution
DOI: 10.4018/978-1-60566-330-2.ch011 for their data-storage and data-exchange needs
Mining Association Rules from XML Documents
(Laurent, Denilson, & Pierangelo, 2003). The concerned with analysing texts within the XML
first immediate problem for the researchers was document, structural clarification is concerned
how to represent the data contained in the old with determining similar documents based on
relational databases using this new format, so their content (Nayak, 2005; Nayak et al., 2002).
various techniques and methodologies have been Discovering association rules is looking for
developed to solve this problem. Next, the users those interesting relationships between elements
realised that they not only required storing the data appearing together in the XML document, which
in a different way, which made it much easier to can be used to predict future behaviour of the
exchange data between various applications, but document. To our knowledge, this chapter is the
they required getting interesting knowledge out first work that aims to put together and study the
of the entire volume of XML data stored as well. existing techniques to perform the mining of as-
The acquired knowledge might be successfully sociation rules out from XML documents.
used in the decisional process to improve business
outcomes. As a result, the need for developing new
languages, tools, and algorithms to effectively Background
manage and mine collections of XML documents
became imperative. The starting point in developing algorithms and
A large volume of work has been developed, methodologies for mining XML documents was,
and research is still pursued to get solutions that naturally, the existing work done in the relational
are as effective as possible. The general idea and database mining area (Agrawal, Imielinski, &
goal for researchers is to discover more power- Swami, 1993; Agrawal & Srikant, 1998; Ashrafi,
ful XML mining algorithms that are able to find Taniar, & Smith, 2005; Ashrafi, 2004; Daly & Ta-
representative patterns in the data, achieve higher niar, 2004; Tjioe & Taniar, 2005). In their attempt
accuracy, and be more scalable on large sets of to apply various relational mining algorithms to
documents. The privacy issue in knowledge dis- the XML documents, researchers discovered that
covery is also a subject of great interest (Ashrafi, the approach could be a useful solution for min-
Taniar, & Smith, 2004a). ing small and not very complex XML documents,
XML mining includes both the mining of but not an efficient approach for mining large and
structures as well as the mining of content from complex documents with many levels of nesting.
XML documents (Nayak, 2005; Nayak, Witt, & The XML format comes with the acclaimed
Tonev, 2002). The mining of structure is seen extensibility that allows the change of structure,
as essentially mining the XML schema, and it that is, adding, removing, and renaming nodes in
includes intrastructure mining (concerned with the document according to the information nec-
mining the structure inside an XML document, essary to be encoded in. Furthermore, using the
where tasks of classification, clustering, or as- XML representation, there are a lot of possibilities
sociation rule discovering could be applied) and to express the same information (see Figure 1 for
interstructure mining (concerned with mining the an example) not only between different XML
structures between XML documents, where the documents, but inside the same document as well
applicable tasks could be clustering schemas and (Rusu, Rahayu, & Taniar, 2005a).
defining hierarchies of schemas on the Web, and In a relational database, it is not efficient to
classification is applied with name spaces and have multiple tables to represent the same data
URIs [uniform resource identifiers]). The min- with different field names, types, and relationships
ing of content consists of content analysis and as the constraints and table structures are defined
structure clarification. While content analysis is at the design time. In an opposite manner, a new
880
Figure 1. Different formats to express the same information using the XML structure
XML document can be added to a collection of Overview of the Generic

existing XML documents even though it represents Association Rule Concepts
the same type of data using a totally different
structure and element names, that is, a different The concept of association rules was first intro-
XML schema. As a result, researchers concluded duced by Agrawal (1993) for relational-database
that the logic of the relational mining techniques data to determine interesting rules that could be
could be maintained, but they needed to assure extracted from some data in a market basket
that the steps of the existing algorithms were analysis. The algorithm is known as the Apriori
looking to the specific characteristics of the XML algorithm, and an example of an association rule
documents. extracted could be “If the user buys the product A,
Among other XML mining methods, asso- he or she will buy the product B as well, and this
ciation rule discovery, and the classification and happens in more than 80% of transactions.” The
clustering of XML documents have been the most generic terms and concepts related to the Apriori
studied as they have a high degree of usability algorithm are as follows. If I represents the set
in common user tasks or in Web applications. In of distinct items that need to be mined, let D be
this chapter, we present a number of techniques the set of transactions, where each transaction T
for mining association rules out of XML docu- from D is a set of distinct items T ⊆ I. An asso-
ments. We chose to analyse this particular type of ciation rule R is an implication X→Y, where X,
mining because (a) it is, in our opinion, the most Y ⊂ I and X ∩ Y = ∅. The rule R has the support
useful for general types of applications in which s in D if s% of transactions in D contain both X
the user just wants to find interesting relation- and Y, and the confidence c if c% of transactions
ships in his or her data and wants help to make in D that contain X also contain Y. If we use a
better business decisions, and (b) the techniques freq(X,D) function to calculate the percentage of
used are easy to understand, replicate, and apply transactions in D that contain X, the support and
for the common user who does not have a high confidence for the association rule R could be
degree of knowledge of mathematics or statistics, written as the following formulas:Support (X→Y)
often required by some techniques for performing = freq (XUY, D) and Confidence (X→Y) = freq
classification or clustering. (XUY, D) / freq (X, D).
The minimum support and minimum confi-
dence are set at the beginning of the mining pro-
881
cess, and it is compulsory that they are observed purchases in a store for March 2005 and June 2005
by the determined rules. In the Apriori algorithm, might come as two separate static XML documents
all the large k-itemsets are determined, start- if the business process stores the data at the end of
ing from k=1 (itemsets with only one item) and each month). Dynamic XML documents contain
looping through D (the set of all transactions) to data that are continuously changing in time (an
calculate its support and the confidence. If they online bookstore, for example, will change its
are not validated against the minimum required, content, represented as an XML document, from
the k-itemset is considered to be not large and is one day to another, or even multiple times dur-
pruned. The algorithm assumes that any subset ing the same day depending on the e-customers’
of items that is not large determines its parent behaviour).
(i.e., the itemset that contains it) to not be large, Most of the work done in the area of mining
and this improves the speed of the process a lot. association rules from static XML documents use
At the end, when all the large itemsets are found, classical algorithms based on the Apriori algo-
association rules are determined from the set of rithm, described before in the overview section,
large itemsets. while a number of non-Apriori-based approaches
have been developed as well. In this chapter we
Overview of XML Association Rules will analyse at least one of each type of algorithms.
In case of dynamic XML documents, the fo-
For the XML documents, finding association cus is on mining association rules out of historic
rules means finding relationships between simple versions of the documents or out of the effective
or complex elements in the document: in other set of changes extracted between two successive
words, finding relationships between substructures versions. The difference between two versions
of the XML document. For example, in an XML of the same XML document is named delta, and
document containing details of the staff members it can be (a) structural delta, when the difference
and students in a computer-science university between versions is done at the schema level, or (b)
department, including details of their research content delta, when the difference is calculated at
publications, an association rule could be “Those the content level (Chen, Browmick, & Chia, 2004).
staff members who publish their papers with X
publisher received an award, and this happens
in 75% of cases.” Later in the chapter (see the Discovering Association
section on Apriori-based approaches), we give Rules from Static
some examples of how the generic concepts of XML Documents
transaction and item are perceived by the XML
association rules. We will also show how the con- As specified in the background section, some
cepts of support and confidence are used by the of the XML association rule mining techniques
presented approaches as they need to be correct use the Apriori general algorithm (Agrawal et
with regard to the total number of XML transac- al., 1993; Agrawal & Srikant, 1998) as a starting
tions that need to be mined. point for developing new methodologies specific
Our analysis is split in two subsections based to the XML document format and extensibility,
on the type of XML documents mined, that is, while completely different techniques have been
(a) static XML documents and (b) dynamic XML developed as well. The following analysis is
documents. Static XML documents contain data split in two subsections depending on the type of
gathered for a specific period of time that do not mining algorithm used, that is, (a) Apriori-based
change their content (for example, details about approaches and (b) non-Apriori-based approaches.
882
Apriori-Based Approaches is viewed as a record relative to the other records

at the same depth or with similar tags.
A first thing to do is to see how the generic con- A simple and direct method to mine association
cepts related to the association rules (mentioned rules from an XML document by using XQuery
in the previous section), that is, transactions and (XQuery, 2005) was proposed by Wan and Dob-
items, are mapped to the particular XML format. bie (2003, 2004). Based on the fact that XQuery
Even though most of the papers detailed further in was introduced by W3C (World Wide Web Con-
the chapter (Braga, Campi, & Ceri, 2003; Braga, sortium) to enable XML data extraction and
Campi, Klemettinen, & Lanzi, 2002; Braga, manipulation, the algorithm is actually an imple-
Campi, Ceri et al., 2002; Wan & Dobbie, 2003, mentation of the Apriori algorithm’s phases using
2004) do not give certain definitions for these the XQuery Language. In Figure 2, we exem-
concepts, we can determine their view on the plify the algorithm on an XML document contain-
matter by analysing the algorithms. If an XML ing information about items purchased in a
document is seen as a tree (see the example in number of transactions in a store (Figure 2a). The
Figure 3), the set of transactions D will be a list algorithm loops through the XML document,
of complex nodes formed by querying the XML generates the large itemsets in the “large.xml”
document for a specific path, a single complex document (Figure 2b), and then builds the asso-
node will form a transaction, and the children of ciation rule document (Figure 2c). For details on
the transaction node will be the items. The main the XQuery code implementation of the apriori
difference from the generic concepts is that, while a function and the other functions involved, we
generic transaction contains only a limited number refer the reader to the original papers (Wan &
of items and is easier to quantify, one XML tree Dobbie, 2003, 2004).
transaction can have a different number of items The significance of this approach is that the
depending on the level of nesting of the document. authors demonstrated for the first time that XML
A similar definition is given in Ding, Ricords, and data can be mined directly without the necessity
Lumpkin (2003), but at a more general level; that of preprocessing the document (for example,
is, all the nesting depths (paths) in an XML docu- mapping it to another format, such as a rela-
ment are considered to be records starting with the tional table, which would be easier to mine). The
root, so for any node in the document, each child algorithm could work very well in case of XML
documents with a very simple structure (as in our
Figure 2. Example of a direct association-rule mining algorithm using XQuery (Wan & Dobbie, 2003,
2004)
883
Figure 3. Example of an XML document presented as a tree (research.xml) with the identified context,
body, and head
example in Figure 2), but it is not very efficient Preprocessing Phase

for complex documents. Also, a major drawback,
assumed by the authors, is that in the XQuery At this stage, a lot of operations are done to pre-
implementation, the first part of the algorithm, pare the XML document for extracting association
that is, discovering large itemsets (Figure 2b), is rules. In the following, we discuss some important
more expensive regarding time and processor terms and concepts appearing during this step,
performance than in other language implementa- noting that this phase is the most extended one
tions (e.g., in C++). This drawback is explained because a proper identification of all the aspects
by the lack of update operations in XQuery: a involved in mining preparation will significantly
large number of loops through the document is reduce the amount of work during the other two
required in order to calculate the large itemsets. phases (extracting and postprocessing rules).
However, the algorithms promise a high speed The concept of the context of the association
when the update operations are finally imple- rules refers to the part(s) of the XML documents
mented in XQuery. that will be mined (similar to the generic concept
Other methodologies for discovering associa- of a set of transactions). Sometimes, we do not
tion rules from XML documents are proposed by want to mine all of the information contained in an
Braga et al. (2003) and Braga, Campi, Ceri et al. XML document, but only a part of it. For example,
(2002); they are also based on the Apriori algo- in an XML document containing university staff
rithm as a starting point and mine the association and student information (see Figure 3), we may
rules in three major steps, that is, (a) preprocess- want to find association rules among people ap-
ing data, (b) extracting association rules, and (c) pearing as coauthors. In this case, the identified
postprocessing association rules. In our opinion, context includes the multitude of nodes relating
due to the specific XML format, when many lev- to publications, no matter if they belong to PhD
els of nesting could appear inside of a document, students or professors. This means the algorithm
simple loops and counts (as in Figure 2) are no will not consider the <PhD_courses> nodes or
longer possible, so the three-step approach seems <Personal_info> nodes as they are not relevant
to be more appropriate for mining various types to the proposed rules to discover.
of XML documents. Context selection refers to the user’s oppor-
tunity to define constraints on the set of transac-
884
tions D relevant to the mining problem (Braga, The working document is defined in the first
Campi, Ceri et al., 2002). Referring again to our line, then the context, body, and head areas are
example (Figure 3), we may want to look for as- defined together with the minimum support and
sociation rules considering all the authors in the minimum confidence required for the rules. The
document, but only for publications after the year WHERE clause allows constraint specification;
2000, so a constraint needs to be defined on the in this example, only publications after 2000 will
“year” attribute of each publication element (not be included in the context of the operator.
visible in the graph, but existing in the original The XMINE RULE operator brings some
XML document). improvements, which could not be solved by the
If we talk about an association rule as an impli- direct association rule mining algorithm in one
cation X→ Y, X is the body of the rule and Y is the step which uses XQuery, described at the begin-
head of the association rule. The body and head ning of the section, as follows:
are always defined with respect to the context of
the rule as the support and the confidence will be •)>> The context, body, and head of the operator
calculated and relevant only with respect to the can be as wide as necessary by specifying
established context. In the XML association rule multiple areas of interest for them as parts
case, the body and the head will be, in fact, two of the XML document or even from differ-
different lists of nodes, that is, substructures of the ent XML documents.
context list of nodes; only nodes from these two •)>> When specifying the context, body, and
lists will be considered to compose valid XML head segments, a variable can be added to
association rules. take some specific values that enhance the
We exemplify the above described concepts, context selection facility.
that is, context identification, context selection, •)>> A GROUP clause can be added to allow the
and the head and body of the rules, by using the restructuring of the source data.
XMINE RULE operator (Braga et al., 2003) on
the working example in Figure 3, that is, the We exemplify how the first feature can be
“research.xml” document. implemented using the same working example,
We visually identify the mentioned concepts that is, the “research.xml” document. Suppose we
in Figure 4, which details the algorithm proposed now want to determine rules between publishers
by Braga et al. (2003) and Braga, Campi, Ceri, and keywords, that is, to find which publishing
et al. (2002). companies are focusing on specific areas of re-
Figure 4. Mining association rules from an XML document using the XMINE RULE syntax
885
Figure 5. The syntax of the XMINE RULE operator introduced by Braga et al. (2003)
search. See Figure 5 for a visual representation al., 2003) because it limits from the beginning the
of the new body and head selections. possibility to find and extract other rules (involv-
The main difference from the one-step mining ing other parts of the documents).
approach (Wan & Dobbie, 2003, 2004) is that the
three-step algorithm (Braga et al., 2003; Braga, Extracting Association Rules
Campi, Ceri et al., 2002; Braga, Campi, Klemet-
tinen et al., Klemettinen, 2002) does not work For the one-step mining approach, Figure 2
directly on the XML document all the way down exemplifies the XQuery implementation of the
to the phase of extracting the association rules; generic Apriori algorithm. It mainly performs the
instead, the first phase, that is, preprocessing, has following steps. Starting from the 1-itemsets (i.e.,
as a final output a relational binary table (R). The itemsets with one single item), a k-itemset (k>1)
table is built as follows (the authors suggest the is built by extending the (k-1)-itemset with a new
use of the Xalan, 2005, as an XPath interpreter item. For each itemset, the support is calculated as
in the actual implementation): (a) The fragments a percentage of the total number of transactions that
of the XML document specified in the context, contain all the items of the itemset. If the itemset
body, and head are extracted and filtered by apis not frequent (large) enough (i.e., its support is
plying the constraints in the WHERE clause (in less than the minimum support required), it will
case one exists), (b) the XML fragments obtained be removed (pruned), and the algorithm continues
by filtering the body and head will become col- with the next itemset until all the large itemsets are
umns in the relational table R, (c) the XML frag- determined. Before the calculation of an itemset’s
ments obtained by filtering the context will become support to decide on pruning or keeping it, the
rows in the table R, and (d) by applying a contains itemset is considered to be a candidate itemset
function (which, for a given XML fragment x and (i.e., possibly large) if all its sub sets are large
an XML fragment y, returns 1 if x contains y, and (i.e., observe the minimum support required).
0 otherwise), the binary relational table R is ob- The association rules are determined from the
tained, which will be used during the rule-extrac- largest itemsets extracted, and for each of them
tion step to determine binary association rules a confidence is calculated as follows: For a rule
applicable to the XML document. X→Y, its confidence is equal to the percentage
The selection done during the preprocessing of transactions containing X that also contain Y.
phase, by specifying the context, the body, and In the three-step approaches presented in the
the head of the association rules, is considered previous subsection, after obtaining the binary
by some researchers not generic enough (Ding et table R in the preprocessing phase, any relational
886
association rule algorithm can be applied (e.g., were based on the Apriori algorithm sequence.
generic a priori) to get the relationship between The main feature is that this framework (Feng,
the binary values in the table, which represent Dillon, Wiegand, & Chang, 2003) considers
the existence of an XML fragment inside another in more detail the specific format of the XML
XML fragment. The steps of the generic Apriori documents, that is, their possible representation
algorithm have been detailed in the previous para- as trees. We recall that at the beginning of the sec-
graph. In the particular case of the binary matrix tion on Apriori-based approaches, we proposed a
R, the rows of the matrix will be transactions to translation of the terms transaction and item into
be mined by the algorithm. The binary knowledge some concepts more specific to XML association
extracted at this step will signify the simultaneous rule mining. The non-Apriori-based framework
presence of fragments from the body or head in discussed in the current section proposes a dif-
the selected context. ferent mapping of the above terms to tree-like
structured XML documents.
Postprocessing Phase The work of Feng et al. (2003) aims to dis-
cover association rules from a collection of XML
After the extraction of the binary association documents rather than from a single document,
rules from the relational table during the second hence each XML document or tree corresponds
step, they will be transformed back into XML- to a database record (transaction), where each
specific representations of the discovered rules. XML fragment (subtree) corresponds to an item
We remember from the preprocessing step that the in the transaction. In this context, the framework
filtered XML fragments obtained by applying the proposed intends to discover association rules
body and head path queries on the XML docu- among trees in XML documents rather than among
ment became columns in the table, while filtered simple-structured items. Each tree is named a
XML fragments obtained by applying the context tree-structured item and is a rooted, ordered tree
path queries became rows. Reversing the process, having its nodes classified into (a) basic nodes with
together with the new knowledge determined, no edges emanating from them and (b) complex
that is, the association rules between the binary nodes, which are internal nodes with one or more
values, we get an XML structure in which each edges emanating from them. In Figure 7 we pres-
<rule> element has two attributes, support and ent some of the concepts introduced to define the
confidence, and two child elements, <body> and framework for mining XML association rules.
<head>, where the fragments of the body and head
participating in the rule are listed. An example
of the result of applying the XMINE algorithm
Figure 6. Example of XML association rules ob-
is presented in Figure 6, in which the following
tained by applying the XMINE RULE algorithm
rules are given: “Author A → Author H has 85%
support and 20% confidence” and “Author H and
Author B → Author A has 70% support and 22%
confidence.”
Non-Apriori-Based Approach
In this section, we present one framework for

discovering association rules that is different
from the earlier described approaches, which
887
Figure 7. Example of two tree-structured items in the framework for mining XML association rules as
proposed by Feng et al. (2003)
In Figure 7, there are two tree-structured items, nodes sharing the same parent. For example,
the <PERSON> and <ITEM> elements, ex- in Figure 7, Posi(n2,4)=last() means the Title
tracted from the order.xml example document node with the Star War Game content is the
(Feng et al., 2003), in which the nodes n1,1, n2,1, title of the last ordered CD.
n2,2, and n2,3 are complex, while n1,2, n1,3, n1,4, n2,4,
n2,5, and n2,6 are basic. The edges inside the trees In this framework, a well-formed tree is a tree
are labeled depending on the type of relationship that observes three conditions: (a) It has a unique
between the nodes. There are two types of labels root node, (b) for any chosen edge in the tree, if
attached to edges: ad (ancestor-descendant) and it is labelled ad, it will link a complex node with
ea (element-attribute). In Figure 7, the edge that a basic node, while if it is labeled ea, the source
connects the PERSON with the Profession node node needs to be a complex node, and (c) all the
is labeled ea because Profession is an attribute of constraints are correctly applied, that is, a level
the PERSON in the XML document. All the constraint can be applied only on an ad edge, while
other edges are labeled ad as they represent con- an adhesion constraint can be applied only on an
nections between a parent node and a child node. ea edge. Using the above described concepts, the
There are three types of constraints that can be subtree concept (subitem) is defined based on the
imposed on nodes and edges, as follows. definition of the subtree relationship. A tree T
with root r is a subtree of the tree T’ with root r’
1. )>> Level constraints: If e is an ad relationship (noted T ≤tree T’) if and only if there is a node n’
nsource→ntarget, Level (e)=m (m integer) means in T’ such that r is a part of n’ (noted r ≤node n’).
that ntarget is the mth descendant of the nsource. We refer the reader to the original paper (Feng et
2. )>> Adhesion constraints: If e is an ea rela- al., 2003) for more details and explanations on
tionship nsource→ntarget, Adhesion(e)=strong these concepts.
means that ntarget is a compulsory attribute of Finally, the association rule is defined as an
nsource, while Adhesion(e)=weak means that implication T1→T2 that satisfies two conditions.
ntarget is an optional attribute of the nsource.
3. )>> Position constraints: They refer to the actual 1. )>> X ⊂ T, Y ⊂ T and X ∩ Y = ∅, where T is the
contextual position of the node among all the set of tree-structured items and
888
2. )>> For any Tm and Tn ∈ (X ∪ Y), there is no tree presented in this chapter consists of the way they
Tp that can satisfy the conditions Tp ≤tree Tm perceive the notion of item, which they consider
and Tp ≤tree Tn. in their mining algorithms. While the former ones
extract the items to be mined as a list of nodes
An example of the association rule in terms by querying the XML document for a specific
of tree-structured items (named XML-enabled path, for the last one, each subtree (substructure)
association rules by the authors) is presented in in the XML document tree representation is an
Figure 8. item and the framework actually looks to discover
The rule exemplified in Figure 8 tells that if a association rules between the substructures of the
male person orders a CD with the title Star War document.
Game, he will also order two books, that is, Star Another significant difference resides in the
War I and Star War II, in this order. Though an number of XML documents allowed by the al-
algorithm to implement the above described gorithms and the degree of the complexity of the
framework is still under development, the obtained documents (levels of nesting). Sometimes we may
association rules are powerful as they address the want to find association rules from a single XML
specific format of the XML documents; the as- document (e.g., books in a library) or from two or
sociated items are hierarchical structures, not more XML documents (e.g., documents containing
simple nodes. Furthermore, they carry the notion books in a library, one containing personal details
of order, as exemplified by rule in Figure 8. of the authors and the third containing sales of the
books for a period of time). If we have a collection
Summary of Association of XML documents, it is probable that we will
Rule Mining Techniques for get more interesting information by analysing all
Static XML Documents the documents together instead of one at a time.
The simple (one-step) XML association rule
To conclude this section, we make some com- mining techniques (Wan & Dobbie, 2003, 2004)
ments on the major differences between the above are considering one single document, with a
discussed XML association rule techniques and simple structure (see Figure 2a), for example, an
the degree of the possible generalization of them, XML document containing transactions in a su-
considering both the number of XML documents perstore, with the corresponding purchased items.
mined at once and the structure of these docu- The authors state that their proposed algorithm
ments, together with some experimental results “works with any XML document, as long as the
of the authors. structure of it is known in advance” (p. 94), but
The main difference between the Apriori-based they consider that applying their algorithm to an
approaches and the non-Apriori-based framework XML document with a more complex structure
Figure 8. An example of the XML-enabled association rule (Feng et al., 2003)
889
is still an open issue from the performance point time in the future for business purposes. In this
of view. case, a high degree of redundancy might appear,
The three-step approaches (Braga et al., 2003; and the user will end up with a large collection
Braga, Campi, Ceri et al., 2002; Braga, Campi, of XML documents in which a large amount of
Klemettinen et al., 2002) are designed to work information is repeated.
with more complex-structured XML documents The issue for researchers was how to efficiently
(see the example in Figure 3 with five levels of store all these versions so the user will be able to
nesting). Still, the structure of the document needs get a historic version of the document with as less
to be known in advance as the context, body, and redundancy of information as possible. Moreover,
head of the association rules should be defined at a new question was raised about what kind of
the beginning of the algorithm. The authors acknowledge can be discovered from the multiple
knowledge that, even if the experiments were done versions of an XML document; the goal in the
without considering efficiency as a main concern, case of mining dynamic XML documents would
the results proved excellent performance when be to find a different type of knowledge than can
using the Xalan (2005). Also, the experimental be obtained from snapshots of data. For example,
results showed that only a small percentage of time some parts of the XML document representing the
was spent for preprocessing and postprocessing the online store could change more often, and some
XML document, while the actual mining was the other parts could change together; for instance,
slowest phase. The authors reckon that any future deletions could appear more often than updates,
step in the XQuery development to allow more and so on. All this information could be usefully
complex conditions in filtering XML documents utilised by the end user in making business deci-
will determine a substantial improvement of the sions related to the online store’s content.
mining step’s efficiency and speed. In this section, we will first refer to the work
done for versioning XML documents, that is,
methodologies that efficiently store the chang-
Discovering Association ing XML documents in a way that allows the
Rules from Dynamic fast retrieval of the historic versions. They will
XML Documents include our own proposed solution to the issue of
versioning dynamic XML documents to collect
As specified in the background section, this section all the changes between versions in a single XML
details some of the work done for dynamic XML document, named consolidated delta. Finally, we
document versioning and mining. A dynamic XML will describe our proposed solution for mining
document is one that is continually changing its association rules from changes supported by the
content and/or structure in time depending on the dynamic XML documents.
data requested to be stored at a certain moment. Most of the methodologies addressing the
An example could be the content of an online issue of versioning XML documents are based
bookstore, where any change in the number of on the concept of the delta document (Cobena,
existing books, their prices, and/or availability Abiteboul, & Marian, 2005; Marian, Abiteboul,
will affect the content of the XML document that Cobena, & Mignet, 2001). This is calculated and
stores this information. The possible user (e.g., the built by comparing two consecutive versions of
online store manager) might decide to store each the XML document and recording the changes
new version of the XML document, which results that have been taking place.
after each change, so he or she would be able to XML versioning techniques come to solve two
refer to the history of the store’s content at any main issues (Zhao, 2004), as follows:
890
1. )>> The querying time can be improved by operations capable to transform the initial version
limiting the amount of data that need to be of the document (Di) into the final version (Di+1).
queried if the result of the same query in the Furthermore, based on the observation that the
previous state of the document is already delta ∆i is not enough to transform Di+1 back into
known. Di, the authors introduce the notion of completed
2. )>> Storing historical structural deltas (the actual delta. This is a delta that contains more informa-
changes) of the XML documents can help tion and works both forward and backward, being
to find knowledge (e.g., association rules) able to obtain Di or Di+1 when the other version
not just for snapshot data (as in mining static is available.
XML documents), but also considering their In our working example (Figure 9), the for-
evolution in time. ward, backward, and completed deltas are shown
in Figure 10.
For a better understanding of the differences In the example in Figure 10, T1 is the tree
between the XML versioning techniques, we rooted at node 11, that is, the <price> node, while
will exemplify them on two versions of an XML T2 is the tree rooted at node 3, the <name> node.
document (catalog.xml), which contains data These two trees will be included in the com-
about some products in an online store (Figure 9). pleted delta XML document. In the delete and
A change-centric management of versions in insert sequences, the first parameters are the par-
an XML warehouse was first introduced by Mar- ent node, the second parameters are the affected
ian et al. (2001). They consider a sequence of node positions, and the third parameters are the
snapshots of XML documents and, for each pair trees rooted at the affected nodes. In the update
of consecutive versions, the algorithm calculates sequences, the first parameters are the affected
a delta document as the difference between them. nodes, the second ones are the new values, while
Delta ∆i is a sequence of update, delete, and insert the third parameters are the old values.
Figure 9. Two consecutive versions of the same XML document, catalog.xml, with corresponding IDs,
in both XML document format and trees
891
Figure 10. Examples of forward, backward, and completed deltas
In this approach (Marian et al., 2001), a pre- For our working example, the minimum edit
sumptive XML warehouse will need to store the script generated by X-Diff would be:E= {delete
initial version of an XML document together with (3), update (5, “Available”), update (7,300), insert
all the completed deltas calculated in time so the (4, (Price, 160)}.
model will be able to successfully solve different As it can be noticed, the insert operation does
versioning requests. At the same time, the authors not include the position of the new inserted node
acknowledge that one of the most important issues because the X-Diff technique is focused on the
in their approach is the storage of the redundant unordered XML trees, and the position of the node
information (e.g., both the old version and new is not considered important for the algorithm.
version of elements consecutively updated will A novel way of storing changes in time with
be stored in the completed deltas). less overhead was proposed by Rusu, Rahayu, and
Another change detection algorithm, X-Diff, Taniar (2005b). In this approach, earlier versions
was proposed by Wang, DeWitt, and Cai (2003), of the documents can be easily queried and the
focusing on unordered XML document trees degree of redundancy is very small. Our algorithm
(where the left-to-right order among siblings is replaces the way of storing differences between
not important). They argue that an unordered tree two versions of an XML document in deltas and
model is more appropriate for most applications keeping all the deltas in the warehouse with a
than the ordered model (where both the ancestor- new concept of consolidated delta, in which
descendant and the left-to-right order among changes between versions are recorded in a new
siblings are important) and propose a methodol- XML document, modified any time a new ver-
ogy that detects changes in XML documents by sion appears. The main idea is to build a single
integrating specific XML structure characteristics (consolidated) XML delta document containing
with standard tree-to-tree correction techniques. all the changes supported by the versioned XML
We do not detail here the X-Diff algorithm, but document in the T1–Tn period of time by introduc-
mainly, it performs the followings steps to deter- ing a new temporal element (namely, <stamp>)
mine the minimum-cost edit sequence that is able to store the changes at each time stamp for each
to transform document D1 into document D2. (a) altered element. Each <stamp> element has two
It parses the D1 and D2 documents and builds the attributes: time to store the time stamp and delta
associated T1 and T2 trees while at the same time, to store the type of change (delta can take one of
it computes an XHash value for every node used the values inserted, modified, or deleted).
to represent the entire subtree rooted at the node. To exemplify the consolidated delta approach,
(b) It compares the XHash values for the roots and Figure 11 shows another set of changes that have
decides if the trees are equivalent (when the XHash been applied to the document in Figure 9. The
values are equal); otherwise, it calculates min(T1, changes between Version 1 and Version 2 are re-
T2) as a minimum-cost matching between trees. corded in the first consolidated delta (left), which
(c) It determines the generated minimum-cost edit is built starting from the initial version (Version 1),
script E based on the min (T1,T2) found at step b. adding the <stamp> elements as explained before.
892
Similarly, after another set of changes happen at If a parent is deleted at the time Ti, all its children
time T3 (Version 3), new <stamp> elements are will be deleted so they will not appear in the
added and the consolidated delta is updated to consolidated delta for that particular time stamp
reflect these (right). Every time the consolidated or for any time stamp after that.
delta is modified to reflect new changes, there To get a high speed in building the consoli-
are rules to be observed in order to increase the dated delta, we assign unique identifiers to ele-
efficiency of the algorithm and eliminate the re- ments in the initial XML document and store the
dundancy as much as possible; we list them here, maximum ID value. When new elements are in-
as follows. (a) If all the children are unchanged, serted in a following version, they will receive
the parent is unchanged. If a parent is unchanged at IDs based on the existing maximum ID so at any
the time Ti, its children are not marked (stamped) time, one element will be uniquely identified and
for that particular time stamp; they will be easily we will be able to track its changes.
rebuilt from the existing previous versions of their The two big advantages of the consolidated
parents. (b) If any of the children are modified, approach are the following: (a) There is a very
deleted, or inserted, the parent is modified. If a small degree of redundancy of the stored data
parent is modified at the time Ti, all its children as unchanged data between versions will not
will be stamped, each with their own status, that be repeated, and (b) it is enough for the user to
is, modified, inserted, deleted, or unchanged. (c) store the calculated consolidated delta to be able
Figure 11. Example of the consolidated delta after two series of changes applied to the initial XML
document catalog.xml
893
to get an earlier version of the document at any has the modified value, if the element is a complex
time. We have tested the algorithm of building the one (it has children), we analyse the changes for
consolidated delta and it has excellent results for each of its children elements; they may have a
various dimensions of XML documents. modified, deleted, added, or unchanged value as
well. If the element is not a complex one, we take
Versioning Dynamic XML its value. (b) When the delta has the inserted
Documents Using the value, if it is a complex element, all its children
Consolidated Delta Approach were inserted, too; so, we take their values as they
are returned for the Ti time stamp. If the element
The consolidated delta is a very efficient tool is not a complex one, we take its value. (c) When
when the user wants to retrieve an old version the delta has the deleted value, if the parent ele-
of the document. Suppose the latest version of ment was deleted (together with its children) at
the document is at the moment Tn in time (see the time Ti, the consolidated delta will contain the
Figure 12), and the user wants to determine the deleted value for the delta and no children details;
effective look (structure and content) for the XML consequently, the element will not appear in the
document at a moment Ti, where i<n (i=3 in the built version document. (d) When the delta has
example in Figure 12). Using the consolidated the unchanged value, we know a complex un-
delta, he or she does not need to re-create the changed element does not include its unchanged
entire set of intermediate documents from Tn to children; so if we find an unchanged element, we
Ti (Tn→Tn-1, Tn-1→Tn-2….Ti+1→Ti). Instead, the will query backward for each of the Ti-1, Ti-2, and
consolidated delta can be directly queried to get so forth, and earlier time-stamp changes until we
the elements that have <stamp> elements with a get to a version without an unchanged delta at-
Ti value of the time attribute. This query will not tribute or until we get to the initial D1 version of
return at once the entire structure and content the document (included in the consolidated delta
of the XML document at the moment Ti — this as a starting point).
would be an ideal output. We still have to query
backward in the history of certain elements, but Mining the Changes Extracted
only for a limited number, that is, the unchanged from Dynamic XML Documents
ones as the modified or inserted elements will
contain the actual values at the time Ti. In our view, there are two different ways to ap-
When an earlier version is required, the con- proach the task of mining dynamic XML docu-
solidated delta document is scanned starting from ments. Supposing there is a collection of versions
the root, and for each element, the algorithm of n dynamic XML documents stored in an XML
determines if the delta attribute has one of the data warehouse, composed by using one of the
modified, deleted, added, or unchanged values, above presented methods, the user might be in-
building, at the same time, the required Di version terested in discovering either of the following:
of the document as follows. (a) When the delta
Figure 12. Using consolidated delta to get an earlier version of an XML document
894
a. )>> Interesting knowledge (in our case, associa- & <description> inserted value=“good book” &
tion rules) that can be found in the collection <price> inserted value=“25”;
of historic versions of the document(s) In a generalized Apriori-based algorithm,
b. )>> Association rules extracted from the actual the set of items I will be the list of all distinct
changes between versions, that is, from the elements from the initial XML document. In our
differences recorded in delta documents example, I = {<catalog>, <product>, <status>,
<name>, <description>, <price>}. The extracted
There was some work done to discover fre- changes will form the set of transactions D, where
quently changing structures in versions of XML each transaction T from D is a set of items from
documents (Chen et al., 2004; Zhao, Bhowmick, I represented by one set of changes extracted for
Mohania, & Kambayashi, 2004; Zhao, Bhowmick, one time stamp Ti, 2<i<n. A possible association
& Mandria, 2004) applicable more to discovering rule will be an implication X→Y where X, Y ⊂ I
the first type of knowledge (Case a above). We and X ∩ Y =∅. The rule’s support and confidence
do not detail them here; instead, we will propose will be calculated with regard to the total number
a novel method of mining changes extracted of changes extracted. Dynamic association rules
from dynamic XML documents (applicable for discovered in this way could give precious infor-
the second type of knowledge, Case b above) by mation about the relationship between changes af-
using the consolidated delta described earlier in fecting specific parts of the initial XML document.
the previous subsection. Mainly, mining is done For example, it could be found that the insertion
by extracting the set of changes for each time Ti of new products determine a fall in the availability
(2<i<n, where n is the total number of versions of certain products. We are currently working on
and Ti is the time of each set of changes) as a set implementing and proving the efficacy of this
of transactions. After that, we mine them apply- mining algorithm.
ing one of the classic algorithms for discovering
association rules, for example, the a priori one Summary of Association
(Agrawal et al., 1993). Because the implementa- Rule Mining Techniques for
tion of the actual algorithm is still under review, Dynamic XML Documents
we will give only a general description of the
technique using an example. In this section, we have presented some of the
We consider again the consolidated delta ex- state-of-the-art work in the area of recording
ample in Figure 10. If we extract only the changes changes between versions of dynamic XML
from the consolidated delta for each of the times documents, detailing more on the consolidated
T2 and T3, we get the following two transactions: delta approach, which is an effective way to store
At time T2 → <catalog> modified & <prod- successive changes of the documents in a single
uct> modified & <name> deleted & <price> document. Then, we have presented an algorithm
inserted value=“160” & <status> modified for extracting a historic version of the document
value=“Available” & <price> modified val- at any time where its versions are stored by us-
ue=300; ing the consolidated delta approach. Finally, we
At time T3 → <catalog> modified & <prod- presented our view on mining the set of changes
uct> modified & <description> modified value=“A extracted for a given period of time.
new stuff” & <price> modified value=“150” & The methods presented for storing the changes
<price> modified value=“400” & <product> in- between versions of the XML documents are all
serted & <status> inserted value=“Not available” using the concept of delta as a difference between
two consecutive versions of the XML document,
895
but each approach comes with its own definition static XML documents need to know from
and implementation as the target is to find the most the beginning which are the specific ar-
efficient representation that is easy to interrogate eas they need to look at to find either the
and mine later on. While the work of Marian et antecedent or the consequent of the asso-
al. (2001) proposes building a consolidated delta ciation rule. In this context, future work is
as a set of instructions able to reverse the initial needed to improve the existing methodolo-
version of the document to the final one and vice gies in terms of generalization (Buchner,
versa for ordered XML documents, the technique Baumgarten, Mulvenna, Bohm, & Anand,
introduced by Wang et al. (2003) is similar but 2000; Garofalakis, Rastogi, Seshadri, &
focuses on unordered XML documents. A different Shim, 1999). Finding algorithms with a
approach is given by Rusu, Rahayu, and Taniar high degree of generalization is imperative
(2005), in which the proposal is to record the as scalability is a priority for the current
historic changes in one single document, named and future XML-driven applications.
consolidated delta, that is easy to be queried when •)>> Mining association rules from dynamic
the user needs to extract an old version of the XML documents (i.e., documents that
document. The same consolidated delta approach change their content in time to allow dif-
can be used to perform the mining of association ferent formats of data): Dynamic mining
rules out of the set of changes applied to the initial is still a very young area in which a lot of
document, returning possible interesting informa- research has been undertaken. From our
tion about the relationships between changes and perspective, intense activity in this field
their influences on the XML document’s behaviour will be noticed soon as Web applications
in the future. are used on a large scale and manipulate
dynamic data. Besides association rules,
researchers are looking to find other types
Future Trends of patterns in dynamic XML documents,
that is, structural changes from an XML
In this section, we present our view on the future document version to another, and content
trends in the area of mining XML documents, changes. Our next research work is to im-
considering how the existing work answers pos- plement and evaluate a mining algorithm
sible user needs. able to discover association rules and other
types of knowledge from the sequence of
•)>> Mining association rules from static actual changes of dynamic XML docu-
XML documents (i.e., documents that ments. The outcome of this work will be
are not changing their content in time): very useful in finding not only what the
In this area, the majority of the research patterns are in the changing documents,
work has been focused not so much on de- but also how they relate to one another and
termining generic association rules (what how they could affect the future behaviour
type of knowledge can be extracted from of the initial XML document.
a certain XML document or from a col-
lection of XML documents), but more
on seeking a confirmation of possible as- Conclusion
sociation rules between elements or parts
of the document. For example, the major- This chapter is a systematic analysis of some of the
ity of the presented algorithms for mining existing techniques for mining association rules
896
out of XML documents in the context of rapid Ashrafi, M. Z., Taniar, D., & Smith, K. (2005).
changes and discoveries in the Web knowledge An efficient compression technique for frequent
area. The XML format is more and more used to itemset generation in association rule mining.
store data that now exist in the traditional rela- In Proceedings of International Conference in
tional-database format, and also to exchange them Advances in Knowledge Discovery and Data Min-
between various applications over the Internet. ing (PAKDD 2005) (LNCS 3518, pp. 125-135).
In this context, we presented the latest dis- Heidelberg, Germany: Springer-Verlag.
coveries in the area of mining association rules
Ashrafi, M. Z., Taniar, D., & Smith, K. A. (2004a).
from XML documents, both static and dynamic,
A new approach of eliminating redundant associa-
in a well-structured manner, with examples and
tion rules. In Database and expert systems appli-
explanations so the reader will be able to easily
cations (LNCS 3180, pp. 465-474). Heidelberg,
identify the appropriate technique for his or her
Germany: Springer-Verlag.
needs and replicate the algorithm in a develop-
ment environment. At the same time, we have Ashrafi, M. Z., Taniar, D. & Smith, K. A. (2004b).
included in this chapter only the research work ODAM: An optimized distributed association
with a high level of usability in which concepts rule mining algorithm. IEEE Distributed Systems
and models are easy to be applied in real situations Online, 5(3).
without imposing knowledge of any high-level
Braga, D., Campi, A., & Ceri, S. (2003). Discover-
mathematics concepts.
ing interesting information in XML with associa-
The overall conclusion is that this chapter is a
tion rules. Proceedings of 2003 ACM Symposium
well-structured tool very useful for understand-
on Applied Computing (SAC’03) (pp. 450-454).
ing the concepts behind discovering association
rules out of collections of XML documents. It is Braga, D., Campi, A., Ceri, S., Klemettinen,
addressed not only to the students and other aca- M., & Lanzi, P. L. (2002). A tool for extracting
demics studying the mining area, but to the real XML association rules. Proceedings of the 14th
end users as a guide in creating powerful XML International Conference on Tools with Artificial
mining applications. Intelligence (ICTAI ’02) (p. 57).
Braga, D., Campi, A., Klemettinen, M., & Lanzi,
P. L. (2002). Mining association rules from XML
References
data. In Proceedings of International Conference
Agrawal, R., Imielinski, T., & Swami, A. N. on Data Warehousing and Knowledge Discovery
(1993). Mining association rules between sets of (DaWak 2002) (LNCS 2454, pp. 21-30). Heidel-
items in large databases. Proceedings of the ACM berg, Germany: Springer-Verlag.
International Conference on Management of Data Buchner, A. G., Baumgarten, M., Mulvenna, M.
(SIGMOD 1993) (pp. 207-216). D., Bohm, R., & Anand, S. S. (2000). Data mining
Agrawal, R., & Srikant, R. (1998). Fast algorithms and XML: Current and future issues. Proceedings
for mining association rules. In Readings in data- of 1st International Conference on Web Information
base systems (3rd ed., pp. 580-592). San Francisco: System Engineering (WISE 2000) (pp. 127-131).
Morgan Kaufmann Publishers Inc.
897
Chen, L., Browmick, S. S., & Chia, L. T. (2004). Nayak, R. (2005). Discovering knowledge from
Mining association rules from structural deltas XML documents. In J. Wong (Ed.), Encyclopedia
of historical XML documents. In Proceedings of of data warehousing and mining (pp. 372-376).
International Conference in Advances in Knowl- Hershey, PA: Idea Group Reference.
edge Discovery and Data Mining (PAKDD 2004)
Nayak, R., Witt, R., & Tonev, A. (2002). Data
(LNCS 3056, pp. 452-457). Heidelberg, Germany:
mining and XML documents. Proceedings of
Springer-Verlag.
the 2002 International Conference on Internet
Cobena, G., Abiteboul, S., & Marian, A. (2005). Computing (pp. 660-666).
XyDiff tools: Detecting changes in XML docu-
Rusu, L. I., Rahayu, W., & Taniar, D. (2005a).
ments. Retrieved February 2006, from http://www.
Maintaining versions of dynamic XML docu-
rocq.inria.fr/gemo
ments. In Proceedings of the 6th International
Daly, O., & Taniar, D. (2004). Exception rules Conference on Web Information System Engi-
mining based on negative association rules. In neering (WISE 2005) (LNCS 3806, pp. 536-543).
Computational science and applications (LNCS Heidelberg, Germany: Springer-Verlag.
3046, pp. 543-552). Heidelberg, Germany:
Rusu, L. I., Rahayu, W., & Taniar, D. (2005b). A
Springer-Verlag.
methodology for building XML data warehouses.
Ding, O., Ricords, K., & Lumpkin, J. (2003). International Journal of Data Warehousing and
Deriving general association rules from XML Mining, 1(2), 67–92.
data. Proceedings of the ACIS 4th International
Tjioe, H. C., & Taniar, D. (2005). Mining asso-
Conference on Software Engineering, Artificial
ciation rules in data warehouses. International
Intelligence, Networking and Parallel/Distributed
Journal of Data Warehousing and Mining, 1(3),
Computing (SNPD’03) (pp. 348-352).
28–62.
Feng, L., Dillon, T., Wiegand, H., & Chang, E.
Wan, J. W., & Dobbie, G. (2003). Extracting
(2003). An XML-enabled association rules frame-
association rules from XML documents using
work. In Proceedings of International Conference
XQuery. Proceedings of the 5th ACM International
on Database and Expert Systems Applications
Workshop on Web Information and Data Manage-
(DEXA 2003) (LNCS 2736, pp. 88-97). Heidel-
ment (WIDM’03) (pp. 94-97).
berg, Germany: Springer-Verlag.
Wan, J. W., & Dobbie, G. (2004). Mining as-
Garofalakis, M. N., Rastogi, R., Seshadri, S., &
sociation rules from XML data using XQuery.
Shim, K. (1999). Data mining and the Web: Past,
Proceedings of International Conference on Re-
present and future. Proceedings of the 2nd Work-
search and Practice in Information Technology
shop on Web Information and Data Management
(CRPIT 2004) (pp. 169-174).
(WIDM 1999) (pp. 43-47).
Wang, Y., DeWitt, D. J., & Cai, J. Y. (2003). X-Diff:
Laurent, M., Denilson, B., & Pierangelo, V. (2003).
An effective change detection algorithm for XML
The XML Web: A first study. Proceedings of the
documents. Proceedings of the 19th International
International WWW Conference (pp. 500-510).
Conference on Data Engineering (ICDE 2003)
Marian, A., Abiteboul, S., Cobena, G., & Mignet, (pp. 519-530).
L. (2001). Change-centric management of ver-
World Wide Web Consortium. (W3C). (n.d.). Re-
sions in an XML warehouse. The VLDB Journal,
trieved February 2006, from http://www.w3c.org
581–590.
898
Xalan. (2005). The Apache Software Foundation: Zhao, Q., Bhowmick, S. S., Mohania, M., &
Apache XML project. Retrieved December 2005, Kambayashi, Y. (2004). Discovering frequently
from http://xml.apache.org/xalan-j/ changing structures from historical structural
deltas of unordered XML. Proceedings of ACM
XQuery. (2005). Retrieved February 2006,
International Conference on Information and
from http://www.w3.org/TR/2005/WD-xque-
Knowledge Management (CIKM’04) (pp. 188-
ry-20050915/
197). Heidelberg, Germany: Springer Berlin.
Zhao, Q., Bhowmick, S. S., & Mandria, S. (2004).
Discovering pattern-based dynamic structures
from versions of unordered XML documents. In
Proceedings of International Conference on Data
Warehousing and Knowledge Discovery (DaWaK
2004) (LNCS 3181, pp. 77-86). Heidelberg, Ger-
many: Springer-Verlag.
This work was previously published in Services and Business Computing Solutions with XML: Applications for Quality
Management and Best Processes, edited by Patrick Hung, pp. 176-196, copyright 2009 by Business Science Reference
899
Section IV
Utilization and Application
This section introduces and discusses the utilization and application of enterprise information systems around
the world. These particular selections highlight, among other topics, enterprise information systems in multiple
countries, data mining applications, and critical success factors of enterprise information systems implementation.
Contributions included in this section provide excellent coverage of the impact of enterprise information systems
on the fabric of our present-day global village.
901
Chapter 4.1
QoS-Oriented Grid-Enabled
Data Warehouses
Rogério Luís de Carvalho Costa
University of Coimbra, Portugal
Pedro Furtado
University of Coimbra, Portugal
AbstrAct do not concern the grid environment and those

which do so, use best-effort oriented strategies.
Globally accessible data warehouses are useful Our experimental results show the importance and
in many commercial and scientific organizations. effectiveness of proposed strategies.
For instance, research centers can be put together
through a grid infrastructure in order to form
a large virtual organization with a huge virtual INtrODUctION
data warehouse, which should be transparently
and efficiently queried by grid participants. As it In the last few years, Grid technology became a
is frequent in the grid environment, in the Grid- key component in many widely distributed appli-
based Data Warehouse one can both have resource cations from distinct domains, which include both
constraints and establish Service Level Objectives research-oriented and business-related projects. The
(SLOs), providing some Quality of Service (QoS) Grid is used as an underlying infrastructure that
differentiation for each group of users, participant provides transparent access to shared and distrib-
organizations or requested operations. In this work, uted resources, like supercomputers, workstation
we discuss query scheduling and data placement in clusters, storage systems and networks (Foster,
the grid-based data warehouse, proposing the use 2001). In Data Grids, the infrastructure is used to
of QoS-aware strategies. There are some works on coordinate the storage of huge volumes of data or
parallel and distributed data warehouses, but most the distributed execution of jobs which consume
or generate large volumes of data (Krauter et al,
DOI: 10.4018/978-1-60566-756-0.ch009
QoS-Oriented Grid-Enabled Data Warehouses
2002; Venugopal et al, 2006). Most of the works SLOs may be used by the Community Scheduler
on data grids considers the use or management to negotiate with Local Schedulers the establish-
of large files, but grid-enabled Database Manage- ment of Service Level Agreements (SLA). But
ment Systems (DBMS) may be highly useful in SLOs can also be used to provide some kind of
several applications from distinct domains (Nieto- differentiation among users or jobs. Execution
Santisteban et al, 2005; Watson, 2001). deadline and execution cost’s limit are some
On the other hand, data warehouses are mostly example of commonly used SLOs.
read-only databases which store historical data We consider here the use of deadline-marked
that is commonly used for decision support and queries in grid-based Data Warehouses. In such
knowledge discovery (Chaudhuri & Dayal, 1997). context, execution time objectives can provide
Grid-based data warehouses are useful in many some differentiation between interactive queries
real and virtual global organizations which are and report queries. For example, one can establish
generating huge volumes of distributed data. that interactive queries should be executed by a 20
In such context, the data warehouse is a highly seconds deadline and that report queries should be
distributed database whose data may be loaded executed in 5 minutes. In fact, different deadlines
from distinct sites and that should be transparently may be specified considering several alternatives,
queried by users from distinct domains. like the creation of privileged groups of users that
But constructing effective grid-based applica- should obtain responses in lower times or like
tions is not simple. Grids are usually very hetero- providing smaller deadlines for queries submitted
geneous environments composed by resources by users affiliated to institutions that had offered
that may belong to distinct organization domains. more resources to the considered grid-based data
Each domain administrator may have a certain warehouse.
degree of autonomy and impose local resource Data placement is a key issue in grid-based
usage constraints for remote users (Foster, 2001). applications. Due to the grid’s heterogeneity and
Such site autonomy is reflected in terms of to the high cost of moving data across different
scheduling algorithms and scheduler architec- sites, data replication is commonly used to improve
tures. The hierarchical architecture is one of the performance and availability (Ranganathan &
most commonly used scheduling architecture in Foster, 2004). But most of the works on replica
Grids (Krauter et al, 2002). In such architecture, selection and creation in data grids consider ge-
a Community Scheduler (or Resource Broker) is neric file replication [e.g. (Lin et al, 2006; Siva
responsible to transform submitted jobs into tasks Sathya et al, 2006; Haddad & Slimani, 2007)].
and to assign them to sites for execution. At each Therefore, the use of specialized data placement
site, a Local Scheduler is used to manage local strategies for the deployment of data warehouses
queues and implement local domain scheduling in grids still remains an open issue.
policies. Such architecture enables a certain degree In this chapter, we discuss the implementation
of site autonomy. of QoS-oriented Grid-enabled Data Warehouses.
Besides that, in Grids, tasks are usually speci- The grid-enabled DW is composed by a set of
fied together with Service Level Objectives (SLO) grid-enabled database management systems, a set
or Quality-of-Service (QoS) requirements. In fact, of tools provided by an underlying grid resource
in many Grid systems, scheduling is QoS-oriented management (GRM) system and hierarchical
instead of performance-oriented (Roy & Sander, schedulers. We combine data partitioning and
2004). In such situations, the main objective is replication, constructing a highly distributed da-
to increase user’s satisfaction instead of achiev- tabase that is stored across grid’s sites, and use a
ing high performance. Hence, the user-specified QoS-oriented scheduling and a specialized replica
902
selection and placement strategy to achieve high from a scalability problem: if a wide variety of
QoS levels. distributed heterogeneous resources is available,
This chapter is organized as follows: in the considering all the resources’ individual char-
next Section we present some background on data acteristics when scheduling job execution may
grids and grid-enabled databases. Then, we discuss become very time consuming. In the hierarchical
QoS-oriented scheduling and placement strategies architecture, a Community Scheduler (or Resource
for the Grid-based warehouse. In the following, Broker) is responsible to assign job execution to
we present some experimental results. Next, we sites. Each site has its own job scheduler (Local
draw conclusions. At the end of the chapter, we Scheduler) which is responsible to locally schedule
present some key terms definitions. the job execution. The Community Scheduler and
Local Schedulers may negotiate job execution
and each Local Scheduler may implement local
DAtA GrIDs AND GrID- resource utilization policies. Besides that, as the
ENAbLED DAtAbAsEs Community Scheduler does not have to exactly
know the workload and characteristics of each
The Grid is an infra-structure that provides available node, this model leads to greater scal-
transparent access to distributed heterogeneous ability than the centralized scheduling model.
shared resources, which belong to distinct sites In the Decentralized model, there is no Central
(that may belong to distinct real organizations). Scheduler. Each site has its own scheduler, which
Each site has some degree of autonomy and may is responsible to schedule local job execution.
impose resource usage restrictions for remote Schedulers must interact to each other in order to
users (Foster, 2001). negotiate remote job execution. Several messages
In the last decade, some Grid Resource Man- may be necessary during the negotiation in order
agement (GRM) Systems [for example, Legion to do good job scheduling, which may impact the
(Grimshaw et al, 1997) and Globus Toolkit (Foster system’s performance.
& Kesselman, 1997)] were developed in order to Some of the GRM systems have built-in
provide some basic functionality that is commonly scheduling policies, but almost all enable the
necessary to run grid-based applications. Autho- user to implement its own scheduling policy or to
rization and remote job execution management use application-level schedulers. In this context,
are among the most common features in GRM some general purpose application level schedul-
systems. Some of them also provide data man- ers were designed [e.g. Condor-G (Frey et al,
agement-related mechanisms, like efficient data 2001) and Nimrod-G (Buyya et al, 2000)]. These
movement [e.g. GridFTP (Allcock et al, 2005)] general purpose generally consider some kind
and data replica location [e.g. Globus Replica of user-specified requirement or QoS-parameter
Location Service – RLS (Chervenak et al, 2004)]. (e.g. job’s deadline), but may fail to efficiently
In terms of grid job scheduling, there are three schedule data-bound jobs.
basic architectures (Krauter et al (2002): central- Query scheduling strategies for data-bound
ized, hierarchical and decentralized. In the first jobs were evaluated by Ranganathan & Foster
one, a single Central Scheduler is used to schedule (2004). Data Present (DP), Least Loaded Sched-
the execution of all the incoming jobs, assigning uling (LLS) and Random Scheduling (RS) were
them directly to the existent resources. Such ar- compared. In RS, job execution is randomly
chitecture may lead to good scheduling decisions, scheduled to available nodes. In LLS, each job
as the scheduler may consider the characteristics is scheduled to be executed by the node that has
and loads of all available resources, but suffers the lowest number of waiting jobs. Both in RS
903
and LLS, a data-centric job may be scheduled marked queries among grid-enabled databases
to be executed by a job that does not store the when several replicas of the same data are present
required data to execute such job. In this case, at distinct sites.
remote data is fetched during job execution. In In Grids, data replicas are commonly used to
the DP strategy, each job is assigned to a node that improve job (or query) execution performance
stores the job’s required input data. Ranganathan and data availability. Best Client and Cascading
& Foster claim that, in most situations, DP has Replication are among the dynamic file replica-
better performance than LLS and RS (as doing tion strategies evaluated by Ranganathan & Foster
data movement across grid’s nodes may be very (2001) to be used in the Grid. In both models, a
time consuming). new file replica is created whenever the number
There are several parameters that should be of access to an existent data file is greater than a
considered when scheduling data-centric jobs. threshold value. The difference among the meth-
These include the size of the job’s input and ods resides on where such new file is placed. The
output data, and the network bandwidth among ‘best client’ of a certain data file is defined as the
grid’s nodes. Park & Kim (2003) present a cost node that has requested for each more times in a
model that use such parameters to estimate job’s certain time period. In the Best Client placement
execution time at each node (both considering strategy, the new replica is placed at the best cli-
that a job can be executed at the submission site ent node. In the Cascading Replication method,
or not, and that it may use local or remote data the new file is placed at the first node in the path
as input). Job execution is scheduled to the node between the node that stores the file that is being
with the lowest predicted execution time. replicated and the best client node.
Although very promising, the grid-enabled The Best Client strategy is used as an inspi-
database management systems were not largely ration for the Best Replica Site strategy [(Siva
adopted for a long time (Nieto-Santisteban et al, Sathya et al, 2006)]. The main different among
2005; Watson, 2001). Watson (2001) proposed the the this strategy and the original Best Client is
construction of a federated system with the use that in Best Replica Site the site in which the
of ODBC/JDBC as interface for heterogeneous replica is created is chosen considering not only
database systems. In more recent work, web ser- the number of access from clients to the dataset,
vices are used as interface to database management but also the replica’s expected utility for each site
systems. Alpdemir et al (2003) present an Open and the distance between sites. Sathya et al (2006)
Grid Services Architecture [OGSA – (Foster et al, also propose two other strategies: Cost Effective
2002)]-compatible implementation of a distributed Replication and Topology Based Replication. In
query processor (Polar*). A distributed query the first one, a cost function is used to choose in
execution plan is constructed by basic operations which site a replica should be created (the cost
that are executed at several nodes. function evaluates the cost of accessing a replica
Costa & Furtado (2008c) compares the use at each site). In the latter, database replicas are
of centralized and hierarchical query schedul- created at the node that has the greatest number
ing strategies in grid-enabled databases. The of direct connections to other ones.
authors present that hierarchical schedulers can Topology related aspects are also considered
be used without significant lose in the system’s by Lin et al (2006) in order to choose replica loca-
performance and can also lead to good levels of tion. The authors consider a hierarchical (tree-like)
achievement of Service Level Objectives (SLOs). grid in which database is placed at the tree root.
In Costa & Furtado (2008b) the authors propose Whenever a job is submitted, the scheduler looks
the use of reputation systems to schedule deadline- for the accessed data at the node in which the job
904
was submitted. If the necessary data is not at such tHE DIstrIbUtED QOs-
node, then the schedulers asks for it at the node’s OrIENtED WArEHOUsE
parent node. If the parent node does not have a
replica of the searched data, then the scheduler Data warehouses are huge repositories of historical
looks for it at the grandparent node, and so on. data. They are subject-oriented: a certain subject
Whenever the number of searched nodes is greater (revenue, for example) is analyzed considering
than a defined value, a new data replica is created. several distinct measures (e.g. time period). Each
Such newly created replica is placed at the node analyses measure domain is called a dimension.
that maximizes the number of queries that can be Therefore, the DW is a multidimensional space.
answered without creating new replicas. Such space is commonly represented in relational
Maximizing the economic value of locally databases as star schemas (Chaudhuri & Dayal,
stored data is the objective of the strategy proposed 1997), with several distinct dimension tables and
by Haddad & Slimani (2007). In such strategy, a huge facts table. The dimensions tables store
there is a price to access each data fragment. information about the analyze measures domains.
Each node tries to foresee the future price of the The facts table stores data that represents the events
fragments and stores the ones that are forecasted of the real world and pointers to dimensions tables.
as the most valuable. Figure 1 presents an example of a star schema (in
Most of the abovementioned strategies are the remaining of this chapter, tables are considered
oriented for file-based grids. Others are related to be conceptually organized in a star schema).
to best-effort oriented scheduling in grid-enabled The Grid-based data warehouse is accessed
databases. But all of them are somehow related by users of geographically distributed sites which
to the aspects we deal with in the next Sections. may or may not belong to the same real orga-
In the next Section, we discuss the architecture nization, but that are put together within a grid
and scheduling for the QoS-oriented grid-based infrastructure. Each site may share one or more
distributed data warehouse. resources to the grid. Examples of possible shared
Figure 1. Sample Star Schema
905
resources are storage systems, computer clusters is considered the first tier and stores cached data.
and supercomputers. Database servers at remote sites compose the
Data warehouses are usually deployed at a second tier. The scheduling algorithm tries to
single site. But that may not be the most effec- use the locally stored data to answer submitted
tive layout in a grid-based DW implementation. queries. If it is not possible, then remote servers
In fact, in such environment, placing the entire are accessed.
database at a single site would be more expensive The Globus Toolkit is used by Wehrle et al
and time consuming than creating a distributed (2007) as an underlying infrastructure to imple-
DW that uses the available distributed resources ment a grid-enabled warehouse. Facts table data
to store the database and to execute users’ queries. is partitioned across nodes participating nodes
It is important to consider that not only users and dimension data is replicated. Some special-
are distributed across distinct grid sites but also ized services are used at each node: (i) an index
that the warehouse’s data may be loaded from service provides information about locally stored
several sites. data; and (ii) a communication service is used to
Hence, in the distributed Grid-based DW, access remote data. Locally stored data is used to
data is partitioned and/or replicated at nodes answer incoming queries. If the searched data is
from distinct sites and may be queried by any not stored at the local node, then remote access
grid participant. is done by the use of the communication service.
This strategy and the abovementioned Olap-
best-Effort Approaches for enabled strategy do not provide any autonomy
Grid-Enabled Warehouses for local domains.
There are some previous works on implementing Distributed Data Placement

and using grid-enabled data warehouses, but most in Qos-Oriented DW
use best-effort oriented approaches, which may
not be the most adequate approach in grid based In data warehouses, users’ queries usually follow
systems (as presented in the previous Section, some kind of access pattern, like geographically
grid scheduling is usually satisfaction-oriented). related ones in which users from a location may
High availability and high performance are the have more interest in data related to such location
main concerns by Costa & Furtado (2006). Each than in data about other locations (Deshpande et al,
participating site stores a partitioned copy of the 1998). That may also be applicable for the grid. For
entire warehouse. Intra-site parallelism is obtained instance, consider a global organization that uses a
by the use of the Node Partitioned Data Warehouse grid-based DW about sales which is accessed by users
(NPDW) strategy (Furtado, 2004). Hierarchical from several countries. The users in New York City,
scheduler architecture is used together with an USA, may start querying data about sales revenue in
on-demand scheduling policy (idle nodes asks Manhattan, and then do continuous drill-up opera-
the Central Scheduler for new queries to execute). tions in order to obtain information about sales in
Such model leads to good performance and high New York City, in New York State and, finally, in the
availability, but also consumes too much storage USA. Only rarely New York users would query data
space, as the whole warehouse is present at each about sales in France. In the same way, users from
participating site. Paris may start querying the database about sales in
The Olap-enabled grid (Lawrence & Rau- France, and then start doing drill-down operations
Chaplin, 2006; Dehne et al, 2007) is a two tier in order to obtain data about sales in Paris and, then,
grid-enabled warehouse. The users’ local domain individually on each of its arrondissements.
906
In order to reduce data movement across sites replicated to distinct sites, what would increase
(improving the system’s performance) the grid- the system’s degree of parallelism. Besides that,
based DW tables are physically distributed across depending on the selection predicate, some queries
different sites. Such distribution is represented at a may access only a set of the smaller fragments,
Global Physical Schema (GPS). Ideally, the physi- which would be faster than accessing the whole
cally used data distribution strategy is transparent original site source-based fragment. These two
to users, which should submit queries considering situations are represented in Figure 2. Therefore,
a unified Logical Model (LM). even at the global level, other partitioning criteria
Grids are highly heterogeneous environments. should be used together with the site source at-
At each site, different types of resources may be tribute. The use of the most frequently used equi-
available (like shared-nothing and shared-disk join attributes as part of the partitioning criteria
parallel machines, for example). It is somewhat for the facts table can improve performance, by
difficult to find an intra-site allocation strategy reducing data movement across sites when execut-
that is optimal in the several possible situations. ing queries [as it does in shared-nothing parallel
Therefore, each site may use its own local physi- machines (Furtado, 2004b)].
cal allocation strategy (e.g. Multi-Dimensional Besides facts table’s partitions, each site should
Hierarchical Fragmentation – MDHF (Stöhr et also store dimension tables’ data. Full replication
al, 2000) or Node-Partitioned Data Warehouse of dimension tables across all sites may be done
strategy - NPDW (Furtado, 2004). Each site’s to reduce inter-site data movement during query
existent relations are represented in a Local Site execution and to improve data availability. Such
Physical Schema (LSPS). This assumption fits strategy is feasible when dimension tables are
well with the idea of domain autonomy, which is small (this also facilitates system management).
one of the grid’s characteristics. But when large dimension tables are present, they
In the generic grid-based DW, nodes from can be fragmented both at intra-site and inter-sites
any site can load data to the database. But the levels in order to improve performance and QoS-
same data cannot be loaded from distinct sites. levels. Intra-site dimension table fragmentation
This leads to the idea that each piece of data has strategy depends on the locally chosen physical
a single site (to which we call Data Source Site) allocation strategy (which is dependent on the
that is its primary source. In order to reduce data type of locally available resources, as discussed
movement across grid’s sites (considering the earlier). Inter-sites large dimension tables’ frag-
abovementioned geographically related access mentation should be done using a strategy similar
patters), each site should maintain a copy of the to the one of facts table fragmentation: initially,
facts data it has loaded into the DW (in this chapter, dimension data should remain at its Data Source
we consider that tables in the LM are organized in Site. Inter-site replication is done when necessary.
a star schema). This generates a globally physi- Derived partitioning of the facts table can also be
cally partitioned facts table which uses the values done, improving the system’s performance as join
of a site source attribute as partitioning criteria. operations can be broken into subjoins that are ex-
Depending on the implementation, the site ecuted in parallel at distinct sites. Although the use
source attribute values may be combined with of facts table derived partitioning depends on the
values of other existent dimensions. In fact, the semantics of stored data, such kind of partitioning
repartitioning of each facts table site source-based should be used together use the aforementioned
fragment into several smaller fragments can partitioning based on the site source attribute.
benefit the system in several ways. For instance, In the case of large dimension tables’ frag-
in such situation, each smaller fragment can be mentation, some data replication may also oc-
907
Figure 2. Examples of benefits on the use of smaller facts table fragments at the global level
cur. For instance, let’s consider a grid-DW of a Derived facts table partitioning may be used. Each
nation-wide retail store. There are several sites site stores the partitions from which the site is
participating in such grid-DW, each one at a the Data Source Site. Large dimension tables are
distinct state. In such warehouse, a (large) dimen- fragmented and small dimension tables replicated
sion table stores information about customers. at all sites. A fragmented site source dimension
Such table may be fragmented according with table (each site storing only its own information)
the location in which the customer buys. Initially, should be used. Such data distribution strategy is
each customer’s information would be at a single represented in Figure 3. Facts table’s fragments are
site. But when a certain customer travels (or replicated across grid’s sites in order to improve
moves) to another state and buys at stores from performance and availability.
such state, his/her information may also appear
in the state’s database. When there is dimension the Qos-Oriented Query scheduling
table (fragment) replication at distinct sites, a
replica consistency strategy may be necessary. Users submit queries to the grid-based DW
There are several works in the literature about considering the Logical Model. Ideally, the
algorithms to efficiently maintain replica con- physically used data distribution strategy should
sistency in distributed and grid-based databases be transparent to users. Hence, the first phase in
[e.g., (Akal et al, 2005; Breitbart et al, 1999; query scheduling is transforming the submitted
Chen et al, 2005)]. query (job) into other ones (tasks) that access
Hence, in the GPS, the facts table is partitioned the physically distributed global relations. Such
by the combination of the site source attribute with transformation is similar to the ones presented at
the other most frequently used equi-join attributes. (Furtado, 2004b).
908
Figure 3. Global physical allocation example
The generated tasks are assigned to sites by the Local Scheduler should transform the glob-
a Community Scheduler. At each site, a Local ally specified query into the ones that should be
Scheduler is responsible to manage task execu- locally executed. Besides that, results merging
tion. If domain specific physical layout is used, should also be done at local site level in order to
the Local Scheduler must also do the conversion send back a single result, which is correspondent
of the globally specified task into other queries to the query (task) the site has received.
that access the physically existent relations. The
Community Scheduler specifies the necessary
requirements (execution deadlines) of each task EVALUAtING IF A sLO
(rewritten query) in order to execute the user’s cAN bE AcHIEVED
query by the desired QoS level. Then, there is a
task execution negotiation phase among the Com- Let’s consider a user-submitted query Q with a
munity Scheduler and Local Schedulers, in order deadline interval d. The system must estimate
to verify if each of the rewritten queries (tasks) the query’s total execution time (tet) and com-
can be executed by its deadline. If any task can- pare it with d in order to verify if the query can
not be executed by the specified deadline, then or cannot be executed by its deadline. When the
user is notified that the required SLO cannot be system estimates that the execution can be done
achieved and a new one must be specified, or the according to the specified SLO (d ≥ tet), it starts
query execution would be canceled. query execution. Otherwise, query execution is not
The Community Scheduler assigns queries started and user is informed that the established
to sites considering the Global Physical Model. SLO cannot be achieved.
Hence, if local domain uses some data partitioning In order to predict the tet value, the system
policy different from the one used at the global estimates the execution time of each of the query’s
level (e.g. large dimension table partitioning, as tasks (task finish time – tft), considering three key
in the abovementioned NPDW strategy), then time components for each query (task):
909
(i) The query execution time at a local site (local On the other hand, the Community Scheduler
execution time - let); does not have control of intra-set data placement
(ii) The necessary time to transfer required data and query execution. Therefore, it is somewhat
to the site (data transfer time – dtt); difficult to make such module estimate tasks’
(iii) The necessary time to transfer the query’s execution time. Such estimation is done by local
results back from the chosen site (results schedulers. In fact, in QoS-oriented scheduling,
transfer time – rtt). the Community Scheduler does not have to know
the exactly necessary time to execute a query:
The tft value of a single task at a certain site local schedulers must commit themselves to
is computed by Equation 1. An upper bound es- execution the assigned queries by a certain time
timated value for the users’ query execution (tet) interval. Such interval is the maximum value that
is obtained by Equation 2. let (mlet) can assume in order to finish the user’s
query execution by the specified SLO.
tft = let + dtt + rtt (1) Hence, for each task, the Community Scheduler
computes the mlet value (Equation 4) and uses
tet = Max(tft) (2) such value as a task deadline when negotiating
with local schedulers.
To estimate the value of tft, the Community
Scheduler must have some estimative of its com- mlet ≤ d − (dtt + rtt ) (4)
ponents. First of all, it predicts the values of dtt
and rtt, with the support of a grid infrastructure Figure 4 presents a general view of the SLO-
network monitor tool [like the Network Weather aware scheduling model.
Service – NWS - (Wolski, 1997)]. Such tool
is used to predict network latency (L) and data
transfer throughput (TT) between sites. The Com- LOcAL scHEDULErs AND
munity Scheduler uses such predicted values for sErVIcE LEVEL AGrEEMENts
network characteristics together with estimated
dataset sizes (obtained by database statistics) to When estimating if a user’s query can be executed
predict dtt and rtt (a predicted transfer time (tbs) by the proposed deadline, the Community Sched-
of a dataset of size z between sites i and j can be uler must consider the necessary time to execute
obtained by Equation 3). each rewritten query at local sites (let). But not
only the Community Scheduler does not have
 z  total control of the execution environment, but
tbsi , j = L +  
TT  (3)
Figure 4. General view of the SLO-aware scheduling model
910
also each site can have local domain policies that tion are QoS-oriented, which means they intend
can constraint the use of local resources by remote to increase the system’s SLO-achievement rate
users. Therefore, the necessary time to execute (SLO-AR).
each task should be predicted by local schedulers. The SLO-AR (Costa & Furtado, 2008b) is
But in QoS-oriented scheduling, each site a performance metric that aims to indicate how
may not inform the Community Scheduler the well a system is performing on executing jobs by
exact predicted query execution time. On the specified service level objectives. It is defined
other hand, local schedulers should commit as the relation between the number of queries
themselves to execute the negotiated query by whose execution finishes by the required deadline
a certain deadline (mlet) that is specified by the (N) and the number of queries in the submitted
Community Scheduler. When a local scheduler workload (W).
agrees to execute a query by a certain deadline, Therefore, the SLO-Oriented replica selection
it makes a SLA (Service Level Agreement) with and placement strategy aims at increasing the
the Community Schedule. number of queries that the system executes by
When a SLA is signed, the local scheduler is the specified SLOs. In order to do that, a benefit-
committed to execute the query by the negotiated based mechanism is implemented by a Replication
deadline. But it can, for instance, reorder local Manager (RM), as it is described below.
query execution (do not have to execute each of the
incoming queries as fast as possible) or change the
number of queries that are concurrently executed INtEr-sItE QOs-OrIENtED
at the local DBMS (multi-programming degree). DyNAMIc rEPLIcAtION
The QoS-OES (Costa & Furtado, 2008) sched-
uler is an example of QoS-oriented query scheduler The RM monitors the number of times that a
that can be used in such context. Such module SLO-objective cannot be achieved due to the
is a generic external scheduler that is used as a inexistence of a certain dataset (e.g. facts table
middleware between the DBMS and its users. The fragment) at a certain site and computes the total
QoS-OES is capable to estimate query execution benefit value (β) that the system would have if
time in a multi-query environment and to commit a replica of such dataset is created. When β is
itself to execute a submitted query by a certain greater than a threshold value, the system considers
deadline, as soon as the specified deadline time is the possibility of creating the data replica at the
greater than the predicted query execution time. evaluated site (at this point we refer that replicas
of input datasets are created. Latter in this chapter,
caching and replication for Qos we discuss how to determine if such datasets are
fact’s table fragments or computed chunks).
Grids are highly heterogeneous environments In order to implement such policy, when the
in which the use of database replicas can lead Local Scheduler cannot execute the task (query)
to great performance improvement and to high by the specified mlet time, it should evaluate if it
QoS levels. But as the problem of choosing the would achieve the specified deadline if a replica
optimal number of database replicas and doing of a given dataset is present at the site.
the optimal placement of such replicas into the When the Local Scheduler predicts that it
nodes of a distributed environment is NP-hard would achieve the required task’s deadline if a
(Loukopoulos & Ahmad, 2000), some heuristics certain replica is locally stored, it informs the
should be used. The heuristics used in this Sec- Replication Manager what is such replica. In such
911
situation, the value of β for the specified dataset that a certain deadline would be achieved if its
is incremented by a certain δ value (benefit of the site stores a local copy of a certain fragment, it
considered input data set replica to the system), as informs the Replication Manager which consid-
represented in Equation 5. However, the δ value ers such fragment as a dataset that is candidate
of each task should vary over time, in order to for inter-site replication. Such fragment may be
differentiate the benefit for old queries from the replicated or not depending on its benefit for the
ones for newer queries. Therefore, a time discount system.
function may be used in order to compute δ, as As discussed before, each site is autonomous
presented in Equation 6. to implement its own data placement (and repli-
cation) strategy. Besides that, each site may also
β = ∑ δi (5) implement its own data caching mechanism. There
i are some caching mechanisms that are benefited
 ∆t  by the multidimensional nature of warehouse data.
− 
δi = e
 λ 
(6) Chunked-based caching (Deshpande et al, 1998;
Deshpande & Naughton, 2000) is one of those
In Equation 6, Δt represents the time window specialized mechanisms for the DW.
between the task execution time (of the query that In chunk-based caching, DW data to be stored
would be benefited by the input dataset replication) in the cache is broken up into chunks that are
and current time and λ enables the use of different cached and used to answer incoming queries.
time intervals [as defined in (Huynh et al, 2006)]. The list of necessary chunks to answer a query is
Whenever β is greater than a threshold value broken into two: (i) chunks that may be obtained
for a certain dataset/site, the site is marked as a (or computed) from cached data; and (ii) chunks
candidate to receive a replica of the considered that have to be recovered from the data warehouse
dataset. Indeed, the replica is immediately cre- database. In such method, sometimes it is possible
ated if there is enough disk space. Otherwise, to compute a chunk from chunks from different
the system would have to evaluate if some of the levels of aggregation (each aggregation level cor-
existent data replicas (of another datasets) should responds to a group-by operation) (Deshpande &
be replaced or not by the new replica candidate. In Naughton, 2000).
order to do that, RM also maintains the benefits Such computed chunk based mechanism may
score of existing dataset replicas. Such score is be implemented by local schedulers to implement
computed in the same way that β and δ values of local caching. But chunks of results or computed
inexistent replicas are computed. If the β value chunks can also be considered as candidates for
of an existing replica is lower than the one of a replication by Replication Manager. In such con-
replica candidate, then a replica replacement is text, when a local scheduler evaluates that a certain
done. Otherwise, the system maintains the already missing chunk would enable the site to execute a
existing replicas. task by a certain deadline that would be achieved
without such chunk, the local scheduler must send
to the Replication Manager the identification of
LOcAL cAcHING AND INtrA- such chunk. Then, the RM would consider the
sItE rEPLIcA cANDIDAtEs chunk as a dataset candidate for replication in
its benefit-based dynamic replica selection and
Facts’ table fragments are natural candidates for placement mechanism.
inter-site replication, as discussed previously in
this Chapter. When a Local Scheduler evaluates
912
EXPErIMENtAL EVALUAtION topology. In such topology, there is no central

root site (eliminating a possible bottleneck in the
The QoS-oriented scheduling and dynamic repli- system) and each site is directly connected to two
cation mechanisms were experimentally evaluated other sites. Data movement in the rings is done in
in a simulation environment. The experimental opposite directions. In all the tests, we consider
setup is composed by 11 sites, which were inspired a data transfer rate of 50Mbps and a latency of
in the experimental testbed used in (Sulistio et al, 10 milliseconds.
2007) and in the LHC Computing Grid Project The considered query workload is composed
sites (Bird et al, 2005). Figure 5 presents the main of 1,000 tasks (re-written queries). Tasks’ sizes
characteristics of the used sites. vary about 2,000 kMIPS ± 30%, which means
A Star Schema-based DW is considered. Facts that a typical task execution would take about 30
table is partitioned into 121 fragments. Fragments minutes in the least powerful site and 40 seconds
sizes are generated considering Pareto’s power-law in the most powerful one. Figure 5 presents the
distribution [which fits well for data grids files number of queries submitted at each site. In order
(Sulistio et al, 2007)], with a 1Gb mean fragment to model that, we consider that each query may be
size. Initially, each site stores the same number of submitted by a user from any site, but the prob-
facts table fragments (11 fragments). ability of a query being submitted by a specific
Two distinct network topologies are used in our site is proportional to the number of DW users
tests (represented in Figure 6). The first one is a at the site. Users’ access patterns were modeled
hierarchical topology, in which sites are organized considering that half of the tasks access data stored
in a binary tree according to their ids [hierarchical at the same site that submitted the job.
topologies are also considered in real projects, like In order to evaluate the effectiveness of the
the LHC Computing Grid Project - data storage QoS-aware scheduling and dynamic replication
and analysis project for CERN’s Large Hadron strategies, we have made several tests using the
Collider (Bird et al, 2005)]. The second network QoS-oriented scheduling and three different data
model is a nonhierarchical topology, dual ring dynamic replication strategies: (i) the QoS-aware,
Figure 5. Experimental setup description
913
Figure 6. Experimentally tested network topologies
(ii) Best Client (BC) and (iii) Data Least Loaded The obtained SLO-AR values are represented
(DLL). Variations of the BC strategy are used by in Figure 7. All the evaluated dynamic replication
Ranganathan & Foster (2001) and Siva Sathya methods lead to good SLO-AR. Such success
et al (2006). The DLL is used by Ranganathan & ensures the quality of the used QoS-oriented
Foster (2004). Both in the BC and in DLL, each hierarchical scheduling model. But the obtained
site monitors the number of data access at the data results also present the benefits of the proposed
replicas it stores, computing the number of times QoS-aware dynamic data replication and place-
the fragment was requested. When such number is ment, as it was the method that leads to the highest
greater than a threshold value, a fragment replica value of SLO-AR.
is created at another site. In BC, the fragment In Figure 8 we present the measured through-
replica is created at the site that has more times put for the evaluated strategies. Once again, the
demanded for the considered fragment. In DLL, proposed QoS-aware replication strategy leaded
fragment replica is created at the site that has the to the highest values in the two tested network
least remaining work to execute. configurations. This happens because such place-
Figure 7. SLO-achievement rate
914
ment strategy has the same objective of the used In Figure 10, we present the number of created
query scheduling strategy. Therefore, not only it replicas per site for the Hierarchical Topology.
increases the number of queries that the Commu- Let’s analyze site 6: such site is almost the most
nity Scheduler agrees to execute but also lead to powerful one and it is the site that have the high-
a better resource utilization than the other replica est number of submitted queries. This is also the
selection and placement methods. site on which the BC method created the highest
The number of created replicas per method was number of data replicas. But as the BC method
almost the same and no method created a huge places too much replicas at such site, the scheduler
number of replicas for the same data fragment assigns too much queries for site 6 (what puts it
(as show in Figure 9). In fact, the success of the into a too high load situation and decreases per-
QoS-aware scheduling method is mostly related formance, leading to a lower SLO-achievement
to its replica placement strategy. rate) or schedules queries to other sites that are
Figure 8. Measured throughput
Figure 9. Number of replicas per facts table fragment - hierarchical topology
915
Figure 10. Number of facts table fragments at each site in hierarchical topology
near site 6 but that do not have the same data cONcLUsION
replicas (in such case, data movement is done
during query execution, which also decreases the Grid-based data warehouses are useful in many
system performance). global organizations which generates huge vol-
On the other hand, the QoS-aware placed umes of distributed data that should be transpar-
high number of replicas at sites 3-5 than the other ently queried by the organizations participant’s.
methods. Such sites are of medium powerful ones In such environment, the Grid is used as basic
and are relatively near of sites 1, 2 and 6 (the infrastructure for the deployment of a large dis-
three sites with the highest number of submitted tributed mostly read-only database. But due to
queries). This way, many of the queries submit- the grid’s special characteristics, like resource
ted at sites 1, 2 and 6 can be executed at sites 3-5 heterogeneity, geographical dispersion and site
with a good performance (no database copy is autonomy, the efficient deployment of huge data
done during execution, only results - which are warehouses over grid-connected sites is a special
relatively small in size - are transferred through challenge.
sites during query execution). In contrast, the In this chapter, we present QoS-oriented
DLL strategy placed many replicas on not so scheduling and distributed data placement strat-
powerful sites (like 10 and 7). This happened as egies for the grid-based warehouse. We discuss
the DLL strategy evaluated that such sites had a the use of a physically distributed database, in
small size of pending work. But such sites take which tables are both partitioned and replicated
too much time to execute a query, and the Com- across sites. The use of facts’ table partitioning and
munity Scheduler rarely assigns query execution replication is particularly relevant as grid users’
for such sites. Therefore, the replicas created by queries may follow geographical related access
DLL were somehow of less utility to the system. patterns. Inter-site dimension tables fragmenta-
tion and replication are done in order to achieve
good performance in query execution but also to
reduce data movement across sites, which is a
costly operation in grids.
916
Incoming queries are rewritten into another Alpdemir, M., Mukherjee, A., Paton, N., Watson,
ones (tasks) that are assigned to sites by a Local P., Fernandes, A., Gounaris, A., & Smith, J. (2003).
Scheduler based on Service Level Agreements OGSA-DQP: A service-based distributed query
between the Community Scheduler and Local processor for the Grid. In Proceedings of UK e-
Schedulers The use of a hierarchical scheduling Science All Hands Meeting.
model leads to good SLO-achievement rates and
Bird, I. & The LCG Editorial Board (2005). LHC
also maintains site autonomy, as each site’s Lo-
Computing Grid Technical Design Report [LCG-
cal Scheduler may implement its own scheduling
TDR-001, CERN-LHCC-2005-024].
strategy.
Dynamic data replication is very important in Breitbart, Y., Komondoor, R., Rastogi, R., Seshadri,
grid-based data bound jobs. In the grid-enabled S., & Silberschatz, A. (1999). Update propagation
warehouse, dynamic replication of facts table protocols for replicated databates. SIGMOD Re-
fragments or of computed chunks is important cord, 28(2), 97–108. doi:10.1145/304181.304191
to improve the systems’ performance. The QoS-
Chaudhuri, S., & Dayal, U. (1997). An over-
oriented dynamic replica selection and placement
view of data warehousing and OLAP tech-
is especially important to increase the SLO-
nology. SIGMOD Record, 26(1), 65–74.
achievement rate in grid-enabled warehouses.
doi:10.1145/248603.248616
Chen, G., Pan, Y., Guo, M., & Lu, J. (2005). An
rEFErENcEs asynchronous replica consistency model in data
grid. In Proceedings of Parallel and Distributed
Akal, F., Türker, C., Schek, H., Breitbart, Y., Grabs, Processing and Applications - 2005 Workshops
T., & Veen, L. (2005). Fine-grained replication (LNCS 3759, pp. 475-484).
and scheduling with freshness and correctness
guarantees. In Proceedings of the 31st interna- Chervenak, A. L., Palavalli, N., Bharathi, S., Kes-
tional Conference on Very Large Data Bases selman, C., & Schwartzkopf, R. (2004). Perfor-
(pp. 565-576). mance and scalability of a replica location service.
In Proceedings of the 13th IEEE international
Akinde, M. O., Böhlen, M. H., Johnson, T., Lak- Symposium on High Performance Distributed
shmanan, L. V., & Srivastava, D. (2002). Efficient Computing (pp.182-191).
OLAP query processing in distributed data ware-
houses. In Proceedings of the 8th international Costa, R. L. C., & Furtado, P. 2008. A QoS-oriented
Conference on Extending Database Technology: external scheduler. In Proceedings of the 2008
Advances in Database Technology. (LNCS 2287, ACM Symposium on Applied Computing (pp.
pp. 336-353. 1029-1033). New York: ACM Press.
Allcock, W., Bresnahan, J., Kettimuthu, R., & Costa, R. L. C., & Furtado, P. (2008). QoS-oriented
Link, M. (2005). The Globus striped GridFTP reputation-aware query scheduling in data grids. In
framework and server. In Proceedings of the Proceedings of the 14th European Conference on
2005 ACM/IEEE Conference on Supercomputing Parallel and Distributed Computing (Euro-Par).
(pp.54-65). Costa, R. L. C., & Furtado, P. (2008). Scheduling
in Grid databases. In Proceedings of the 22nd
international Conference on Advanced informa-
tion Networking and Applications – Workshops
(pp. 696-701).
917
Deshpande, P., & Naughton, J. F. (2000). Aggre- Huynh, T. D., Jennings, N. R., & Shadbolt, N.
gate aware caching for multi-dimensional queries. R. (2006). An integrated trust and reputation
In Proceedings of the 7th international Conference model for open multi-agent systems. Autonomous
on Extending Database Technology: Advances in Agents and Multi-Agent Systems, 13(2), 119–154.
Database Technology vol. 1777 (pp. 167-182). doi:10.1007/s10458-005-6825-4
Deshpande, P. M., Ramasamy, K., Shukla, A., & Krauter, K., Buyya, R., & Maheswaran, M.
Naughton, J. F. (1998). Caching multidimensional (2002). A taxonomy and survey of grid resource
queries using chunks. In A. Tiwary & M. Franklin management systems for distributed computing.
(Eds.), Proceedings of the 1998 ACM SIGMOD Software, Practice & Experience, 32(2), 135–164.
international Conference on Management of Data doi:10.1002/spe.432
(Seattle, Washington, United States, June 01 - 04,
Lawrence, M., & Rau-Chaplin, A. (2006). The
1998) (pp. 259-270). New York: ACM Press.
OLAP-Enabled Grid: Model and query process-
Foster, I. Kesselman, C. Nick, J., & Tuecke, S. ing algorithms. In Proc. of the 20th international
(2002). The physiology of the grid: An open grid Symposium on High-Performance Computing in
services architecture for distributed systems in- An Advanced Collaborative Environment (HPCS).
tegration (Globus Project Tech Report).
Lima, A., Mattoso, M., & Valduriez, P. (2004).
Foster, I., Kesselman, C., Tsudik, G., & Tuecke, S. Adaptive virtual partitioning for OLAP query
(1998). A security architecture for computational processing in a database cluster. In Proceedings
grids. In Proceedings of the 5th ACM Conference of the Brazilian Symposium on Databases (SBBD)
on Computer and Communications Security. CCS (pp. 92-105).
‘98 (pp. 83-92).
Lin, Y., Liu, P., & Wu, J. (2006). Optimal placement
Foster, I. T. (2001). The anatomy of the grid: En- of replicas in data grid environments with locality
abling scalable virtual organizations. In Proceed- assurance. In Proceedings of the 12th international
ings of the 7th international Euro-Par Conference Conference on Parallel and Distributed Systems
on Parallel Processing (LNCS 2150, pp. 1-4). - Vol 1 (2006) (pp. 465-474).
Furtado, P. (2004). Workload-based placement Loukopoulos, T., & Ahmad, I. (2000). Static and
and join processing in node-partitioned data ware- adaptive data replication algorithms for fast in-
houses. In Proceedings of the 6th International formation access in large distributed systems. In
Conference on Data Warehousing and Knowledge Proc. of the 20th Intern. Conference on Distributed
Discovery (LNCS 3181, pp. 38-47). Computing Systems (ICDCS).
Furtado, P. (2004). Experimental evidence on Nieto-Santisteban, M. A., Gray, J., Szalay, A.,
partitioning in parallel data warehouses. In Pro- Annis, J., Thakar, A. R., & O’Mullane, W. (2005).
ceedings of the 7th ACM international Workshop When database systems meet the grid. In CIDR
on Data Warehousing and OLAP (pp. 23-30). (pp. 154-161).
Haddad, C., & Slimani, Y. (2007). Economic Park, S., & Kim, J. (2003). Chameleon: A resource
model for replicated database placement in Grid. scheduler in a data grid environment. In Proceed-
In Proceedings of the Seventh IEEE international ings of the 3st international Symposium on Cluster
Symposium on Cluster Computing and the Grid Computing and the Grid. IEEE Computer Society.
(pp. 283-292). IEEE Computer Society.
918
Poess, M., & Othayoth, R. K. (2005). Large Wolski, R. (1997). Forecasting network perfor-
scale data warehouses on grid: Oracle database mance to support dynamicscheduling using the
10g and HP proliant servers. In Proc. of the 31st network weather service. In Proceedings of the
international Conference on Very Large Data 6th IEEE international Symposium on High Per-
Bases (pp. 1055-1066). formance Distributed Computing (August 05 - 08,
1997) (pp. 316). IEEE.
Ranganathan, K., & Foster, I. (2004). Computa-
tion scheduling and data replication algorithms for
data Grids. In Grid resource management: State of
the art and future trends (pp. 359-373). Norwell, KEy tErMs AND DEFINItIONs
MA: Kluwer Academic Publishers.
Community Scheduler: It is a specialized
Roy, A., & Sander, V. (2004). GARA: A uniform middleware, responsible for matching users’ jobs
quality of service architecture. In Grid resource requirements with the available resources in a
management: State of the art and future trends grid by the interaction with local schedulers. It
(pp. 377-394). Norwell, MA: Kluwer Academic assigns jobs to sites through a process that, besides
Publishers. requirement matchmaking, can also comprise
Siva Sathya, S., Kuppuswami, S., & Ragupathi, some kind of negotiation with local schedulers.
R. (2006). Replication strategies for data grids. Sometimes it is also called Resource Broker or
International Conference on Advanced Comput- Meta-scheduler.
ing and Communications. ADCOM 2006 (pp Data Grid: A Grid environment whose ser-
123-128). vices are mainly used to deal with (including to
store, process, replicate and move) huge volumes
Stöhr et al, 2000 Stöhr, T., Märtens, H., & Rahm, of distributed shared data or over which are ex-
E. 2000. Multi-Dimensional Database Allocation ecuted grid-based applications that consume or
for Parallel Data Warehouses. In Proceedings of generate huge volumes of data.
the 26th international Conference on Very Large Grid: The term Grid is a basic infrastructure
Data Bases. 273-284. used to interconnect and provide access to widely
Transaction processing council benchmarks distributed, and possibly heterogeneous, shared re-
(2008). Retrieved from http://www.tpc.org sourced that may belong to distinct organizations.
Grid Resource Management System: It is the
Venugopal, S., Buyya, R., & Ramamohanarao, resource management system that runs over the
K. (2006). A taxonomy of Data Grids for dis- grid and is used to manage the available shared
tributed data sharing, management, and pro- resources, providing a wide range of services (like
cessing. ACM Computing Surveys, 38(1), 3. efficient data movement, replica management
doi:10.1145/1132952.1132955 and remote job submission and monitoring) to
Watson, P. (2001). Databases and the grid. UK grid-based applications.
e-Science Technical Report Series. Grid-Enabled Databases: A set of Database
Management Systems (DBMS) which are physi-
Wehrle, P., Miquel, M., & Tchounikine, A. (2007). cally distributed and are queried by grid users
A grid services-oriented architecture for efficient through the use of a middleware together with a
operation of distributed data warehouses on Grid Resource Management System.
Globus. In Proceedings of the 21st international Quality-of-Service (QoS): The term was first
Conference on Advanced Networking and Ap- coined in the networking-related field in order to
plications (AINA) (pp. 994-999). identify the ability of a certain technology to do
919
resource reservation in order to provide different characteristics (like performance and availabil-
priority to distinct applications or users. More ity) of the provided services. It may also define
recently, it can also be used as a users’ satisfac- guarantees and penalties (for the case of non
tion degree or the ability to provide predictable compliance with the SLA).
performance levels that are according to users’ Service Level Objective (SLO): It is a tar-
expectations. get value used to measure the performance of a
Service Level Agreement (SLA): An agree- service provider in what concerns to a specific
ment that is firmed between a service provider characteristic, like response time or throughput.
and a service consumer, and which defines the Its definition may also contain information about
service levels (possibly in terms of Service Level how the SLO is measured and the measurement
Objectives) that should be provided for several period.
plex Construction, edited by Ladjel Bellatreche, pp. 150-170, copyright 2010 by Information Science Reference (an imprint
of IGI Global).
920
921
Chapter 4.2
EIS Systems and Quality
Management
Bart H.M. Gerritsen
TNO Netherlands Organization for Applied Scientific Research, The Netherlands
This chapter discusses the support of quality Quality Management systems

management by Enterprise Information Systems.
After a brief introduction in ISO9001, one of the Over the last decades, enterprises and other organiza-
principle and widest-spread quality management tions from large to small have come to implement
frameworks, this chapter discusses the design and quality management systems (QMS). Large Scale
implementation of a typical QMS and in particu- Enterprises (LSE’s) and Small and Medium Enter-
lar of key performance indicators, indicating the prises (SME’s) alike, decided to apply QMS to get
present state of performance in the organization. grip on the product and business process quality
While analyzing design and implementation issues, level customers nowadays expect. Many SME’s
requirements on the supporting EIS system will be initially did so “because customers ask for it”. While
derived. Finally, the chapter presents an outlook customer satisfaction is a pivotal factor indeed, learn-
onto future developments, trends and research. This ing to master and apply quality principles correctly
chapter reveals that key performance indicators also assists in increased employee involvement and
can be well integrated in EIS systems, using either productivity, preventing defects from occurring,
relational or object-oriented storage technology. reducing costs and production times. The key to
achieving this is a timely and correct alignment of
DOI: 10.4018/978-1-60566-892-5.ch017
EIS Systems and Quality Management
the delivered quality in business processes at all ISO 9001 does not prescribe any quality
levels in the organization, board to shop floor. The management system in particular but frames the
information needed to know and control quality process of designing, implementing and operating
performance goes hand in hand with other daily one, defining guiding principles, requirements
operational information within the organization and and key elements it ought to contain for proper
consequently, quality information will typically functioning: the what to, not the how to. Orga-
be residing in emerging Enterprise Information nizations can tailor and scale a QMS framework
Systems (EIS). This is why in this chapter we will to their own needs and chose the implementation
discuss quality management within the context they see fit, as long as the standardized good qual-
of EIS systems, seen from the angle of SME’s. A ity management practices remain honored. The
QMS is not the same as an information system; an detailed design and operating of a QMS is critical
information system (e.g., an EIS system) supports to its success, however, and ultimately critical to
the implementation of a QMS. the success of the organization as a whole.
IsO 9001 research Questions and Approach
One of the principle and widest-spread standards Designing a fit-for-purpose QMS requires thor-
to design and implement a QMS is ISO 9001, ough understanding of business strategy and busi-
belonging to the ISO 9000-family of standards. ness processes and the readiness to align the QMS
The most recent version of this standard is ISO with the business processes, vice versa. Generally,
9001:2008. Figure 1 based on the 2006 ISO an- alignment and fine tuning is something for the long
nual survey figures (ISO, 2006) shows the world haul, and of continuous concern. Niven, in (Niven,
wide adoption of the standard. 2005) estimates that at present at most 10% of
Figure 1. Global uptake of ISO 9001:1994 (solid bars) and 9001:2000 (hatched bars) up to 2006; China
is now the country with the largest number of ISO 9001-based QMS (approx. 200000), comparable to
Europe as a whole. Source: (ISO, 2006).
922
the organizations actually achieves their strategic • How to store recorded quality performance
objectives. The problem is often (particularly for in the EIS system?
small SME’s) that expertise is lacking and attention • How to determine correct timing and pre-
slips away after the QMS has been introduced; dictive power for lead quality performance
compare (Woodhouse, 2004). This is like buying indicators, allowing off-target perfor-
an advanced piece of equipment without learning mance and defects to be remedied before
to operate it optimally: the cost is taken while the repair becomes impossible?
deeper benefits never come in sight. Ultimately, • How to determine whether the hierarchy
quality is something to be embedded and fused of performance indicators coherently and
with daily operational processes in order to be unambiguously supports strategic control?
effective and efficient; quality drives productivity. • How to attain a flexible QMS implementa-
tion in which organizational change is ad-
Research Questions opted swiftly and which allows for appro-
priate up- and downscaling of activities?
In view of emerging EIS systems, the question • How to retrieve the correct actual qual-
is how to assure proper and lasting alignment ity performance data again from the EIS
and how to obtain a coherent view on quality system;
performance across the organization, such that it • How to archive and maintain a history of
can be managed and kept inline with the quality quality performance data within the EIS
targets defined. More specifically, research ques- system, allowing reconstruction at any one
tions arising, are: time in the past of the then present quality
performance?
• How to define a quality performance strat- • How to effectively and efficiently build up
egy and measurable quality performance an evidence-based track record of qual-
targets; how to store this typically unity performance within the EIS system,
structured strategy description in an EIS for auditing, approval and endorsement
system? purposes?
• How to measure actual quality perfor- • How to combine secondary EIS data with
mance of each of the business processes? quality performance data in root cause
• How to design and implement the quality analysis so as to explore the deeper causes
performance indicators (often referred to of ill-performance?
as: key performance indicators, or KPI’s) • How to combine performance data across
for each of the business processes in the organizational borders, in case of supply
organization, including strategic control chain, delivery chain or partners delivering
processes? bundled products;
• How to determine a working definition, a
proper format and a fit-for-purpose accu- Quality in this context is not just product qual-
racy for each of the performance indica- ity; it is also about controlling all processes so that
tors; does the EIS systems support storage the outcome is under control and about manage-
of this format? ment aspects targeting customer satisfaction. A
• How to compute and aggregate actual more accurate definition of quality will be given
quality performance figures from EIS sys- further down. A (key) performance indicator is a
tem residing data? metric expressing the result of a measurement of
923
the performance of a business process. KPI’s take • EIS systems do support the access rights
the form of business data typically processed and and roles to grant all employees proper ac-
stored in a database on a computer system. Key cess to the shared parts of the QMS and
performance indicators will be explained in detail, quality performance information, and to
in the sections ahead. The term key performance block improper modifications or manipu-
indicator (KPI) has become so common in practice lations of managed data;
that we also use KPI to refer to quality perfor- • EIS systems are efficient enough to sup-
mance. A lead performance indicator expresses a port recording of performance data and up-
future performance, for instance expected annual and downscaling efficiently;
turnover based on current performance. Lead in- • EIS systems can combine quality perfor-
dicators and lag indicators (which record results mance data of suppliers, wholesalers, re-
in the past) will also be discussed in detail in the tailers, etc., all having impact on customer
sections ahead. perceived quality and satisfaction;
In the past two decades, great advancements • EIS systems support the kind of transac-
have been made with respect to the above research tions needed to operate a QMS;
questions, see for instance (Adams & Frost, 2008; • EIS systems support reporting of qual-
Ahmad & Dhafr, 2002; Cobbold & Lawrie, 2002b; ity performance at all levels in the
Hernandez-Matias, Vizan, Perez-Garcia, & Rios, organization;
2008; Hubbard, 2007; Kaplan & Norton, 1996b; • EIS systems require different skills when
Kaplan & Norton, 2001b; Kaplan & Norton, containing quality performance data;
2001a; Kaplan & Norton, 2001a; Kaplan & Nor-
ton, 2006; Kaplan & Norton, 2004a; Niven, 2005; This chapter addresses these issues.
Woodhouse, 2004). None of these contributions In order to understand the requirements on the
discusses the above research questions in the EIS system we need to understand how quality
context of EIS systems, however. Consequently, performance is measured, the role of key per-
it is unclear whether: formance indicators and bad performance alerts
therein and finally the role of root cause analysis,
• EIS systems are suited to store structured in case management intervention is needed.
and unstructured pieces of data and infor-
mation related to a QMS; Research Approach
• EIS systems adequately support storage of
causal and other relationships; Approaches to design effective QMS systems
• EIS systems allow lead indicators to quan- covering all relevant aspects in an organization
titatively forecast future scores, trends and have been primarily developed in the nineties and
performance; the beginning of this century (see the early work
• EIS systems can provide evidence-based in (Kaplan & Norton, 1992). Kaplan and Norton
quality track records; can it keep track of take the constantly transforming organization and
approvals and results of plan-do-check-act the organization’s strategic planning as the start-
cycles for instance; can it reconstruct qual- ing point, transforming organizational strategy
ity performance at any one moment back into measurable strategic objectives to which a
in time, for instance in case of a customer target can be assigned. A Strategy Map (Kaplan
claim or complaint? Can it show (evidence & Norton, 2004a; Kaplan & Norton, 2004b; Ka-
of) continuous improvement over a certain plan & Norton, 2004c; Kaplan & Norton, 2004c)
period for instance? and a Destination Statement (Cobbold & Lawrie,
924
2002a) define the strategic change desired and a guiding principles and main elements. After this
time frame for the transition, while a Balanced background information, we further focus on the
Score Card (BSC) is commonly used to combine design and characteristics of KPI’s needed to learn
different views and aspects of the transition in a the requirements for their embedding in an EIS
single overview (Cobbold et al., 2002b; Kaplan system. The characteristics of properly designed
et al., 1992; Kaplan & Norton, 1993; Kaplan & KPI’s will be discussed, with special attention to
Norton, 1996a; Kaplan et al., 1996b; Kaplan et the dynamics (the timing, say) of lead indicators.
al., 2001a; Kaplan et al., 2001b). This makes the Concluding this chapter, an outlook onto the future
BSC one of the output forms, to be delivered by of performance-based management will be given.
the EIS system; more on this later.
Strategic objectives to transform the organiza-
tion and the products and services it delivers, can IsO 9001 OVErVIEW
only be successfully accomplished if its critical
success factors (CSF’s) are satisfied. Causal rela- brief History of IsO 9001
tionship analysis reveals how one critical success
contributes to the next critical success. Critical Quality initiatives go as far back as the fifties and
success factors can be monitored using KPI’s. In Japan is generally seen as the cradle of industrial
fact, this cascade shows how strategic planning quality programming. Deming’s early work on
is made measurable and manageable down to statistical process control (SPC) is generally seen
the operational level; the bottom up cascade of as the birth of quality control, one of the main
KPI’s reports how well the organization’s actual elements of quality management. The idea behind
performance contributes to realizing the desired quality management efforts is that by controlling
transformation at strategic level. This approach the variability in every process, a greater consis-
leads to very flexible, effective quantitative quality tency in output results can be obtained. During
management system implementations and com- the seventies and eighties, quality management
bining them with the efficiency of EIS systems and QMS systems gained worldwide attention.
appears very attractive. Flexibility is an important Starting out with US Army and NATO initiatives,
aspect in the context of the growing demand for most notably AQAP, the idea of quality manage-
mass customization and agile manufacturing; ment gradually invaded industry at large. Out of
more on this later. various national standardization efforts, the desire
For the research approach followed in this grew to come to a standardized worldwide qual-
chapter, we will adopt the above outlined approach ity management framework. To that extent ISO
and analyze in every step the requirements and and more specifically its TC 176/SC2 committee
consequences for the implementation in an EIS started working on this global standard, beginning
system. This chapter thus seeks to present answers of the eighties, resulting in the first version of
to the above defined research questions. Not all the ISO 9001 in the late eighties. After its initial
aspects are equally important: emphasis will be on 1987 version, ISO 9001 was updated in 1994
the design and embedding in an EIS of the KPI’s. (ISO 9001:1994) and 2000 (ISO 9001:2000) and
lastly in 2008, formally tagged ISO 9001:2008.
Organization of this chapter Today, over a million organizations worldwide
have adopted one of the above management
The remainder of this chapter is organized as fol- frameworks, with highest penetrations in the
lows: firstly, a brief overview of ISO 9001 will realms of engineering and material technologies
be presented, among other things discussing its (International Organization for Standardization
925
ISO, 2008). Some 100 countries participate in whether a private enterprise, public agency or
and/or follow the ISO 9001 standardization ef- non-governmental organization: an organization
fort nowadays, with over 150 countries being an must constantly strive to a lasting and better sat-
ISO-member and users spread across nearly 180 isfaction of its customers (shareholders, partners,
countries (Figure 1). the public, the environment …). The constant
transition organizations find themselves in is
Other Quality Management also approached as a quality controlled business
standards process (the strategic control process) and thus
subject to quality management.
With environmental objectives becoming ever Main principles contributing to the above
more compelling, ISO 9001-based QMS systems philosophy are:
are occasionally replaced or complemented by
ISO 14001-based systems. The ISO 14000 family • A written QMS must be available and ac-
of standards deals with environmental manage- cessible to every employee and customer;
ment. Total Quality Management (TQM) can • Targeted actions and performance must be
be regarded as a broader and deepened form of verified and validated periodically;
quality management. It lays strong emphasis on • The QMS must take a quantified approach:
quality awareness everywhere in the organization it must regularly record quality perfor-
and in its supply and delivery chains. It shares, mance scores (or: KPI scores) and set them
among others, the quality performance manage- off against agreed target values;
ment characteristics with ISO 9001. • A process-based approach, with plan-do-
check-act cycles (PDCA) to drive quality
Main Principles of IsO 9001 and to repair defects;
• Procedures to report and handle defects;
Satisfied customers stand central in the ISO • Active promotion, commitment and in-
9001 and all quality management is directed to volvement of senior management;
just that. Keeping customers satisfied is not just • Central roles and responsibilities shall be
a matter of products being free of deficiencies. formalized and assigned;
The customer’s explicit and implied needs and • Regular auditing (internally, externally, or
expectations should be satisfied by the product by the customer) shall be conducted;
and evidently, the product must comply with • The QMS must be effective and efficient
regulatory and generally implied sustainability and support continuous improvement.
demands. All this makes quality a subjective at-
tribute that depends on customer perception. That There is no room and no need to go into more
is why an organization must make quality explicit detail here. In the sections following, the relevant
at beforehand and approve to live up to it so that details will be further discussed in relation to their
customers can rely on the design, manufacturing, embedding in an EIS system.
environmental footprint, servicing, etc. to be inline
with their expectations. All businesses processes Main Elements of a typical
should be mastered such that process output is IsO 9001-based QMs
guaranteed to be within specified quality bounds,
contributing to a customer satisfying product, A typical ISO 9001-based QMS consists of the
now and in the future. No organization is static, following main elements:
926
• A handbook describing the QMS in detail, FrOM strAtEGy tO KPI’s

containing mandatory procedures such as
a reporting and repairing defects proce- Definitions
dure, continuous improvement and audit-
ing procedures, as well as other operational In this chapter, we will understand quality in a
procedures; context of producers and consumers of products
• A Strategy map, describing the organiza- and services. Quality is not limited to product
tion’s planned transition; quality: it extends through all business processes
• Balanced Score Card to consistently report involved in the creation of the product, from com-
on KPI scores; ponent design in the supply chain to the services
• A collection of CSF’s and the causal analy- on a delivered product. Quality in this context
sis of the hierarchy of CSF’s; is defined as the degree to which the inherent
• Infrastructure to regularly measure actual characteristics of a product or a service and the
performance through KPI scores reporting; way it reaches the consumer fulfill consumer’s
• A data collection of KPI scores and targets; needs and expectations. A quality management
• Decision making to drive plan-do-check- system (QMS) is a system supporting a system-
act cycles and continuous improvement; atic approach to monitor, control and manage an
• An auditing regime to audit the adherence organization’s quality performance. A QMS is
to the quality management as laid down in driven by an underlying information system; an
the handbook; information system is a set of cooperating com-
• Certificate of compliance to the own ponents with a structural organization, with the
ISO 9001-based QMS, to assure custom- aim to capture, process, manage and distribute
ers, shareholders or anyone else of this data and information.
compliance. A critical success factor (CSF) is a core area
or a limited number of areas in which satisfac-
Notice that ISO 9001 does not prescribe all tory results will induce successful results in the
these elements in detail and leaves room for al- environment around these area(s). A CSF com-
ternatives for some of these elements. This holds monly represents that core action, achievement or
for instance for a Strategy Map and for a Balanced performance that if completed successfully, sets
Score Card. Most organizations will prefer to forth successful achievement of the rest. A (key)
integrate these elements in their ISO 9001-based performance indicator (KPI) is a metric expressing
QMS, however, and their use has become com- the result of a measurement of the performance of
mon practice. Also note that ISO 9001 does not a business process. Quality management following
prescribe storing KPI scores in a computer system, ISO 9001 takes a quantitative approach, which
let alone in an EIS. When preferred so, KPI scores entails a need for a-priori known and quantified
may be paper-based, but in practice virtually all quality targets. KPI scores are set off against qual-
organizations will opt to store them in a database ity targets and should be within an a-priori defined
on a computer. bandwidth around the target value. A Balanced
Score Card (BSC) is a management overview that
groups KPI values logically, according to the per-
spective they cover, typically: customer-oriented
processes, operational performance, financial
performance, and what is commonly denoted as
learning (experience, knowledge and growth). The
927
BSC (Kaplan et al., 1992) not only reports on KPI the above marked process. Once the design has
scores, but has been designed such during QMS been validated, it can be operated.
design, that it also assists in the understanding
of collective performances and mutual relation- stEP 1: starting from strategy
ships among individual objectives: what drives
what. Business Activity Monitoring (BAM) is the At board level, a balance has to be found between
acquisition, processing and presentation of real strategic planning and strategic control. The
time information (e.g., KPI scores) on activity and pressure on organizations to timely adjust their
performance within an organization. strategy is ever increasing, due to reasons such
as globalization, sustainability, exploding energy
A Generic stepwise Approach costs, rapidly changing consumer preferences,
mass customization and many others. First step
Following the research approach as outlined in for an organization is to assess its present state
the Introduction and the above definitions, we of excellence, for instance by means of capabil-
can now compile the following generic approach ity-maturity-modeling (CMM), benchmarking,
to designing and implementing a QMS, linking position auditing or using any other Business
KPI’s to strategy, and the realization of strategic Excellence Model approach.
transition to actual performance. Following the A Balanced Scorecard (BSC) captures central
Kaplan and Norton approach top down – bottom KPI’s from various strategic perspectives: custom-
up in an iterative process, we arrive at the steps er-oriented processes, operational performance,
as in Table 1. financial performance, and what is commonly
In practice, the above stepwise approach is denoted as the learning, (experience, knowledge…)
neither fully top-down nor bottom-up, but a mid- and growth perspective. The anatomy of the BSC
dle-out process that iterates until an effective and looks like in Figure 3. The relative weights put on
efficient network of coherent indicators has been each of the perspectives may vary from organization
established. A scorecard design and the indicators to organization and is a matter of how organizations
populating it are in fact strongly intermingled, see themselves, their orientation, and the maturity
and simultaneously and iteratively developed in level they position themselves in. A successful
Table 1. stepwise translation of strategy and strategic change into KPI’s and verification that the KPI’s
as designed contribute to understanding quality performance up to board level.
STEP 1: At board level, strategic objectives are being outlined in the Strategic Planning process and translated into measurable objectives,
for instance by collecting them in a destination statement, describing what should be realized within say three years from now. Strategic
change (transition) is thus put under Strategic Control. Global CSF’s are being identified, a first version of a BSC is being designed
STEP 2: At the management levels below board level (Production Control), these objectives and their CSF’s are further worked out in
smaller (measurable) goals, linking up the operational processes. A cause-effect analysis is used to identify and verify causal relationships
(Causal Analysis) between CSF’s at subsequent levels;
STEP 3: Per CSF and for each operational process, one or more KPI’s are determined, along with all their characteristics.
STEP 4: Next, bottom up a verification and calibration process is conducted to learn how lower level KPI scores aggregate at the next higher
level and how their indication contributes to the performance measurement at the next level. Aspects like interpretation, accuracy, missing
values, lead times etc. are verified and validated;
STEP 5: Finally, arriving back at the board level, the final verification is being conducted, which should reveal as to whether the progress
towards reaching the destination can be monitored and controlled indeed (Strategic Control), whether ill-performance is timely signaled
and risks and opportunities can be monitored and controlled adequately on the basis of BSC and its KPI’s.
928
Figure 2. Generic approach to design and implement a QMS, using an EIS system.
Figure 3. Anatomy of a BSC. What drives what (arrows) depends (among others) on the organization’s
maturity level, orientation and ambitions.
transition to a next maturity level can be interpreted & Rosenbaum, 2003; Brun, Caridi, Salama, &
as a demonstration of continuous improvement. Ravelli, 2006).
A first version of the BSC is typically designed What (KPI in a) perspective drives what other
in parallel to the design of Strategy Map and Des- (KPI in a) perspective, is among other things de-
tination Statement. More detailed scorecards may pending on the organization’s maturity level. Many
be designed afterwards and cascading scorecards of today’s organizations find themselves in a stage
may be used to aggregate results from various in which operational performance drives financial
levels of the organization (Bukh & Malmi, 2005). performance. Learning organizations (knowledge-
Suppliers performances can be also be linked oriented, human capital-oriented organizations)
up (Angerhofer & Angelides, 2006; Bostorff go one step further: self-organization and learning
929
competence drive internal performance which stEP 2: towards csF’s

in turn drives financial. Learning organizations
will typically work out and weight the Learning Research on critical success factors (CSF’s) is
& Growth box correspondingly. somewhat interwoven with that of KPI’s (Forster
Overlooking the above considerations, the fol- & Rockart, 1989; Rockart, 1986). Today, there is
lowing requirements on EIS systems to support a substantial body of literature available on criti-
the implementation of QMS systems, emerge: cal success factors, performance indicators and
REQUIREMENT 1: Apart from structured measurement as a whole. See for instance (Niven,
information, an EIS system must also be capable 2005) and (Hubbard, 2007). For their statistical
of containing, handling and archiving unstructured analysis background, a classic resource is (Dixon
information like the Strategy Map and Destina- & Massey, 1983).
tion Statement Requirements on EIS systems:
REQUIREMENT 2: An EIS system must be REQUIREMENT 9: An EIS system shall be
capable of storing and retrieving CMM-scores as capable of containing, handling and archiving
input information for the next assessment of the causal analysis results as unstructured information;
CMM-stage. Factors contained in the EIS system
Assumptions, conditions and limitations shall be
must be labeled as enablers and results; Maturity
stored along with the causal analysis itself
level assessments must be validated and validation
REQUIREMENT 10: Historic (regular) veri-
shall be supported by the EIS system
fication and compliance analysis can be stored as
REQUIREMENT 3: The impact of amend-
ments to one of the strategic objectives in terms structured information in the EIS system
of CSF’s and KPI’s affected, can be assessed by REQUIREMENT 11: CSF failure analysis
the EIS can be stored as unstructured information in the
REQUIREMENT 4: The other way around, EIS system
each KPI and each CSF is linked to (one or more) REQUIREMENT 12: Each occurrence of
strategic objectives; these relationships can be CSF qualification has an explicit validation, by
described for the purpose of causal analysis a hierarchy of validators (explicit roles of type
REQUIREMENT 5: Actors can be described Actor); each validation is assigned a time stamp
on a per-data item or a per-relationship basis, and can be traced back
along with access rights, modification and archival REQUIREMENT 13: Once validated, CSF
roles included qualifications shall be protected against amend-
REQUIREMENT 6: Each of the entities and ment; they can only be superseded by a revision
the relationships has an ownership defined REQUIREMENT 14: Release and version
REQUIREMENT 7: Each occurrence of KPI management shall apply to all CSF qualifications;
scores (and possibly also CSF qualification) has REQUIREMENT 15: Each CSF has one or
an explicit validation, by a hierarchy of validators more relationships with strategic objectives and
(explicit roles); Each validation shall be assigned with one or more KPI’s; Relationships of the type
a time stamp many-to-many shall be supported
REQUIREMENT 8: An EIS system must REQUIREMENT 16: The sensitivity analysis
be capable of reporting quality performance in a of CSF compliance to KPI scores can be recorded
form resembling the designed BSC with the CSF description in the EIS system
930
stEP 3: towards KPI’s industry, to express plant performance, equipment

performance, system performance, but also service
Some organizations tie CSF’s and KPI’s one to providing performance, training performance,
one; whilst it is good practice to specify CSF’s expert consultancy performance, or anything
for each strategic objective and cover all CSF’s similar. OEE combines three major factors in a
by one or more KPI, additional KPI’s may be re- single KPI: availability, performance and qual-
quired, e.g. to monitor performance in lower level ity, TEEP ads a fourth factor; the loading. OEE
processes. Abundant KPI collection on the other seeks to jointly measure planned productive time,
hand, should be avoided, just like over-accuracy performance in that time compared to the nominal
and over-confidence. Accuracy means a low bias of performance and good quality production rates.
real and measured data. Nowadays, organizations Defined like this, world class OEE’s are approx.
typically have some 10..25 CSF’s and KPI’s and 85%, resulting from >99% quality, approx. 95%
experience says that in any one decision making performance and 90% availability. A 100% OEE
process no more than a few (4..8) KPI’s should means no downtime (other than scheduled), no
be involved (Woodhouse, 2000). slow production and no defects.
Pioneering work on KPI’s has been done in the Calculation examples can readily be found in
early sixties by Daniel (Daniel, 1961). The devel- literature and on the Internet. OEE computations
opment of Lean Manufacturing (LM) and Total are also supported in many commercial software
Productive Maintenance (TPM), mainly in the Jap- systems, and can be integrated with many com-
anese automotive industry during the seventies, led mercial ERP-systems and Plant Management
to the development of principle compound KPI’s Systems.
like Overall Equipment Effectiveness (OEE) and
Total Effective Equipment Productivity (TEEP) the sMArt Paradigm
(Ahmad et al., 2002; Hubbard, 2007; Mather,
2003; Woodhouse, 2000), later supplemented In the design of KPI’s, the SMART-paradigm is
by On Time In Full from Supplier (OTIFS) and frequently being practiced, e.g. (Ahmad et al.,
On Time In Full to Customer (OTIFC) (Ahmad 2002). A SMART KPI has the following properties:
et al., 2002), covering supply and delivery chain
respectively. The OEE is a KPI widely used in
Table 2. Left to right: building op compound types of KPI’s out of simple types and their evolution into
benchmark KPI’s like OEE. Left of the solid bar are the structured KPI types, right of the solid bar are
the unstructured types. They can either be processed into structured types or stored as unstructured
raw data.
Structured Unstructured Assessment type

Standard/ best practice/
Simple Compound Complex
benchmark
- Elapsed time - Rates - OEE - Data mining Self-assessment
- Lead time - Specific cost - TEEP - Web log pruning Mutual assessment
- Cost/expenditure - ABC cost - OTIFS - Questionnaire Certification
- Unit count - Stocks - OTIFC - Interview
- Employee count - Fees/wages - ... - Benchmarking
- Customer count - Customer profiles - Reviews
- ... ... - Audit
- ...
931
Table 3. SMART KPI characteristics.
S pecific The object of measurement must be unambiguously specified

M easurable What the KPI is designed to quantify must be measurable and an adequate measurement method must be avail-
able
A chievable The measurement must be achievable in the time frame given at reasonable cost at the accuracy specified
R elevant The KPI must add to an assessment of the performance result and be relevant to the decision making process,
in the way designed
T Ime-based The KPI must be measurable as long and as often as required and at the frequency needed. Discrete measurements
must be comparable against the measurements at other moments in time, to monitor its development
specific KPI’s ... and you can’t measure what you can’t
describe
Occasionally, sheer competing KPI’s can be
found. As an example, consider a research orga- In the context of this discussion, a measure-
nization’s ICT-network. The yearly operational ment is defined as follows (Hubbard, 2007): a
cost per seat of the network may be one KPI measurement is a set of observations that reduces
(target: < 8000 € per seat per year) while yearly uncertainty where the result is expressed as a
productive research hours may be another KPI quantity. Notice that a measurement is never an
(target: > 1200 hours per researcher per year) and exact number: there is always some uncertainty
the quality of research (target: >85% satisfactory associated with each measurement. A clear and
to the customer) a third. Perfect corrective and concise description is needed of what we want
preventive maintenance by the ICT department to measure exactly: the object of measurement.
may easily bring that first target in reach. But Just a name of some phenomenon or variable
at the same time, excessive network down time (indicator) is not enough. The same goes for
may obstruct reaching productive research hours a measurement method. Apart from the quite
(second KPI scores) and jeopardize the quality of measurable phenomena, organizations have a
work. By combining the three KPI’s into a single number of less-obviously-measured-but-vital-
OEE (Table 2), optimization can be done by low- to-know things, like quality of management and
ering the one or raising the other but not at the employee motivation. In modern literature, these
expense of the others and whilst optimizing the phenomena are commonly known as intangibles,
common (global) outcome. Neither availability e.g. (Hubbard, 2007).
of the network alone, nor performance in terms
of productive hours alone, nor quality alone is Lag and Lead Indicators
specific enough for the overall objective; only
the OEE, taking all three as input, multiplying KPI’s can be subdivided in lag and lead indicators
them, is specific enough. (Nudurupati, Arshad, & Turner, 2007; Woodhouse,
2000). A lag indicator reports on a past perfor-
Measurable KPI’s mance, lead indicators are indicators that flag
the advent of an event or state while emerging.
A well-known quote in quantified quality manage- Examples are:
ment reads (Kaplan et al., 2004c):
• Tooling speed decrease may indicate tool
You can’t manage what you can’t measure ... wear, causing productivity to be in peril;
932
• A rising chisel temperature may indicate level to obtain proper productivity. According to
wear and breakdown to arrive; the measured KPI values (black solid dots), this
• Decreasing income tax agency website level is predicted to be reached after 3-4 days; the
visits may indicate a better understood in- predicted failure point is indicated in the diagram.
come tax form and consequently less pro- The alert level is predicted to be reached slightly
cessing and reviewing capacity needed at after 3.4 days; an alert with the KPI score at day
the agency next year; three allows for intervention lead time of 0.8 days.
• Employee dissatisfaction may be a lead-in If not sufficient, the alert level can be raised and
for employee absence due to illness. the alert will be issued earlier. This example is
highly simplified, to illustrate the various aspects
Lead indicators require careful (lead) timing, discussed inhere. In real practice, correct timing
depending on the underlying process dynamics of lead variables is much more complicated.
and the intervention model. With process dynam- Furthermore, it is good practice to verify and
ics being such that the targeted value (controlled validate adequate functioning of the KPI, includ-
object) can grow out of control in say 4 days, a ing timing and alert levels (the dynamics of the
monthly KPI is useless as a predictor. An hourly KPI). Scenario play, Monte-Carlo simulation
or perhaps a daily KPI would be sufficient, al- and determining confidence intervals can help
lowing for timely intervention. Critical alert to estimate reliable alert and intervention levels,
levels and lead time to intervention should also so that the actual maintenance moment is right
be taken into account. Lead time to intervention to stay on-target. The whole process of alerting,
(total response time included) should be sufficient responding and scheduling of the intervention
to allow for intervention as designed to unroll. (in total: lead time to intervention) must fit in
Timing is generally not a (big) issue for lagging the time span between alert and failure point,
KPI’s, which commonly average or sum up and together with the intervention time itself. For
log past performances over some time interval. further details, refer to (Brun et al., 2006; Dixon
Figure 4 shows an example of timing of a et al., 1983; Edgar, 2004; Hernandez-Matias et
lead variable. The tooling speed indicates tool al., 2008; Hubbard, 2007).
wear, which must stay above a critical bottom
Figure 4. Timing example for a tooling speed KPI. After 3 to 4 days, the tooling speed sinks below a
critical bottom level and the tool needs replacement and/or readjustment. Measuring the KPI value
daily predicts that failure point (here, after approx. 3.8 days) with sufficient accuracy to allow a timely
alert for intervention.
933
Achievable KPI’s KPI’s must be separated from analysis informa-

tion to drill down the causes of malfunctioning
Occasionally, the ideal KPI cannot be measured and off-target performance. Taking a traffic light
in the process, while a close-to-ideal KPI is read- as a metaphor; you do not need to know exactly
ily available: a matter of what we want to know why the traffic light turned to red in order to drive
versus what we can tell you (Figure 5). The trade home safely. It may be interesting to know but
off is the cost (risk) of not knowing (exactly) what basically, all you need to know for a safe trip is:
needs to be known, versus using what is readily green means drive on, red means stop. The same
available. A KPI does not need to be perfect and holds for KPI’s and the accuracy and uncertainty
can never reveal everything about the true state of associated with a KPI. Accuracy is a characteristic
a process much like a dashboard does not tell every of a measurement having low systematic error,
detail about a car, but enough to drive it safely. whereas precision refers to a low random error
Figure 5. Finding KPI’s means balancing what we need to know and the penalty of not knowing versus
what we can tell you and the value of that information.
Figure 6. KPI scores versus actual state; erroneous missing of bad performance (type-II errors) should
be zeroed out. False red alerts (type-I) are costly and slow down performance, but are not immediately
catastrophic. The balance is a costs/risk tradeoff.
934
(Hubbard, 2007). Simplicity of computation and measurements pay off as long as they cost less
interpretation is to be traded off against chances than the EOL reduction gained (Figure 7).
of a false alert (a false red, taking a traffic light This theoretical framework may help to de-
metaphor again) or an unjustified performance termine the “right” quality of measurements and
OK (unjustified green). See Figure 6. What false information to support decision making. Com-
rates to tolerate is a matter of balancing the costs puting EVI values at beforehand is complicated,
and the risks. however. Hubbard suggests to compute the EVI
This brings us with the issue of the value of to exclude uncertainty altogether, i.e. computing
knowing and the cost of not knowing: the more the Expected Value of Perfect Information (EVPI).
we know, the smaller the uncertainty and the better Since it excludes the entire uncertainty (EOL-
the decision making, but also: the higher the cost. after = 0), by definition EVPI is the EOL of the
Where is the tradeoff? What are the chance and the chosen alternative without additional information.
cost of being wrong? Generally, lost opportunities Examples can be found in (Hubbard, 2007).
are capitalized and information that can reduce
the chances of missing an opportunity is assigned standardized and benchmark KPI’s
a value (Hubbard, 2007; Woodhouse, 2004). The
cost of being wrong is the difference between the For many applications, lists of useful KPI’s have
wrong chosen alternative (based upon the current been compiled and disclosed through internet.
information) and the best alternative (should one Branches and professional communities in chemistry,
have had perfect information). Opportunity Loss construction, health care, etc. are starting to collect
(OL) is the cost made for an alternative that turns and standardize KPI’s, to support self-assessment
out to be wrong, Expected Opportunity Loss and benchmarking. For further details and pointers
(EOL) is thus the chance times the cost of choos- to resources, see (Califf, Gibbons, Brindis, & Smith,
ing a wrong alternative. Reducing the uncertainty 2002; Campbell, Roland, & Buetow, 2000; Dolan,
about the best alternative reduces the chances of 2001; Drummond, O’Brien, Stoddart, & Torrance,
making the wrong choice and hence the EOL. 1997; Gouscos, Kalikakis, Legal, & Papadopoulou,
The difference in EOL before and after additional 2007; Holden, 2006; Nudurupati et al., 2007; Puig-
measurement is the Expected Value of Information janer & Guillen-Gosalbez, 2008; Schultink, 2000;
(EVI) resulting from those measurements. Extra Van den Eynde, Veno, & Hart, 2003).
Figure 7. The reduction of EOL (solid curve) versus the additional measurement cost (dashed curve) to
achieve that. As long as the loss reduction ∆loss exceeds the cost of additional measurements ∆cost, extra
measurements pay off, modified after (Hubbard, 2007).
935
stEP 4: bottom up verification KPI score. Transactions supported by the EIS

must be associated with each of the equivalence
A number of important issues shall be verified classes (including the boundary values);
once the KPI’s have been designed: REQUIREMENT 18: For each KPI, the EIS
can contain, handle and archive a measured object
• Adequacy of the accuracy of each of the description, measurement method description and
KPI’s measured on the measured object measurement equipment description;
with the measurement method identified REQUIREMENT 19: The accuracy and
(Hubbard, 2007); confidence interval description (unstructured) of
• Confidence intervals on KPI scores; each of the KPI’s can be stored in the EIS system;
• The identification of outliers on the mea- REQUIREMENT 20: An EIS system must be
sured KPI’s; capable of containing, processing, reproducing
• The impact of a missing value or a late ar- and archiving historical time series of KPI scores,
rival; what if for some reason a KPI cannot along with its validations;
be measured (timely)? REQUIREMENT 21: An EIS system shall
• The adequacy of the lower bound and/or be capable of containing, handling and archiving
the upper bound (boundary values may be pointers to root cause analysis objects, to support
one-sided, yielding a half-open interval) failure analysis;
around the target value, and the so obtained REQUIREMENT 22: An EIS system shall
equivalence classes of values; allow for readjustment of accuracy of KPI’s;
• The association of transaction process- REQUIREMENT 23: An EIS system shall be
ing procedures with each of the equiva- capable of supporting variable domain-based and
lence classes and boundary values; state-based transaction processing;
Underperformance may require transac- REQUIREMENT 24: An EIS system shall
tion processing in the EIS system that dif- support triggering and transaction processing
fers from the transaction processing in case cascading on predefined variable hierarchies, to
performance is within limits; allow for compound recomputation of a series
• Reliability and effectiveness of the verifi- of KPI scores;
cation and validation procedure; REQUIREMENT 25: An EIS system shall
• The sensitivity of KPI’s and CSF’s; support exclusive modification (locking mecha-
• What to do with prospected results by lead nisms) and roll-back mechanisms in case or errors;
indicators; should a production forecast be REQUIREMENT 26: An EIS system shall
corrected as soon as a lead performance in- support the validation of predefined variable hi-
dicator predicts the next charge of products erarchies, to allow for validations of compound
to have bad quality? KPI values;
REQUIREMENT 27: Lead indicators require
Important requirements on EIS systems: timely refreshment of data it needs for its calcula-
REQUIREMENT 17: KPI’s are compound tions. Version matching is generally not enough.
values, with a target value, at least one lower or EIS systems should be capable of working with
upper bound and possibly a second boundary the refreshment rates required by lead indicators.
around the target value; EIS systems shall be ca- All data must be time stamped;
pable of storing this compound valued variable. REQUIREMENT 28: For each lead indica-
Equivalence classes must be stored (or computed) tor, an EIS can contain, handle and archive an
by the EIS system, so as to classify the incoming intervention model. An intervention model in
936
this context is a collection of planned preventive REQUIREMENT 31: An EIS system shall
or corrective actions needed to restore proper allow for a documented validation up to strategic
quality performance. An intervention model is a control level;
compound structured object capturing information REQUIREMENT 32: An EIS can present
on actors to intervene, when intervention should pointers to root cause analysis objects associated
take place (lead time), the actions to undertake, to each KPI to allow verification of the indication
the (elapsed) time required and the way in which reflected by the KPI;
intervention is to be approved. Apart from a
scheduled intervention, the EIS shall support the
recording of a realized intervention, registering QMs AND EIs systEMs
what action has been conducted. Alternately, an
EIS may store a service level agreement (SLA), The above listed requirements partly overlap.
as a model of intervention; Furthermore, they have to be evaluated against
REQUIREMENT 29: An EIS system shall EIS technological characteristics as we understand
be capable of conducting transaction processing, them today. This will be done next.
alerting and reporting synchronously with limited
latency; (distributed) system latency and human Data types required
response time need to be taken into account when
verifying and validating lead indicator dynamics, Evaluating the above requirements, the following
just like time zone differences data types shall be supported by the EIS:
Documents (unstructured data types) may be
stEP 5: Verification contained:
strategic control
• In original format by a Document type;
Finally, verification and validation is carried out as • In restricted format (e.g., PDF) by a
to whether on-target KPI scores adequately indicate Document type or in a BLOB;
progress towards strategic change accomplish- • Scanned, in image format by a Document
ment, and reversely whether off-target KPI scores type or in a BLOB;
relate to strategic control under-performance. It • Or in the form of a hyperlink to an external
is also important to verify if all relevant aspects document;
of strategic change (transition) are covered by
the KPI’s. Commonly, the report mechanism up Typical EIS systems are expected to sup-
to board level is the BSC on which the KPI’s are port these types (except Document type), both
collected in a coherent scheme (Figure 3). Deeper
analysis can help to verify whether the indications Table 4. QMS-required data types.
from the designed BSC and their mutual relation-
ships provide a valid image of the current state of Structured data types Unstructured data types
organizational performance they reflect. Sensitivity • Regular types R-DBMS (int, • Document type
analysis, confidence intervals, scenario play can all String…), or: • BLOB data type
• Standard object types in • Hyperlink data type
be helpful instruments to investigate that. OODB
Important for the requirements on EIS systems: • Date Time type
• Actor type
REQUIREMENT 30: An EIS system shall be • Transaction
capable of presenting KPI values organized as in • Release/Version
• Flag data type (validated,
the designed BSC; archived, …)
937
systems using relational data storage technology system. The first stage (data creation) may take
(RDBMS) and object-oriented storage technology place outside the scope of an EIS system, but data
(OODBMS). acquisition may be supported. Statistical analysis
is another function that may but not necessarily
KPI score Lifecycle is supported by an EIS system.
Measured KPI scores are to be evaluated
A KPI score lifecycle looks as follows; also refer against target values. The ISO principle of continu-
to (Nudurupati et al., 2007): ous improvement entails a regular sharpening of
the target value; an organization should evaluate
• data creation whether the target can be raised to a higher perfor-
◦ data source management mance level. Target values are coupled one-to-one
◦ data measurement equipment to the KPI, also have a history and versions, but
management typically have a lifecycle and refreshment cycle
◦ measured object collection much longer than that of the KPI itself, typically
◦ data measurement and acquisition yearly. Also, access rights and credentials needed
◦ data collection and structuring/ to modify target values are completely different
packaging from those of KPI scores themselves.
◦ data registration
◦ data transmission or reporting relationships
• data processing
◦ data intake With respect to relationships: optional and manda-
◦ data structure validation tory relationships of the type 0-to-1, 1-to-1, 1-to-n
◦ data completeness and integrity check and n-to-m should be supported by the EIS. This
◦ data access rights validation includes not only relationships among structured
◦ (singular) data quality control types but also among structured and unstructured
◦ data analysis and validation types and among unstructured types. EIS systems
◦ data equivalence estimation and may be expected to comply with this requirement.
transaction processing assignment
• data storage Ownership
◦ data ownership assignment
◦ data access rights assignment All data and all relationships can be assigned an
◦ data release and version assignment ownership (type Actor) and access rights. Higher
◦ data relationships (description etc.) stages of Business Excellence Models, Capability
assignment Maturity Models, etc. require that KPI’s and all
• data statistical quality control constituent measurements are documented and
• data and relationship interpretation managed, for instance by a Custodian, or a Data
• data distribution Manager. Change proposals for measurements and
• data release and version management change proposals for EIS systems must be merged
• data quality control and auditing and become in fact one and the same.
• data archiving
• data destruction transaction Processing
There are no stages in the above lifecycle that Apart from the regular transaction processing,
are not supported in some form by a typical EIS the following specific processing features shall
938
be offered by the EIS system: and also stored in or aggregated from a single EIS
subsystem, to maintain optimal system character-
• Locked processing, single item or cascad- istics with respect to:
ing, with roll-back mechanism;
• Constrained insertion and deletion • Flexibility;
to enforce mandatory data item and • Modularity;
relationships; • Extendibility;
• Archiving data items with all related data;
• Modify Actors (for instance Ownership) See Figure 8. Flexibility, modularity and
on a group (release, version…) of data; extendibility are important aspects of system
• Online validation and online auditing, in a integration and system architecture. In the em-
work flow manner; bedding of KPI’s no unnecessary subsystem
• Manage (contain, handle and archive) un- dependencies are to be introduced. A proper
structured data items; Strategic objectives-CSF’s-Process and KPI’s
• Verify the validity of an external reference mapping onto subsystems may help to achieve
(hyperlinked object); this: Customer-relationship KPI’s in the CRM
• Manage externally referenced data objects subsystem, resource-related KPI’s in the ERP
(documents); subsystem, etc.
For lead KPI’s to function correctly, retrieval,
Preferably: computation, and in fact the availability of the EIS
(subsystems) must satisfy the timing requirements
• Apart from sorting and searching struc- as specified. For further details, refer to (Wier,
tured types (according to their value), Hunton, & HassabElnaby, 2007).
searching the content of unstructured types
is preferred;
FUtUrE trENDs
Data Volume
Issues, controversies, Problems
The EIS system must be capable of storing the and Potential solutions
KPI scores over a longer period of time. Archived
data may reside online or moved to some back- Organizations and their role in society changed
ground storage (tape, vault, WORM …). During and will continue to change. A historical over-
the design of the QMS, an estimate can be made view of enterprises and their industrial, economic
with respect to the expected data volume. and societal position can be taken from (Mokyr,
2005). Edgar, in (Edgar, 2004) identifies four
Data Distribution epochs, with quality movement, the previous
epoch, merging with the present 21st –century
Data distribution is typically through a web inter- enterprise view epoch. Traditionally, (for profit)
face and is not further discussed here. organizations sought to raise business revenues
and shareholder values through high outputs while
EIs subsystems minimizing productions costs. Today, product
quality and operational excellence are generally
Different KPI’s may originate from different EIS no longer sufficient to survive. Firstly, both LSE’s
subsystems. Preferably, KPI’s are associated with and SME’s are facing globalization forcing them
939
Figure 8. Flexibility, modularity and extendibility are important aspects of system integration and system
architecture. In the embedding of KPI’s no unnecessary subsystem dependencies shall be introduced. A
proper Objectives-CSF’s-Process and KPI’s mapping onto subsystems may help to achieve this.
to reconsider the organization of their activities; in the twenty first century, organizations cannot
outsourcing, supplier chain networks, strategic operate in splendid isolation: they are (forced to
partner alliances, bundled products, alternate be) aware and confronted with public demands
delivery channels like online sales and the like, in on sustainability, setting forth ethical, societal,
a more or less global context. In the US in 2005, environmental expectations to comply with.
less than 5% of the retail activities took place on Organizations are foreseen to transform even
the Internet, in 2010 this figure is expected to go more rapidly in the future, to keep up with chang-
up to some 13% (Johnson, 2005). There is no room ing economic and societal demands (Amaravadi
to go further into these developments, see for in- Ch, 2003; Ein-Dor, 2003; Fuller, 2003a; Fuller,
stance (Gerritsen, 2008), also see (Hoogeweegen, 2003b; Hassan & McCaffer, 2002; Hoogeweegen
van Liere, Vervest, Hagdorn van der Meijden, & et al., 2006; Stevenson, 2000; Tsoukas & Shep-
de Lepper, 2006) for further details. herd, 2004; Warhurst, 2005). Contrary to classi-
Secondly, customers, having a global overview cal economical theory, China managed to show
over available options, prices and conditions, tremendous growth in SME start-ups without
nowadays find themselves in a strongly developed having a capitalist regime (Fuller, 2003b). Or-
consumers market, and act conformingly. More- ganizations will further virtualize (participate in
over, customers came to demand personalization, complex, geographically dispersed, dynamic and
options, connectivity and additional bundled fea- constantly changing alliances), become more open
tures through appropriate packaging, commonly and transparent to gain trust and confidence, and
grouped under the term mass customization. A much more agile when it comes to opting-in on
third factor is formed by increasingly compelling new opportunities. Organizations will understand
regulations and compliance demands. A fourth and exploit the fact that they are visible 24/7:
important factor is the increase of financial and branding, imaging, reputation and all other shine
fiscal models organizations (including SME’s) factors (Woodhouse, 2004), become dominant,
can choose from. Organizations develop or recon- although it is up to the customer to evaluate and
sider their strategic asset portfolio and changed acknowledge these qualities. Company culture will
their views on investments accordingly. Finally, radically change: the distance between managers
940
and workers will diminish and shared company Branches, collectives, interest groups, etc.
knowledge will become a central asset. will seek to standardize KPI’s and large-scale
Virtualization will also bring about the need programmes and databases will emerge on in-
to share formally strictly internal data and infor- ternet (www.kpilibrary.com) to support self- and
mation with supply and delivery chain and with mutual excellence assessment, ultimately lever-
partners (Puigjaner et al., 2008). The intellectual ing the global quality of life (e.g. HEDIS, UN
property concerns will become more important, Habitat (Holden, 2006). Emerging examples are
compared to today. Sharing knowledge (learning the Capability Maturity Model modeling (CMM)
organization) empowers virtual companies and and EFQM assessment techniques (European
partnerships but new management models are Foundation Quality Management), the South
needed to render this process successful. African Excellence Model, all developed along
QMS-es tend to integrate with other per- the lines of TQM (Ahmad et al., 2002; Cobbold et
formance management frameworks, like asset al., 2002a; Woodhouse, 2000). The construction
management (see Figure 9). The recent PAS industry developed the Construction Best Practice
55, ISO 9001, ISO14001 and other develop- Programme (CBPP) that serves similar purposes
ments are expected to converge into a generic (Nudurupati et al., 2007).
performance-oriented management framework.
Lifecycle management will become a leading
principle. Product, material and knowledge loops cONcLUsION
will be closed, which means that for instance car
manufacturers take in cars they produced at the For large as well as medium and small enterprises
end of the lifecycle. Manufacturers will more and and other organizations, designing, implementing
more act as an owner, offering the product as an and using a QMS can naturally and swiftly be
asset to customers. integrated in EIS systems using either relational
Figure 9. The embedding of quality management in an asset value-centered asset management framework,
after (Woodhouse, 2004). Further development may converge into a generic performance management
framework (© 2004, The Woodhouse Partnership Ltd. Used with permission)
941
or object-oriented storage technology. A number of Ahmad, M. M., & Dhafr, N. (2002). Establish-
requirements must be met, however, as discussed ing and improving manufacturing performance
in this chapter. Both structured and unstructured measures. Robotics and Computer-integrated
data and information shall be accommodated for, Manufacturing, 18, 171–176. doi:10.1016/S0736-
as well as historical data, verification, validation 5845(02)00007-8
and approval information, version and release
Amaravadi Ch, S. (2003). The world and busi-
and extensive roles and access rights models.
ness computing in 2051. The Journal of Strategic
The design of adequate KPI’s is a key step and
Information Systems, 12, 373–386. doi:10.1016/j.
interpreting them correctly during operational
jsis.2001.11.012
use, requires thorough cause-effect and accuracy
and sensitivity knowledge on the KPI. Report- Angerhofer, B. J., & Angelides, M. C. (2006). A
ing of the recorded KPI scores commonly takes model and a performance measurement system for
place using a balanced scorecard, which groups collaborative supply chains. Decision Support Sys-
KPI scores logically, perspective by perspec- tems, 42, 283–301. doi:10.1016/j.dss.2004.12.005
tive, so as to express maximum insight and as-
Bostorff, P., & Rosenbaum, R. (2003). Supply
sist maximally in adequate decision making. At
chain excellence; A handbook for dramatic im-
various organizational levels up to board level,
provement using the SCOR model. New York:
this evaluation instrument serves to monitor the
AMACOM; American Management Association.
delivery of performance and quality according to
the targets agreed upon. This approach facilitates Brun, A., Caridi, M., Salama, K. F., & Ravelli, I.
factual decision making and management control (2006). Value and risk assessment of supply chain
processes. Large, medium and small enterprises management improvement projects. International
and other organizations may implement systems Journal of Production Economics, 99, 186–201.
like this to increase customer satisfaction with doi:10.1016/j.ijpe.2004.12.016
zero defects whilst saving resources. Databases
Bukh, P. N., & Malmi, T. (2005). Re-examining
of more or less best practice KPI’s are starting to
the cause-and-effect principle of the balanced
appear on internet, realm-by-realm.
scorecard. In G.Jonsson & J. Mouritsen (Eds.),
In the future, more generic forms of perfor-
Accounting in Scnadinavia - Northern Lights (pp.
mance management are going to be seen, emerging
87-113). Malmo: Liber & Copenhagen Business
from the merger of present forms, like quality
School Press.
management and asset management. Although
ISO 9001 was considered in the above discus- Califf, R. M., Gibbons, R. J., Brindis, R. G., &
sion, all that has been said also applies to other Smith, S. C. (2002). Integrating Quality into the
quality approaches, like Total Quality Manage- Cycle of Therapeutic Development. Journal of the
ment (TQM), for ISO 9001 extensions like ISO/ American College of Cardiology, 40(11), 1895–
TS 16949:2002 for the Automotive industry and 1901. doi:10.1016/S0735-1097(02)02537-8
largely also for ISO 14001:2004.
Campbell, S. M., Roland, M. O., & Buetow, S.
A. (2000). Defining Quality of Care. Social Sci-
ence & Medicine, 51, 1611–1625. doi:10.1016/
rEFErENcEs
S0277-9536(00)00057-5
Adams, C. A., & Frost, G. R. (2008). Integrating
sustainability reporting into management prac-
tices. Accounting Forum.
942
Cobbold, I., & Lawrie, G. (2002a). Classification Fuller, T. (2003b). Small bisness futures in society
of balanced scorecards based on their intended (Introduction). Futures, 35, 297–304. doi:10.1016/
use. PMA Conference. Berkshire, UK: 2GC Ltd. S0016-3287(02)00082-4
Cobbold, I., & Lawrie, G. (2002b). The devel- Gerritsen, B. H. M. (2008). Advances in Mass
opment of the balanced scorecard as a strategic customization and adaptive manufacturing. In
management tool. PMA Conference. Berkshire, I. Horvath & Z. Rusak (Eds.), TMCE 2008 (pp.
UK: 2GC Ltd. 869-880). Delft, Netherlands: Delft University.
Daniel, R. D. (1961). Management information Gouscos, D., Kalikakis, M., Legal, M., & Papado-
crisis. Harvard Business Review, 39(Sept-Oct). poulou, S. (2007). A general model of performance
and quality for one-stop e-Government service
Dixon, W. J., & Massey, F. J. (1983). Introduc-
offerings. Government Information Quarterly, 24,
tion to statistical analysis. (3rd ed.) New York:
860–885. doi:10.1016/j.giq.2006.07.016
McGraw-Hill Book Company.
Hassan, T. M., & McCaffer, R. (2002). Vision of
Dolan, P. (2001). Output measures and valuation in
the large scale engineering construction industry in
health. In M.F.Drummond & A. McGuire (Eds.),
Europe. Automation in Construction, 11, 421–437.
Economic evaluation in health care (pp. 46-67).
doi:10.1016/S0926-5805(01)00074-7
Oxford: Oxford University Press.
Hernandez-Matias, J. C., Vizan, A., Perez-Garcia,
Drummond, M. F., O’Brien, B., Stoddart, G. L., &
J., & Rios, J. (2008). An integrated modelling
Torrance, G. W. (1997). Methods for the economic
framework to support manufacturing system
evaluation of health care programmes (2nd ed.).
diagnosis for continuous improvement. Robotics
Oxford: Oxford University Press.
and Computer-integrated Manufacturing, 24,
Edgar, Th. F. (2004). Control and operations: 187–199. doi:10.1016/j.rcim.2006.10.003
When does controllability equal profitability?
Holden, M. (2006). Urban Indicators and the Inte-
Computers & Chemical Engineering, 29, 41–49.
grative Ideals of Cities. Cities (London, England),
doi:10.1016/j.compchemeng.2004.07.013
23(3), 170–183. doi:10.1016/j.cities.2006.03.001
Ein-Dor, Ph. (2003). The world and business com-
Hoogeweegen, M., van Liere, D. W., Vervest, P.
puting in 2051: from LEO to RUR? The Journal
H. M., Hagdorn van der Meijden, L., & de Lepper,
of Strategic Information Systems, 12, 357–371.
I. (2006). Strategizing for mass customization by
doi:10.1016/j.jsis.2001.11.011
playing the business networking game. Decision
Forster, N. S., & Rockart, J. F. (1989). Critical Support Systems, 42, 1402–1412. doi:10.1016/j.
success factors: An annotated bibliography (Rep. dss.2005.11.007
No. CISR WP No. 191, Sloan WP No. 3041-89).
Huang, H.-C. (in press). Designing a knowledge-
Cambridge, MA: Sloan School of Management,
based system for strategic planning: A balanced
MIT.
scorecard perspective. Expert Systems with Ap-
Fuller, T. (2003a). If you wanted to know the plications.
future of small business what questions would
Hubbard, D. W. (2007). How to measure anything;
you ask? Futures, 35, 305–321. doi:10.1016/
Finding the value of intangibles in business. John
S0016-3287(02)00083-6
Wiley & Sons, Inc.
943
Hwang, W. T., Tien, W. T., & Shu, C. M. (2007). Kaplan, R. S., & Norton, D. P. (2001b). Trans-
Building an executive information system for forming the balanced scorecard from performance
maintenance efficiency in petrochemical plants measurement to strategic management: Part II. Ac-
-- an evaluation. Trans IChemE, Part B . Pro- counting Horizons, 15(2), 147–162. doi:10.2308/
cess Safety and Environmental Protection, 85, acch.2001.15.2.147
139–146. doi:10.1205/psep06019
Kaplan, R. S., & Norton, D. P. (2004a). How
International Organization for Standardization strategy maps frame an organization’s objectives.
ISO. (2008). ISO in figures for the year 2007 Financial Executive, 20(2), 40–45.
Geneva: ISO Central Secretariat.
Kaplan, R. S., & Norton, D. P. (2004a). Measur-
ISO. (2006). [ISO International Organization for ing the strategic readiness of intangible assets.
Standardization. Retrieved from ttp://www.iso. Harvard Business Review, 82(2), 52–63.
org]. Survey (London, England), 2006.
Kaplan, R. S., & Norton, D. P. (2004b). Strategy
Johnson, C. (2005). US e-commerce: 2005 to maps: Converting intangible assets into outcomes.
2010, a five year forecast and analysis of US Boston, MA: Harvard Business School Press.
online retail sales. Forrester Research.
Kaplan, R. S., & Norton, D. P. (2004c). Strategy
Kaplan, R. S., & Norton, D. P. (1992). The bal- Maps; Converting Intangible Assets into Tangible
ance scorecard - measures that drive performance. Outcomes. Boston: Harvard Business School
Harvard Business Review, 70(1), 71–79. Press.
Kaplan, R. S., & Norton, D. P. (1993). Putting the Kaplan, R. S., & Norton, D. P. (2004c). The
balanced scorecard to work. Harvard Business strategy map: Guide to aligning intangible as-
Review, 71(5), 134–140. sets. Strategy and Leadership, 32(5), 10–17.
doi:10.1108/10878570410699825
Kaplan, R. S., & Norton, D. P. (1996a). The bal-
anced scorecard. Boston, MA: Harvard Business Kaplan, R. S., & Norton, D. P. (2006). Alignment:
School Press. Using the balanced scorecard to create corporate
synergies. Boston, MA: Harvard Business School
Kaplan, R. S., & Norton, D. P. (1996b). Using
Press.
the balance scorecard as a strategic management
system. Harvard Business Review, 74(1), 75–85. Kim, H.-S., & Kim, Y.-G. (2008). A CRM perfor-
mance measurement framework: its development
Kaplan, R. S., & Norton, D. P. (2001a). The
process and application. Industrial Marketing
strategy-focused organization. Strategy and
Management.
Leadership, 29(3), 41–43.
Mather, D. (2003). CMMS: A timesaving imple-
Kaplan, R. S., & Norton, D. P. (2001a). Trans-
mentation process. Boca Raton, FL: CRC Press.
forming the balanced scorecard from performance
measurement to strategic management: Part I. Ac- Mokyr, J. (2005). The gifts of Athena; Historical
counting Horizons, 15(1), 87–106. doi:10.2308/ origins of the knowledge economy. Princeton, NJ:
acch.2001.15.1.87 Princeton University Press.
Niven, P. R. (2005). Balanced scorecard diag-
nostics; Maintaining maximum performance.
Hoboken, NJ: John Wiley & Sons, Inc.
944
Nudurupati, S., Arshad, T., & Turner, T. (2007). Ugwu, O. O., & Haupt, T. C. (2007). Key per-
Performance measurement in the construction formance indicators and assessment methods for
industry: An action case investigating manufac- infrastructure sustainability -- a South African
turing methodologies. Computers in Industry, 58, construction industry perspective. Building and
667–676. doi:10.1016/j.compind.2007.05.005 Environment, 42, 665–680. doi:10.1016/j.build-
env.2005.10.018
Puigjaner, L., & Guillen-Gosalbez, G. (2008).
Towards an integrated framework for supply chain Van den Eynde, J., Veno, A., & Hart, A. (2003).
management in the batch chemical process indus- They look good but don’t work: a case study of
try. Computers & Chemical Engineering, 32, 650– global performance indicators in crime prevention.
670. doi:10.1016/j.compchemeng.2007.02.004 Evaluation and Program Planning, 26, 237–248.
doi:10.1016/S0149-7189(03)00028-4
Rockart, J. F. (1986). A primer on critical success
factors. In C.V.Bullen (Ed.), The rise of managerial Warhurst, A. (2005). Future roles of business in
computing: The best of the center for Information society: The expanding boundaries of corporate
Systems research (pp. 383-423). Cambridge, MA: responsibility and a compelling case for part-
Sloan School of Management, MIT. nership. Futures, 37, 151–168. doi:10.1016/j.
futures.2004.03.033
Schultink, G. (2000). Critical environmental indi-
cators: Performance indices and assessment meth- Wier, B., Hunton, J., & HassabElnaby, H. R.
ods for sustainable rural development planning. (2007). Enterprise resource planning systems
Ecological Modelling, 130, 47–58. doi:10.1016/ and non-financial performance incentives: The
S0304-3800(00)00212-X joint impact on corporate performance. Int.J.of
Accounting Information Systems, 8, 165-190.
Stevenson, T. (2000). Will our futures look dif-
ferent, now? Futures, 32, 91–102. doi:10.1016/ Woodhouse, J. (2000). Key performance indica-
S0016-3287(99)00069-5 tors. Retrieved from http://www.TWPL.com
Tsoukas, H., & Shepherd, J. (2004). Coping with Woodhouse, J. (2004). Closing the loop: sustain-
the future: developing organizational foresight- able implementations of improvements. In ERTC
fulness (Introduction). Futures, 36, 137–144. Reliability & Asset Management Conference; Oil,
doi:10.1016/S0016-3287(03)00146-0 Gas, Petrochem & Power Industries.
945
946
Chapter 4.3
A Procedure Model for a SOA-
Based Integration of Enterprise
Systems
Anne Lämmer
sd&m AG, Germany
Sandy Eggert
University of Potsdam, Germany
Norbert Gronau
Enterprise systems are being transferred into a Enterprise resource planning systems (ERP sys-
service-oriented architecture. In this article we tems) are enterprise information systems designed
present a procedure for the integration of enter- to support business processes. They partially
prise systems. The procedure model starts with or completely include functions such as order
decomposition into Web services. This is followed processing, purchasing, production scheduling,
by mapping redundant functions and assigning dispatching, financial accounting and controlling
of the original source code to the Web services, (Stahlknecht & Hasenkamp, 2002). ERP systems
which are orchestrated in the final step. Finally an are the backbone of information management in
example is given how to integrate an Enterprise many industrial and commercial enterprises and
Resource Planning System and an Enterprise focus on the management of master and transac-
Content Management System using the proposed tion data (Kalakota & Robinson, 2001). Besides
procedure model. ERP systems, enterprise content management
A Procedure Model for a SOA-Based Integration of Enterprise Systems
systems (ECM systems) have also developed into SOA is suitable for a process oriented, distributed
companywide application systems over the last integration (Schönherr, 2005). However, the ad-
few years. ECM solutions focus on indexing all dressed goals of component orientation and SOA
information within an enterprise (Müller, 2003). are similar: different enterprise systems are con-
They cover the processes of enterprise-wide nected through one interface, and a cross-system
content collection, creation, editing, managing, data transfer and the reusage of objects or com-
dispensing and use, in order to improve enterprise ponents is enabled. Thereby a service represents
and cooperation processes (Koop, Jäckel, & van a well-defined function which is generated in
Offern, 2001; Kutsch, 2005). In order to man- reaction to an electronic request (Burbeck, 2000).
age information independently, ECM combines The SOA approach offers a relatively easy way
technologies such as document management, to connect, add and exchange single services,
digital archiving, content management, workflow which highly simplifies the integration of similar
management and so forth. The use of ECM sys- systems (e.g., enterprise take-over). Moreover,
tems is constantly on the rise (Zöller, 2005). This SOA offers a high degree of interoperability and
leads to an increasing motivation for enterprises modularity (Behrmann & Benz, 2005), which
to integrate the ECM systems within the exist- increases the adaptability of enterprise systems
ing ERP systems, especially when considering (Gronau et al., 2006).
growing international competition. The need for The SOA approach is based on the concept of
integration is also eminently based on economi- service. The sender wants to use a service and in
cal aspects, such as the expense factor in system doing so the sender wants to achieve a specific
run time (Schönherr, 2005). For a cross-system result. Thereby the sender is not interested in
improvement of business processes, enterprise how the request is processed or which further
systems have to be integrated. requests are necessary. This is the idea of SOA,
where services are defined in a specific language
and referenced in a service index. Service request
rELAtED WOrK and data exchange occur via use of predefined
protocols (Dostal, Jeckle, Melzer, & Zengler,
service Oriented Architecture as an 2005; Küster, 2003).
Integration Approach This service orientation can be used on
different levels of architecture. The grid archi-
A number of integration approaches and concepts tecture is a common example of infrastructure
already exist. They can be differentiated by in- level (Bermann, Fox, & Hey, 2003; Bry, Nagel,
tegration level (for example data, functions or & Schroeder, 2004). On the application level an
process integration) and integration architecture implementation usually takes place in terms of
(for example point-to-point, hub & spoke, SOA) Web services.
(Schönherr, 2005). This article presents an ap- The use of Web services offers the possibil-
proach to integrating enterprise systems by way ity of reusing raw source code, which is merely
of building up service-oriented architectures. transferred to another environment (Sneed, 2006).
This integration approach is of special interest The benefit of this transfer is the re-usage of per-
and will be described in more detail. fected (old) algorithms. The main disadvantage
The concept of service orientation is currently is the necessity of revising the raw source code
being intensively discussed. It can be differenti- in order to find possible dependencies (Sneed,
ated from component orientation by its composi- 2006). This is also true for enterprise systems. It
tion and index service (repository). Additionally, is not efficient to reuse the entire old system, but
947
rather only significant parts of it. To accomplish or dependencies of code parts or components.
this it is necessary to deconstruct the old enter- Other types of symptoms are possible. In general,
prise system and to locate the source code parts the answer to questions concerning the measure
which can effectively be reused. Our approach of interesting items provides hints for possible
uses self-diagnosis for finding these source code symptoms.
locations. This analysis will be considered in the Self-diagnosis can be categorized by the
third integration step. symptom acquisition method. Active and passive
self-diagnosis must also be distinguished. In this
self-Diagnosis context, the program or source code is the crucial
factor for a division between active and passive
As just described, our approach uses self-diagnosis self-diagnosis. A fundamental basis for either
for location of useful source code. For this, the alternative is an observer or monitor.
method of self-diagnosis will be presented and the Using passive self-diagnosis, the monitor
differences to other approaches will be shown. detects and collects symptoms and information.
Some approaches for transformation of legacy- It can either be activated automatically or manu-
systems into a SOA exist already. However, these ally (Gronau et al., 2006). If you know which
approaches see the whole system as one service. items need to be observed and the point where
The system gets a service description for using this information can be gathered, you only have
this service in a SOA. Our approach differs in that to monitor this point. This is what passive self-
it deconstructs the system for a tailored need. For diagnosis does. For example: if you want to know
this, the method of self-diagnosis is used. how often a button is pressed, you have to find
Self-diagnosis can be defined as a system’s where the button-event is implemented in the
capacity to assign a specific diagnosis to a detected code and observe this button-event.
symptom. The detection of symptoms and assign- In active self-diagnosis, the program’s func-
ment are performed by the system itself without tion or modules are the active elements. They
any outside influence (Latif-Shabgahi, Bass, & send defined information to the monitor and act
Bennett, 1999). The mechanism of self-diagnosis independently if necessary. The monitor is used
has been detected surveying natural systems; it can as receiver and interprets the gathered informa-
partly be applied to artificial systems as well. tion and symptoms. The main advantage of ac-
The first step of self-diagnosis is the detection tive self-diagnosis is the possibility of detecting
of symptoms. Usually the detection of one existing new symptoms, even if no clear diagnosis can
symptom is not sufficient to make an indisputable be made before the problems become acute and
diagnosis. In this case, more information and data are forwarded to other systems. In contrast, us-
have to be gathered. This can be described as ing passive self-diagnosis, the monitor can only
symptom collection. In a second step the symp- inquire about specific data. In this case, a response
toms are assigned to a specific diagnosis. Depend- or further examination is only possible if the
ing on the diagnosis, corresponding measures can problem is already known. For example: if you
be taken (Horling, Benyo, & Lesser, 2001). do not know the location of all the buttons and
Symptoms are a very abstract part of self- or the code component for the button-event, you
diagnosis. These symptoms can be a high network will have to recognize all events with their initial
load in distributed systems, missing signals, or point and filter them with the monitor. The monitor
buffer overload of the hardware layer. For enter- does not have to know how many buttons exist or
prise systems the symptoms can be the frequency where their code is located, but the buttons have
of usage of user interface elements by the user
948
to “know” to register with the monitor. These are services, which is orchestrated in the last step.
the requirements of active self-diagnosis. The process includes taking the old ERP
The assembly of diagnosis points depends on system; deconstructing it into different abstrac-
the application context and the software system. tion levels such as functional entities or business
The required time and effort cannot be specified; objects, searching for redundant entities, allo-
it depends on the design and implementation of cating the code fragments dependent on these
the software system. functional entities and encapsulating them. This
Self-diagnosis can also be employed for the results in many independent functional entities,
examination of source-code usage and interde- which can be described as a service. They have
pendences. Depending on the desired information, different abstraction levels and have to compose
different points of diagnosis have to be integrated and orchestrate with, for example BPEL-WS.
into the source-code. This composition and orchestration is the way
Different points of diagnosis have to be deter- of integration.
mined in order to allow for the allocation of code
parts to various fields and functions. Therefore Deconstruction of systems
context, programming language, and software
system architecture must be considered. First, the systems which are to be integrated are
Our approach uses this method to locate code deconstructed into services. The challenge of
parts that can be collected into components. As this step depends on the number of particular
we will demonstrate later in this article, we need services, which could span the range from one
to locate functions and enterprise systems busi- single service per system, up to a definition of
ness objects. every single method or function within a system
This method can be used for the detection of as a service. In the case of a very broad defini-
code parts which are possible services. Diag- tion, the advantages, such as easy maintenance
nosis points must thereby be integrated into the and reuse and so forth, will be lost. In case of a
system source code, and software dependencies very narrow definition, disadvantages concerning
analyzed. performance and orchestration develop; the con-
As we discussed earlier in this article, the figuration and interdependencies of the services
main challenges in integration of legacy enter- become too complex.
prise systems like ERP and ECM are, first, the This article proposes a hierarchical approach
deconstruction and second, the allocation of code. which describes services of different granular
To address these challenges, we have developed a qualities on three hierarchical levels. Areas of
procedure model which will be described next. function of a system are described as the first
of these levels (Figure 1, Part 1). For example,
an area of functions could include purchase or
PrOcEDUrE MODEL sales in the case of an ERP system or, in the case
of ECM systems, archiving or content manage-
In the following, a procedure model that integrates ment. An area of function can be determined on
general application systems within a company is the abstract level by posing questions about the
presented. The procedure model begins with the general “assigned task” of the system. The dif-
deconstruction of systems into Web services. This ferences between the three hierarchical levels
is followed by a mapping of redundant functions can be discovered by answering the following
and the assignment of original source code to Web questions:
949
Figure 1. Procedure model for the integration of application systems

Task based decomposition of systems in three hierarchical levels
1
Question 1: What are the tasks of the particular system?
Result : The services on the first level which constitute the basic task.
Question 2: Which functionality derives from every task?
Result: The services on the second level which are contributed by the
different functions
Question 3: Which business objects are utilised by both systems?
Result: Number of business objects which will be used as basic objects in

both systems, e.g. article data, customer data or index data .
Preparation of the integration and mapping of redundant functions

2
Question 1: Which tasks, functions and basic functions appear more than
once?
Result: List of possible redundant functions
Question 2: Are they redundant , i.e. superfluous , or do they provide different

services?
yes
Question 3: Can they be combined by an appropriate programming ?
Detection and assignment of services to code fragements

3
Step 1: Definition of concepts for the points of diagnosis depending on the

systems and interesting information about the source code
yes
Step 2: Programming and integrating of the markers
Step 3: Analysing the collected data
Step 4: Reengineering of redundant services depending on the answer to

question 3 of part 2
Orchestration of web services

4
Step 1: Selection of a description language for web services (e.g. WS-

BPEL)
Step 2: Wrapping of the original source code into a web service
Step 3: Modelling of the business process which is important for the

integrating systems
950
1. Question: What are the tasks of the particu- Preparation and Mapping
lar system? The answers resulting from this
step correspond to services on the first level, The main advantage of Web service architecture
which constitute the general task—for ex- is the high degree of possible reuse. By division
ample sales, purchasing, inventory manage- into three hierarchical levels, a detection of similar
ment or workflow management, archiving functions is made possible, especially on the level
or content management. These tasks are of functionality and business objects. In some
abstract and describe main functionalities. cases an adjustment of the functions is neces-
They consist of many other functions which sary in order to serve different contexts of use.
are the objects of the next level. Therefore, the next step consists of integration
2. Question: Which functionality derives on different levels and the mapping of identical
from every single task? The answers to functions (Figure 1, Part 2). This step poses the
this question correspond to the services following questions:
on the second level that are contributed by
the various functions. These functions are 1. Question: Which tasks, functions and busi-
more detailed than the general tasks. They ness objects appear more than once? For
describe what the tasks consist of and what example, most applications contain search
they do—for example, calculate the delivery functions. Some applications have functions
time, identify a major customer, or constitute for check in and check out. ERP systems
check-in and e-mail functionalities. For these calculate the time for many things with the
functions the application needs data, which same algorithm under different names.
can found in the third level. 2. Question: Are these multiple functions and
3. Question: Which business objects are uti- objects-redundant, that is superfluous, or do
lized by both systems? This answer consists they provide different services? Some func-
of the number of business objects that will tions may have the same name but perform
be used as basic objects in both systems, different tasks.
for example article data, customer data or 3. Question: Can these multiple functions and
index data. objects be combined by way of appropriate
programming? For the functions ascertained
In this procedure model, all possible levels in Question 2 to be similar functions with
of service deconstruction are addressed; yet the different names the possibility of integrating
realization on all hierarchical levels constitutes them into one has to be analyzed.
an individual task.
The result of this step is a 3-stage model dis- The advantage of this mapping is the detection
playing the services of an application. The data- of identical functions, which may by only named
level, that is the integration of databases, is not differently while completing the same task. In
further examined at this point since it is not an doing so, the benefit of reuse can be exploited to a
integral part of our model, the aim of which is to high degree. Additionally, this part of the survey
wrap functions as Web services without altering allows for a minimization of programming, due
them or the original source code. The data level to the encapsulation of multiple functions. Only
is not touched by this process. those functions which share a high number of
similarities, but nevertheless complete different
tasks, have to be programmed differently; they
can be merged by reprogramming.
951
It is important to note that for this part the language, making possible the reuse of source
deconstruction consists of an abstract level and code in service oriented architecture.
is in the functional view. In the following step, If redundant services have been detected in
this will change from a functional view to a code Part 2, which need to be reengineered, then the
view. reengineering happens now.
Detection and Assignment of Orchestration of Web services

services to code Fragments
The results of stage 3 are the described Web
The next step brings the biggest challenge, namely services. These have to be connected with each
the transformation of existing applications into other depending on the business process. This
service oriented architecture. Until now, services orchestration takes place in several steps (Figure
have been identified by their tasks, but the cor- 1, Part 4).
relation to existing source code still needs to be First, the context must be defined; second,
done. This is going to be accomplished in the next the service description language has to be se-
step (Figure 1, Part 3). lected; and third, the Web services need to be
Self-diagnosis is used at this point to integrate combined.
earlier defined points of diagnosis into the source A four-stage procedure model for a service-
code. These points of diagnosis actively collect oriented integration of application systems has just
usage data and facilitate conclusions concerning been described. This process holds the advantages
the fields and functions via their structure. The of a step-by-step transformation. The amount
structure of the points of diagnosis depends on the of time needed for this realization is consider-
context of their application and on the software ably higher than in a “big bang” transformation,
system. It is not possible to describe the complexity however, a “big bang” transformation holds a
of the process, which also depends on the structure higher risk and therefore requires high-quality
and programming of the software systems. preparation measures. For this reason, a “big
As we discussed earlier in Section 2.2, the bang” transformation is dismissed in favor of a
points of diagnosis depend on what needs to be step-by-step transformation.
observed. Here we want to know which code There is yet another important advantage in
fragments share correspondences and execute the integration or deconstruction of application
the identified functions in the functional view. systems into services when carried out in several
From this follows the necessity of a monitor. For steps. First, a basic structure is built (construction
example, the points can be every method call in of a repository, etc.). Next, a granular decomposi-
the source code of an ERP system. If the user tion into Web services occurs on the first level,
calls a function, the points of diagnosis have to thereby realizing a basic transformation of a
inform the monitor that they were called. The service oriented concept. Following this, Web
monitor has to recognize and to analyze which services of the second and third hierarchical level
method calls belong together. can be integrated step-by-step. This reduction into
Now the code fragments are analyzed and services provides high quality integration.
assigned to the functions identified in Part 1, and The procedure model we just presented is
the wrapping of code fragments into Web services very abstract. Therefore, a practical example for
can be started. This step necessitates the usage two enterprise systems, ERP and ECM, will be
of the existing source code and the description given in Part 4.
of relevant parts with a Web service description
952
EXAMPLE OF APPLyING tHE incoming mails and documents. In order to scan

PrOcEDUrE MODEL and save the incoming invoices of suppliers, the
module of the ECM System “document manage-
It is necessary to develop a general usage approach ment” is used. The access to the invoice is made
and to test it on ERP and ECM systems since no possible through the ERP system.
concrete scenario of these technologies in regard In the future, a SOA-based integration of both
to praxis as of yet exists (Issing, 2003). The aim of enterprise systems can be reasonably expected
this example of use is to describe the integration under the aspect of business process improve-
of both company-wide systems, ERP and ECM, ment. Referring to the example mentioned above,
using our presented approach. the “portal management” component could be
In what follows, we present a case study of a used to access, search, and check-in all incoming
situation of integration of two systems: an ERP documents. What follows is a description, in four
and an ECM system. A German manufacturer parts, of the integration based on the procedure
of engines and devices administer a complex IT model we presented in Part 3.
landscape. This IT landscape includes, among
others, two big enterprise systems. One of them Part 1: segmentation of ErP and
is the ERP system “Microsoft Dynamics NAV” EcM systems into services
and the other is the ECM system “OS.5|ECM”
of Optimal Systems. The ERP System includes According to the procedure model (Figure 1),
modules such as purchasing, sales and inventory the individual systems will be separated into
management. The ECM system consists of mod- independent software objects, which in each case
ules such as document management, archiving and complete specified functions or constitute business
workflow management. In the current situation objects. The segmentation is structured in three
a bidirectional interface between both systems bottom-up steps (Figure 2).
exists. One example of a business process in Identification is based on the answers to ques-
which both systems are used is the processing of tions concerning main tasks of specific systems.
Figure 2. Segmentation of services

Basic Functions ERP (selection) Basic Functions ECM (selection)
masta data content workflow

purchase sales archiving
management management management
repository inventory document

article management collaboration portal management
management management management
Areas of Functions (selection) Areas of Functions (selection)
check in identify delivery time check out check in email connection create index te rms
Business objects (selection)
customer articles index data prices master data business partner
953
The basic functions of an ERP system are pur- such as invoices or new articles. The estimation
chasing, sales, master data management, inven- of analogy of different functions, particularly
tory management und repository management. in enterprise systems where implementation is
Document management, content management, different, lies in the expertise of the developer.
records management, workflow management and Another example is the service “check in/check
portal management are basic functions of ECM out”. This service is a basic function of both ERP
systems. Subsequently, the areas of functions and ECM systems and is now to be examined for
are disaggregated into separate tasks. Business possible redundancy. After determining that the
objects are classified such as the business object services “check in” or “check out” are equal, the
“article” or “customer”. Thus, segmentation in service will be registered as a basic function only
areas of functions, tasks of functions and business once. Services which are not equal but related
objects is achieved and a basis for the reusage of will be checked in another step and either unified
services is created. with suitable programming or, if possible, spilt
into different services. The results of this step are
Part 2: Preparation of Integration/ the classification of services from ERP and ECM
Mapping systems into similar areas and the separation of
redundant services. The following table shows
The results of the first step of the segmentation are examples of separated services.
separation of services of differentiated granularity By this separation of both enterprise systems, a
per system. According to the procedure model, the higher degree of re-usage and improved complex-
mapping on the different areas will be arranged ity handling of these systems is achieved. For the
in the second step. For that purpose, the potential application of services, a service-oriented archi-
services described will be examined for similari- tecture (SOA) which defines the different roles
ties. On every level of hierarchy, the functional of participants is now required (Burbeck, 2000).
descriptions (answers to questions in Part 1) of ser-
vices are checked and compared with each other. Part 3: Detection and Assignment of
If functions or tasks are similar, they will have services to code Fragments
to be checked for possibility of combination and
be slotted for later reprogramming. One example As already described in the general introduction,
of such similarity between functions is “create the identification of functions to be segmented
index terms”. Most enterprise systems include in the source code constitutes one of the biggest
the function “create index terms” for documents challenges in a transfer to service-oriented archi-
Table 1. Examples of separate services of ERP and ECM systems
Separate services ERP ECM

Purchase Content management
Sales Archiving
Basic Functions
Article management Document management
Repository management Workflow management
Check in E-mail connection
Areas of Functions Identify delivery time Save document
Check out Create index terms
954
tecture. As part of this approach, the method of Web services with help of workflow technologies
self-diagnosis is suggested. Appropriate points (Leymann, 2003). In BPEL, a process is defined
of diagnosis will be linked to the source code in which is started by a workflow system in order
order to draw conclusions from used functions to to start a business process.
associated class, method or function in the original Web services are addressed via a graphical
source code. Through the use of aspect oriented representation with a modelling imagery of WS-
programming, aspects can be programmed and BPEL. The business process is modelled inde-
linked to the classes and methods of the application pendently from the original enterprise systems.
system. Necessary data, such as the name of the Since in the first integration step, the systems
accessed method, can be collected by accessing were separated by their tasks and functions, now
the respective classes and methods (Vanderperren, all of the functions are available for the business
Suvée, Verheecke, Cibrán, & Jonckers, 2005). process as well.
Based on a defined service, “order transaction”,
all the names of methods which are necessary
for the execution of “order transaction” must be cONcLUsION
identified. To wrap the service “order transaction”,
for example to combine it with a Web service The procedure model for the integration of ap-
description language, the original methods need plication systems as it has been presented in this
be searched for and encapsulated. Additionally, paper is an approach that has been successfully
the reprogramming of redundant functions is deployed in one case. Currently the assignment
part of the phase of identification and isolation ability and the universality are being tested. The
of services. This, as well, is only possible if the self-diagnosis, that is the assignment of source
original methods are identified. code to services via aspect oriented programming,
constitutes a bigger challenge.
Part 4: Orchestration of Web A verification of costs and benefits cannot
services be given sufficiently; however, several examples
show convincing results and suggest a general
The last integration phase is used to compile Web transferability. The complexity in such a realiza-
services. The previous steps had to be completed tion cannot be specified. Particularly for bigger
in preparation for the procedure model. The Web and complex systems, the cost-to-benefit ratio
services now are completely described and have has to be verified. Despite this, it must be rec-
a URI to be accessed. Now, only the composition ognized that the assignment of code fragments
and the chronology of requests of the specific Web to functions is not an easy task. If one observes
services are missing. For the orchestration the every method call a high number of calls must
Web service business process execution language be analyzed. Visualization can be helpful for
(WS-BPEL) is recommended. The WS-BPEL was analyzing, since method calls belonging together
developed by the OASIS-Group and is currently will build a cluster in the emerging network. The
in the process of standardization (Cover, 2005). observation of method calls is possibly not the
If the Web services present a function with a most optimal way for very complex systems. If
business process, the WS-BPEL is particularly the functional view of services in Part 1 is not
suitable for orchestration of Web services (Lübke, part of the business object layer, but only of the
Lüecke, Schneider, & Gómez, 2006). Essentially, general task layer, one can reduce the numbers of
BPEL is a language to compose (Leymann & diagnosis points. The possibilities depend on the
Roller, 2000) new Web services from existing programming language and their constructs.
955
Finally, the approach presented above de- Bry, F., Nagel, W., & Schroeder, M. (2004).
scribes a procedure model for service-oriented Grid computing. Informatik Spektrum, 27(6),
integration of different application systems. The 542-545.
integration proceeds using Web services which
Burbeck, S., (2000). The Tao of e-business
thereby improve the integration ability, interoper-
services. IBM Corporation. Retrieved January
ability, flexibility and sustainability. The reusable
12, 2008, from http://www.ibm.com/software/
Web services facilitate the extraction of several
developer/library/ws-tao/index.html
functions and combination of these into a new
service. This allows for reuse of several software Cover, R. (2004). Web standards for business
components. process modelling, collaboration, and choreog-
Altogether, Web services improve the adapt- raphy. Retrieved January 12, 2008, from http://
ability of software systems to the business pro- xml.coverpages.org/bpm.html
cesses and increase efficiency (Hofman, 2003). To
Dostal, W., Jeckle, M., Melzer, I., & Zengler, B.
give an example of the realization of the procedure
(2005). Service-orientierte architekturen mit Web
model, an integration of an ERP- and ECM-system
services [Service oriented architectures with web
was chosen. The reasons for this choice consist
services]. Spektrum Akademischer Verlag.
in targeted improvement of business aspects and
increasing complexity of both application systems. Gronau, N., Lämmer, A., & Andresen, K. (2006).
Dealing with this complexity makes integration Entwicklung wandlungsfähiger auftragsab-
necessary. Through mapping, redundant functions wicklungssysteme [Development of adaptable
can be detected and as a consequence, a reduction enterprise systems]. In N. Gronau, A. Lämmer
of the complexity is made possible. Regarding the (Eds.), Wandlungsfähige ERP-Systeme (37-56).
adaptability and flexibility of affected application Gito Verlag.
systems, Web services are a suitable approach for
Gronau, N., Lämmer, A., & Müller, C. (2006).
integration. In particular, it is the reuse of services
Selbstorganisierte dienstleistungsnetzwerke im
and an adaptable infrastructure which facilitate
maschinen- und anlagenbau [Self organized
the integration.
service networks at engineering]. Industrie-
In addition to all of this, we expect to discover
Management, 2, 9-12.
additional advantages concerning maintenance
and administration of affected application sys- Hofmann, O. (2003). Web-services in serviceori-
tems. entierten IT-Architekturkonzepten [Web services
in service oriented concepts of IT architecture].
In H.-P. Fröschle (Ed.), Web-services. Praxis der
rEFErENcEs Wirtschaftinformatik, HMD 234, dpunkt Verlag,
Wiesbaden 2003, S.27-33.
Behrmann, T., & Benz, T. (2005). Service- Horling, B., Benyo, B., & Lesser, V. (2001) Using
oriented-architecture-ERP. In T. Benz (Ed.), self-diagnosis to adapt organizational structures.
Forschungsbericht 2005 Interdisziplinäres Institut Computer Science Technical Report TR-99-64,
für intelligente Geschäftsprozesse. University of Massachusetts.
Berman, F., Fox, G., & Hey, T. (2003). Grid Issing, F. (2003). Die softwarestrategie für web-
computing. Making the global infrastrucure a services und unternehmensinfrastrukturen der
reality. Wiley. firma Sun Microsystems [Software strategy
for web services and enterprise infrastructures
956
of Sun Microsystems]. In H.-P. Fröschle (Ed.), tributed Processingm, PDP’99 IEEE Computer
Web-Services (pp. 17-26). Praxis der Wirtschaf- Society (pp. 97-104).
tinformatik, HMD 234, dpunkt Verlag.
Lübke, D., Lüecke, T., Schneider, K., & Gómez,
Kalakota, R., & Robinson, M. (2002). Praxishand- J. M. (2006). Using event-driven process chains
buch des e-business [Practice e-business]. Book of of model-driven development of business appli-
practice financial times. Prentice Hall, 317ff. cations. In F. Lehner, H. Nösekabel, & P. Klein-
schmidt (Eds.), Multikonferenz wirtschaftsinfor-
Koop, H. J., Jäckel, K. K., & van Offern, A. L.
matik 2006 (pp. 265-279). GITO-Verlag.
(2001). Erfolgsfaktor content management—Vom
Web-content bis zum knowledge management Müller, D. (2003). Was ist enterprise-content-
[Success factor enterprise content management— management? [What is enterprise content
from web-content to knowledge management]. management?] Retrieved January 14, 2008,
Vieweg:Verlag. from http://www.zdnet.de/ itmanager/strat-
egie/0,39023331,2138476,00.htm
Küster, M. W. (2003). Web-services—Verspre-
chen und realität [Web services—promises and Schönherr, M. (2005). Enterprise applikation
reality]. In H.-P. Fröschle (Ed.), Web-services (pp. integration (EAI) und middleware, grundlagen,
5-15). Praxis der Wirtschaftinformatik, HMD architekturen und auswahlkriterien [Enterprise
234, dpunkt Verlag. application integration (EAI) and middleware,
fundamentals, architectures and criteria of choice].
Kutsch, O. (2005). Enterprise-content-manage-
ERP Management, 1, 25-29.
ment bei finanzdienstleistern—Integration in
strategien, prozesse und systeme [Enterprise con- Scheckenbach, R. (1997). Semantische geschäfts-
tent management at financial service provider— prozessintegration [Semantic integration of busi-
Integration at strategies, processes and systems]. ness processes]. Deutscher Universitäts:Verlag.
Deutscher Universitäts: Verlag.
Stahlknecht, P., & Hasenkamp, U. (2002). Einfüh-
Kuropka, D., Bog, A., & Weske, M. (2006). Se- rung in die wirtschaftsinformatik [Introduction to
mantic enterprise services platform: Motivation, business computing] (10th ed). Springer Verlag.
potential, functionality and application scenarios.
Sneed, H. M. (2006). Reengineering von legacy
In Proceedings of the 10th IEEE International
programmen für die wiederverwendung als web
EDOC Enterprise Computing Conference (pp.
services [Reengineering of legacy software for
253-261). Hong Kong.
reuse as web services]. In Proceedings zum Work-
Leymann, F., & Roller, D. (2000). Production shop Software-Reengineering und Services der
workflow—Concepts and techniques. Prentice Multikonferenz Wirtschaftsinformatik.
Hall International.
Vanderperren, W., Suvée, D., Verheecke, B.,
Leymann, F. (2003). Choreography: Geschäftspro- Cibrán, M. A., & Jonckers, V. (2005). Adaptive
zesses mit web services [Choreography: Business programming in JAsCo. In Proceedings of the
processes with web services]. OBJECTspektrum, 4th International Conference on Aspect-Oriented
6, 52-59. Software Development. ACM Press.
Latif-Shabgahi, G., Bass, J. M., & Bennett, S. Zöller, B. (2005). Vom archiv zum enterprise
(1999). Integrating selected fault masking and content management [From archive to enterprise
self-diagnosis mechanisms. In Proceedings of content Management]. ERP Management, 4,
the 7th Euromicro Workshop on Parallel and Dis- 38-40.
for Reliable and Scalable Operations, edited by Nijaz Bajgoric, pp. 265-276, copyright 2010 by Information Science Refer-
ence (an imprint of IGI Global).
957
958
Chapter 4.4
Size Matters!
Enterprise System Success in
Medium and Large Organizations
Darshana Sedera
Queensland University of Technology, Australia
Organizations invest substantial resources in ac- Enterprise System (ES) is an ideology of planning and
quiring Enterprise Systems, presumably expecting managing the resources of an entire organization in
positive impacts to the organization and its functions. an efficient, productive, and profitable manner, and
Despite the optimistic motives, some Enterprise is manifested in the form of configurable information
System projects have reported nil or detrimental system packages (Laukkanen, Sarpola et al. 2007). The
impacts. This chapter explores the proposition Enterprise System vendors promote a fully integrated
that the size of the organization (e.g. medium, core business processes through the organization where
large) may contribute to the differences in benefits seamless integration of the information flowing from
received. The alleged differences in organizational one functional area to the other.Amongst the myriad of
performance are empirically measured using a prior benefits, Enterprise Systems said to deliver key benefits
validated model, using four dimensions employing like: cost reduction, productivity improvement, quality
data gathered from 310 respondents representing improvement, customer service improvement, better
27 organizations. resource management, improved decision-making and
planning, and organizational empowerment (Shang
DOI: 10.4018/978-1-59904-859-8.ch016 and Seddon 2002).
Size Matters!
Organizations devote substantial resources and many not receiving anticipated benefits, where
and time on acquiring an Enterprise System, the traditional market leveling with ES vendors
presumably expecting positive impacts to the moving into the SME market segment – this chap-
organization and its functions. These extensive ter discusses whether the organization size has an
ES implementations are typically measured in influence over the benefits brought-to-bear by the
millions of dollars Pan et al (2001), and for many Enterprise System. This study aims to contribute to
organizations they represent the largest single IT the encyclopedia by investigating the relationship
investment. The substantial resource requirements of organizational size with the performance of the
for Enterprise Systems have restricted Enterprise system (commonly referred to as System Success).
System market to the Medium-Large organizations Although prior research (Raymond 1985; DeLone
with many suggesting that ES are best suited for 1988; Raymond 1992; Lai 1994) has contributed
Large Corporations (Hillegersberg and Kumar to our understanding of IS and organization size,
2000). Recent changes in market place, wherein few have empirically assessed influence of orga-
the demand for Enterprise Systems from large nizational size for contemporary IS success. More
organizations has plateau, vendors are attempting importantly, instead of resorting to the customary
to shift their emphasis into the Small-Medium approach of considering large and medium-sized
Enterprises (SMEs) with scaled-down ES prod- organizations as one homogenous group receiving
ucts (Piturro 1999; Everdingen, Hillegersberg et equal benefits, this study aims to bring forth the
al. 2000). differences between these two groups using four
Measuring the impacts of Enterprise Systems system related dimensions. This study presented
takes on special importance since the costs and herein investigates the influence of organization
risks of these large technology investments rival size on ES performance. ES impacts are empiri-
their potential payoffs. Often carefully rational- cally measured using information received from
ized in advance, ES investments are too seldom 310 responses representing 27 organizations that
systematically evaluated post-implementation had implemented a market leading Enterprise
(Thatcher and Oliver 2001). Welsh and White System solutions in the second half of 1990.
(1981) differentiated the small and large organi- The chapter begins with a historical overview
zations using such aspects like time, skills, and of literature on size as an important determinant.
resources – where the medium organizations The broad contextual overview begins by differ-
lacking all three compared to their counterparts. entiating characteristics of the medium vs. large
D’Amboise and Muldowney (1988) argue that the organizations and demonstrating the impact of
lack of resources has made smaller organizations such contextual factors on System success. The
more vulnerable to the environmental effects research context is introduced next followed by
and misjudgments forcing them to allocate more discussions on the research methodology and data
time to adjusting to, rather than devoting time on collection instrument. The final section demon-
predicting and controlling. The resource lack of strates the observed differences between the two
constraints has been found to hinder IT adoption organizational sizes and research implications.
(Baker 1987; Cragg and Zinatelli 1995; Iacovou,
Benbasat et al. 1995; Proudlock, Phelps et al.
1999), and to negatively affect IS implementation bAcKGrOUND
success (Thong 2001) and IT growth (Cragg and
King 1993) in SMEs. Prior research suggests that organizational context
With the aforementioned background – where is a determinant of Information System (IS) suc-
organizations devote huge resources acquiring ES cess. Researchers have concluded that medium
959
Size Matters!
organizations have distinctive and unique needs ficulty associated in small businesses information
compared to large organizations (Raymond 1985; systems: (1) software problems, (2) hardware
DeLone 1988; Lai 1994) and therefore, the re- problems and (3) implementation problems.
search findings of large organizations cannot be DeLone (1981) studied the relationship be-
generalized to small to mid-sized firms. tween the size of a manufacturing firm and IS usage
Schultz and Slevin (1975) and Ein-Dor and and concluded that firm size is: (1) directly related
Segev (1978) were among the very first point out to the age of the firm’s computer operations, (2)
the importance of organizational factors in man- inversely related to the amount of external pro-
aging Information Systems. In their early work, gramming that is used, (3) directly related to the
Ein-Dor and Segev (1978) proposed a framework portion of revenues allocated to Electronic Data
after studying Management Information System Processing (EDP), and (4) inversely related to the
(MIS) in which they identified organization size percentage of EDP costs that are used for computer
as a critical variable. Ein-Dor and Segev (1978) equipment. He also explained that smaller firms
identified ten (10) organizational variables with experience more computer related problems than
direct or indirect influence on the impact of an their larger counterparts. Melone (1985) found
IS. The identified variables are: (1) organization that managers in small to mid-sized organiza-
size, (2) maturity, (3) structure, (4) time frame, tions rate accounting and inventory control as the
(5) psychological climate towards [CB] IS, (6) most frequently used and important applications,
organizational situation, (7) rank of responsible and reported that inventory control was the most
executives, (8) location of responsible executives, problematic aspect of computer usage in such
(9) steering committee location and rank and (10) organizations. Nickel and Seado (1986) reputed
resources. They found that the organization size similar findings using 121 small businesses stat-
had special importance because of its influence ing that budgeting and inventory control were
on resource availability, requirements necessary the primary uses of IS in small organizations.
for integration of professional units within an Farhoomand and Hrycyk (1985) reported lack of
organization, degree of formalization of orga- technical staff as a substantial issue for the small
nizational systems, and lead time for planning to mid-sized companies.
and implementation. Furthermore, Ein-Dor and A study by Cooley et al. (1987) identified the
Segev (1978) recognized organization size as importance of user-friendly interfaces and lower
an uncontrollable variable and stated that [CB] implementation costs as key factors affecting end
IS projects are less likely to succeed in smaller users in small to mid-sized organizations. Mon-
organizations compared to larger organizations. tazemi (1988) investigating the aforementioned
Bilili and Raymond (1993) described SME preposition, confirmed the impact of organization
decision making process as reactive, informal, size on end user satisfaction.
and intuitive. (Doukidis, Lybereas et al. 1996; An organization has two basic options when it
Proudlock, Phelps et al. 1999) asserted the op- decides to implement a computerized application;
portunistic, day-to-day focus of the small to (1) to have its own staff develop the software, or
mid-sized organizations in relation to Informa- (2) to acquire packaged software from a vendor
tion Systems. Whisler (1970) studied nineteen (Raymond, 1985). Turner (1992) stated that as
insurance companies and reputed that firm size a firm increases in size, it would demand more
was directly related to performance of IS. Cheney sophisticated software. Even though that this argu-
(1983) identified various factors that would affect ment is intuitive, it suggests a correlation between
a small business firm’s success or failure in using organization size and the adoption of package
information systems and found three areas of dif- software. Turner (1992) specifically emphasized
960
Size Matters!
the importance of small to mid-tier organizations ogy adoption. They found that as business size
obtaining assistance from external sources rather increased, the importance of expectations from
than developing applications in house. To the the [social] environment increased. However, they
contrary, Raymond (1985) found that small firms observed a negative correlation with the impor-
are capable of developing, implementing and tance of intra-firm consequences and control over
administering their own applications in-house, the potential barriers for IS adoption. Hong and
compared to their larger counterparts, specifying Kim (2001) explored the ‘fit perspective’ in 34
that small organizations could maintain an IS Enterprise System installations where organiza-
with minimal financial, technical and personnel tional size was implicitly considered as a critical
requirements. The resource constraints has lead contingency variable.
SMEs to follow an incremental approach to IT In classifying organizations in the small →
investments, which, in turn, may result in isolated medium → large spectrum, many authors use the
and incompatible systems, as well as decreased number of employees as the sole classifier. For
flexibility (Levy and Powell 1998). Raymond example in a recent study (Laukkanen, Sarpola
(1992) emphasized on the advantages of small to et al. 2007), SMEs are defined here as enterprises
mid-sized firms in developing in-house applica- with fewer than 250 employees, wherein the small
tions rather than adopting packaged software from organizations defined as companies with less than
commercial vendors. They further added that end 50 employees and large organizations are simply
user computing, (where the user have direct control classified as those companies that do not meet
over their computing needs) is more appropriate the definition of SMEs and have more than 250
to such organizations than adopting a packaged employees (Chau 1994; Chau 1995). However,
software application. it is unrealistic to associate a 50 staff (or less)
Many researchers have alluded to the skill organizations with a large scaled traditional En-
scarcity for information systems in Small to terprise System implementations as such systems
mid-sized organizations (Bilili and Raymond are targeted at larger counterparts. Though the
1993; Levy and Powell 1998; Mitev and Marsh number of employees in a company provides some
1998). Laukkanen et al. (2007) suggest that the indication to the size of the Enterprise System,
resource constraints faced by SMEs may hinder at times it can be quite misleading due to not all
their ability to maintain technology up to date, employees having access to an Enterprise System.
while at the same time forcing them to consider For example, in a Health and Pharmaceutical
their investments in IT as something that should organization – where the majority of the staff is
last for a long time (Levy and Powell 1998). on medical duties (e.g. doctors and nurses) – the
Soh et al. (1992) investigated the importance of actual Enterprise System users will be a small
external consultants on computerization success proportion of the total number of employees.
in small businesses, concluding that the level of In recent years, Sedera et al (2003) suggest the
computer system usage in small businesses with use of number of user licenses to determine the
consultants is higher than that of small businesses size of the organization for Enterprise System
without consultants. Further, they added that small discussions. They suggest keeping 1000 (or over)
businesses that engage consultants are less likely concurrent user licenses as a benchmark for large
to complete there IS project on time and within organizations and anything below classifying as
budget. Harrison (1997) using the Theory of medium enterprises.
Planned Behaviour (TPB) to explained technol-
961
Size Matters!
tHE stUDy cONtEXt tHE sUrVEy
The empirical data collection was conducted A survey instrument was designed to opera-
across 27 public sector agencies running live tionalize 27 measures of ES-success depicted
the market leading Enterprise System. These 27 in Figure 1 (See details in (Gable, Sedera et al.
agencies were the first Australian state agencies 2003; Sedera and Gable 2004). All items were
to have implemented common financial manage- scored on a seven-point Likert scale with the end
ment software state-wide namely. In 1995 the state values (1) ‘Strongly disagree’ and (7) ‘Strongly
Government commenced implementation of the Agree’, and the middle value (4) ‘Neutral’. The
Financials module across all state Government draft survey instrument was pilot tested with a
agencies (later followed by Controlling, Materi- selected sample of staff of the state Government
als Management and in some agencies Human Treasury Department. Feedback from the pilot
Resources) and soon became one of the largest round respondents resulted in minor modifications
Enterprise System installations in Australia. The to survey items. The survey gathered additional
state Government approach was very much fo- demographic details on respondents’ employment
cused on using the Enterprise System as a common title (e.g. Director, Business Analyst, Applica-
reporting and financial management tool. The tion programmer). Furthermore, the respondents
objectives of the new financial system were to were asked to provide a brief description of
provide a financial management system to state their involvement with the Enterprise System.
Government agencies that would: (1) support the Supplementary information on the organizational
‘Managing for Outcomes’ (MFO) framework and structure, characteristics of the Enterprise System
financial management improvement activities, (2) (i.e. modules in use and hardware in place) and
encourage best practice resource management the number of users in each agency was gathered
across state Government, (3) facilitate the consoli- from objective sources.
dation of state Government financial information, In addition to the 27 items of Figure 3, the
(4) meet the business needs of agencies and (5) questionnaire included two criterion items aimed
achieve economies of scale in main operations. at gauging the respondent’s perception of overall
Despite the claimed benefits by most of the ES-success: (1) ‘overall…the impact of [the name
agencies, a relatively smaller agency that provides of the Enterprise System] on the agency has been
corporate services to a group of other agencies positive’ and (2) ‘overall… the impact of [the
demonstrated their dissatisfaction with their Enter- name of the Enterprise System] on me has been
prise System. Even though the Enterprise System positive’.
provided rich functionality to this organization,
the senior management believed that the system
in place was too complex and too expensive to rEsULts AND ANALysIs
operate in a smaller organization. After three years
of using the implemented Enterprise System, A total of three hundred and nineteen (319)
the agency decided to replace that with a local, responses from twenty-seven (27) public sector
small scaled Enterprise System. The contextual agencies were received. Nine responses were
background further questions the preposition removed from the analysis due to missing values
in discussion – whether the organizational size and perceived frivolity. Using the number of SAP
influences the benefits you receive. user licenses, the sample was divided into two
962
Size Matters!
Figure 1. The measures of the ES-success measurement model
Figure 3. Results of t-test for the criterion measure (alpha = 0.05)
mutually exclusive representing the respondents ernment. All participated agencies having: (1) the
from medium organizations and large organiza- same Enterprise System software application, (2)
tions. Organizations with more than 1000 SAP user the similar versions of the Enterprise System, (3)
licenses considered as large agencies and the rest in the same phase of the ES life cycle, and (4)
were medium agencies. Additional criteria were installed Financial Accounting and Controlling,
established (i.e. Number of employees, dispersion Materials Management modules created a unique
of the organization) to be used in the grouping ex- homogeneous environment increasing the com-
ercise to supplement the principal criterion, where parability of the data.
the initial classification was unclear. Figure 2(a) The one-way analysis of variance (ANOVA)
shows the break down of organizations, classified F-test was chosen as the method for conducting
into medium and large organizations and Figure the analysis of the Likert scale data. For each
2(b) shows the classification of respondents segre- variable measured with Likert scale the statistics
gated into the two agency cohorts. All indications reported include the arithmetic mean (mean) and
suggest that this distribution is representative of standard deviation (std dev.) of the responses in
users of the Enterprise System in the state Gov- each company group, the significance of group
Figure 2. Respondent classification
963
Size Matters!
mean differences (sig.) indicated by F-test, and mean scores of each success dimension and their
the group sizes (n). If a significant difference was corresponding measures.
found in ANOVA at the 0.05 level, paired t-tests
comparisons were conducted to see which of
the company groups differ from each other. The systEM QUALIty
criterion item (Overall, the impact of [the name
of the Enterprise System] on the agency has been The quality of a system under investigation is a
positive) was used to establish the peripheral dif- multifaceted phenomenon. The system quality
ferences between the two cohorts. Analysis of our construct is designed to capture how the system
survey data indicates that significant differences performs from a technical and design perspective.
exist between the medium-sized, and large or- Nine validated measures have been employed
ganization with a high F value of 5.22. It is also in this study. Results depicted in Figure 5, it is
observed that the large organizations demonstrate evident that for all 09 measures on Enterprise
a higher mean score for the criterion item over System Quality, large organizations demonstrate
the medium organizations. relatively larger mean scores compared to their
Encouraged by the findings above, using the smaller counterparts. Moreover, the following
averages of each of the four success dimensions, observations are made in relation to the measures
we conducted an independent sample t-test to of system quality. All, but two measures (flexibility
further explore the differences between the two and customization), have received below scale-
organizational sizes with regards to the dimen- median (4) values for the large organizations. To
sions. the contrary, for medium organizations, 6 of the
Results depicted in Figure 4 indicate significant 9 measures report below scale-median scores.
differences between the two organizational sizes in The two cohorts demonstrate substantial dif-
three of the four success dimensions. The differ- ferences in mean scores across IT-sophistication
ences are observed in System Quality, Individual and the employment of system features. Similarly,
Impacts and Organization Impacts. the mean scores on Enterprise System meeting
Having determined that two organizational the requirements of the organization demonstrate
sizes demonstrate significant differences in rela- differences between the medium and large co-
tion to the Enterprise System success dimensions, horts, with large reporting high mean scores. It
the chapter now attempts to identify where those is noticed with some interest that system centric
differences exists and whether an Enterprise Sys- attributes such as: integration and customization
tem investment favors a particular organizational do not demonstrate strong differences between
size. In order to establish this, we now look at the the two cohorts.
Figure 4. Results of t-test for the success dimensions (alpha = 0.05)
964
Size Matters!
Figure 5. Mean scores for system quality
INFOrMAtION QUALIty conciseness, convenience and relevance as being

key aspects of Executive Information Systems
Measures of information quality focus on the information quality. Results of the exploratory
output – both on-screen and reports – produced survey and expert workshops revealed context-
by the system, and the value, usefulness or relative specific measures of information quality and
importance attributed to the output by the users. thus significant changes have been made to the
In an early leading study of IS success, Bailey and sub-constructs of information quality. The study
Pearson (1983) identified nine characteristics of employs the six validated measures from the ES-
information quality: accuracy, precision, currency, Success of Gable Sedera et at 2003; Sedera Gable
timeliness, reliability, completeness, conciseness, 2004. The Figure 6 depicts the mean scores for
format and relevance. Sirinivasan (1985) added the six measures of Information Quality. Similar
‘understandability’ of information as another to System Quality dimension discussed above,
important sub-construct; while Saaksjarvi and large organizations demonstrate a higher mean
Talvinen (1993) employed content, availability, score for all six measures.
accuracy as sub-construct measures of information Though all mean-scores reported for Informa-
quality in their study of marketing information tion Quality were higher for large organizations,
systems. Rainer (1995) found accuracy, timeliness, substantial differences were observed only in
Figure 6. Mean scores for information quality
965
Size Matters!
relation to Information Relevance and the format individual benefits received to the medium scaled
of the reports. organizations with substantial differences in all
four perspectives.
INDIVIDUAL IMPAct
OrGANIZAtIONAL IMPActs
Individual impact is concerned with how the
Enterprise System influences the performance The impact of an Enterprise System on organiza-
of the individual user. Individual impact tends to tional performance is some-what difficult to isolate
encompass a broad range of subjective measures from the general organizational performance
such as: confidence in decisions made, improve- indicators. The eight measures of the ES-success
ments in decision-making, and the time to reach a measurement model purportedly isolate the impact
decision (Ein-Dor, Segev et al. 1981; Sirinivasan of the system with the one of the organization. The
1985; Kim and Lee 1986). Dickson et al. (1977) analyses of the mean scores for the 08 measures
provided early insights into Individual Impact support all observations above favoring the large
citing decision quality, decision time, decision organizations.
confidence, and estimated outcomes. Though It is also observed that the medium organiza-
individual productivity and decision making ef- tions report substantially lower mean scores for
fectiveness have been mentioned in prior studies five out of the eight measures. The biggest dif-
as impacts from the system, the potential benefits ferences were seen in facilitation of e-business
from the Enterprise System exceeds those and in- through Enterprise Systems and the introduction
cludes aspects such as – facilitating organizational of optimal business process changes. The three
learning and information recall and awareness cost related aspects: (1) reduction of organization
through organizational transparency. costs, (2) reduction of staff and (3) operational
Observing mean scores reported in Figure 7, costs reductions have received below median
similar patters are observed to the previous two scores for both organizational cohorts (See Figure
success dimensions with large organizations 8 for details).
receiving more benefits from the Enterprise Sys-
tem. More importantly, the mean score analysis
demonstrates a bleak picture of the amount of
Figure 7. Mean scores for individual impacts
966
Size Matters!
Figure 8. Mean scores for organization impacts
cONcLUsION employed the number of Enterprise System User

Licenses to classify the organizations in to the two
This chapter explored whether there are significant organizational sizes. The authors argued the bias
differences exist between medium-sized and large and the influence that other factors (e.g. number
enterprises regarding the success of the Enterprise of employees and budget) may bring forth in a
System. In doing so, it gathered data from 310 study of this nature.
respondents from 27 organizations that had imple- The results empirically provide evidence to a
mented a market leading Enterprise System in the well-known anecdote that traditional Enterprise
second half of 1990. The impact of the Enterprise Systems are better suited for the Large Organiza-
System was empirically measured using validated tions. The results demonstrated significant differ-
constructs measures of the Enterprise System ences between the medium and large organizations
Success Measurement Model (Gable, Sedera et in relation to Enterprise System Quality, Impacts to
al. 2003; Sedera and Gable 2004). the Individuals, and Impacts to the Organization.
The findings indicate that organization size No differences were observed in relation to the
undoubtedly has a strong influence over the ben- Quality of Information derived from the system.
efits received from the system. Though there have Substantial differences between the two types
been other studies investigating various aspects of of organizations were observed in relation to
Enterprise System (e.g. differences between small, Individual and Organizational impacts – raising
medium and large organizations on Enterprise Sys- concerns over the suitability of Enterprise Systems
tem initial objectives and constraints of adoption for medium sized organizations.
(Laukkanen, Sarpola et al. 2007), implementation The result also demonstrated that some of the
issues (Mabert, Soni et al. 2003), this is one of the common system related issues, such as customiza-
first to demonstrate the differences in relation to tion, are equally deterrent to both organizational
the outcomes of the system. types. Similarly, the innate Enterprise System
The homogeneity in the study context – where advantages like the integration are equally ben-
all the sampled organization having implemented eficial to both organizational types.
the same Enterprise System, similar modules and We recognize two attributes that may have
are at the same phase of the lifecycle – provided a contributed to the under-performance of Enter-
distinct strength to the study, where the results are prise System in medium tiered organizations.
less vulnerable to extraneous factors. The study The economies of scale could be one the lead-
967
Size Matters!
ing factors that hamper the results of medium rEFErENcEs

tiered organizations. From the study findings it
is evident that medium sized organizations have Bailey, J. E., & Pearson, S. W. (1983). Develop-
received reasonable benefits through System ment Of A Tool For Measuring And Analyzing
and Information Quality dimensions. However, Computer User Satisfaction. Management Science,
the stark differences observed in Individual and 29(5), 530–545. doi:10.1287/mnsc.29.5.530
Organization Impacts suggests that though the Baker, W. H. (1987). Status of information man-
system and information quality were adequate, agement in small businesses. Journal of Systems
mid-sized companies have failed to attain cost, Management, 38(4), 10–15.
productivity and resource benefits. Specifically,
in resource-demanding ERP investments, the Bilili, S., & Raymond, L. (1993). Information
larger enterprises have been found to be able to technology: Threats and opportunities for small
take advantage of economies of scale and, hence, and medium-sized enterprises. International Jour-
compared to their larger counterparts, smaller nal of Information Management, 13(6), 439–448.
companies are faced with a relatively bigger doi:10.1016/0268-4012(93)90060-H
commitment when adopting Enterprise Systems
Chau, P. Y. K. (1994). Selection of packaged
(Mabert, Soni et al. 2000).
software in small businesses. European Journal of
Secondly, the resource limitations character- Information Systems, 3(4), 292–302. doi:10.1057/
ized in medium organizations is identified as ejis.1994.34
another probable contributor to the poorer success
reported by the medium-sized companies. Akin Chau, P. Y. K. (1995). Factors used in the selec-
to a popular view where practitioners argue that tion of packaged software in small businesses:
‘implementing an Enterprise System is just the views of owners and managers. Information &
beginning’, organizations are required to make Management, 29(2), 71–78. doi:10.1016/0378-
continuous investments into optimizing an Enter- 7206(95)00016-P
prise System. The resource scarcity of mid-sized
Cheney, P. H. (1983). Getting The Most Out Of
organizations may not allow further investments
Your First Computer System. American Journal
into the Enterprise System on training, upgrades,
of Small Business, 7(4), 476–485.
business process improvements and organizational
change management practices. Cooley, P. L., & Walz, D. T. (1987). A Research
The findings are particularly important to the IT Agenda For Computers And Small Business.
practitioners (and academics alike) to understand American Journal of Small Business, 11(3),
the diversity of impacts received from Enterprise 31–42.
System and the importance of contextual factors.
Cragg, P. B., & King, M. (1993). Small-firm com-
At a time where the Enterprise System vendors
puting: motivators and inhibitors. MIS Quarterly,
are moving aggressively towards scaled-down
17(1), 47–60. doi:10.2307/249509
systems specifically targeting at small organiza-
tions, the study results provide some caution over Cragg, P. B., & Zinatelli, N. (1995). The evolution
the claimed benefits of Enterprise Systems. of information systems in small firms. Information
& Management, 29(1), 1–8. doi:10.1016/0378-
7206(95)00012-L
968
Size Matters!
d’Amboise, G., & Muldowney, M. (1988). Man- Gable, G., Sedera, D., et al. (2003). Enterprise
agement theory for small business: attempts and Systems Success: A Measurement Model. Pro-
requirements. Academy of Management Review, ceedings of the 24th International Conference
13(2), 226–240. doi:10.2307/258574 on Information Systems, Seattle, Washington,
Association for Information Systems.
DeLone, W. H. (1981). Firm Size And The Char-
acteristics Of Computer Use. MIS Quarterly, 5(4), Harrison, D. A., Mykytyn, J. P. P., & Riemensch-
65–77. doi:10.2307/249328 neider, C. K. (1997). Executive Decisions about
Adoption of Information Technology in Small
DeLone, W. H. (1988). Determinants Of Success
Business: Theory and Empirical Tests. Information
For Computer Usage In Small Business. MIS
Systems Research, 8(2), 171–196. doi:10.1287/
Quarterly, 12(1), 50–61. doi:10.2307/248803
isre.8.2.171
Dickson, G., & Senn, J. (1977). Research In
Hillegersberg, J. V., & Kumar, K. (2000). ERP
Management Information Systems: The Minne-
experience and evolution. Communications of
sota Experiments. Management Science, 23(9),
the ACM, 43(4), 23–26.
913–923. doi:10.1287/mnsc.23.9.913
Hong, K.-K., & Kim, Y.-G. (2001). The Critical
Doukidis, G. I., & Lybereas, P. (1996). Informa-
Success Factors For ERP Implementation: An
tion systems planning in small businesses: A
Organizational Fit Perspective. Information &
stage of growth analysis. Journal of Systems and
Management, 40(1), 25–40. doi:10.1016/S0378-
Software, 33(2), 189–201. doi:10.1016/0164-
7206(01)00134-3
1212(95)00183-2
Iacovou, C. L., & Benbasat, I. (1995). Electronic
Ein-Dor, P., & Segev, E. (1978). Organizational
data interchange and small organizations, adoption
Context And The Success Of Management Infor-
and impact of technology. MIS Quarterly, 19(4),
mation Systems. Management Science, 24(10),
465–485. doi:10.2307/249629
1064–1077. doi:10.1287/mnsc.24.10.1064
Kim, E., & Lee, J. (1986). An Exploratory Con-
Ein-Dor, P., Segev, E., et al. (1981). Use Of
tingency Model Of User Participation And MIS
Management Information Systems: An Empirical
Use. Information & Management, 11(2), 87–97.
Study. Proceedings of the 2nd International Confer-
doi:10.1016/0378-7206(86)90038-8
ence on Information Systems, Cambridge, Mas-
sachusetts, Association for Information Systems. Lai, V. S. (1994). A Survey Of Rural Small Busi-
ness Computer Use: Success Factors And Deci-
Everdingen, Y., & Hillegersberg, J. (2000).
sion Support. Information & Management, 26(6),
ERP adoption by European midsize companies.
297–304. doi:10.1016/0378-7206(94)90027-2
doi:10.1145/332051.332064 Laukkanen, S., & Sarpola, S. (2007). Enter-
prise size matters: objectives and constraints
Farhoomand, F., & Hrycyk, G. P. (1985). The
of ERP adoption. Journal of Enterprise In-
Feasibility Of Computers In The Small Business
formation Management, 20(3), 319–334.
Environment. American Journal of Small Busi-
doi:10.1108/17410390710740763
ness, 9(4), 15–22.
969
Size Matters!
Levy, M., & Powell, P. (1998). SME flex- Rainer, J. K. R., & Watson, H. J. (1995). The
ibility and the role of information systems. Keys to Executive Information System Success.
Small Business Economics, 11(2), 183–196. Journal of Management Information Systems,
doi:10.1023/A:1007912714741 12(2), 83–99.
Mabert, V. A., & Soni, A. (2000). Enterprise Re- Raymond, L. (1985). Organizational Charac-
source Planning Survey Of U.S. Manufacturing teristics And MIS Success In The Context Of
Firms. Production and Inventory Management Small Business. MIS Quarterly, 9(1), 37–52.
Journal, 41(2), 52–58. doi:10.2307/249272
Mabert, V. A., & Soni, A. (2003). The impact of Raymond, L., & Bergeron, F. (1992). Personal
organizationsize onen terprise resource planning DSS success in small enterprises. Information &
(ERP) implementations in the US manufacturing Management, 22(5), 301–308. doi:10.1016/0378-
sector. Omega, 31, 235–246. doi:10.1016/S0305- 7206(92)90076-R
0483(03)00022-7
Saaksjarvi, M. T. V., & Talvinen, J. M. (1993).
Melone, S. C. (1985). Computerising small Integration And Effectiveness Of Marketing Infor-
business information systems. Journal of Small mation Systems. European Journal of Marketing,
Business Management, (April): 10–16. 27(1), 64–79. doi:10.1108/03090569310024567
Montazemi, A. R. (1988). Factors Affecting Schultz, R. L., & Slevin, D. P. (1975). Implemen-
Information Satisfaction In The Context Of The tation and organisational validity: An empirical
Small Business Environment. MIS Quarterly, investigation. Implementing operational research /
12(2), 238–256. doi:10.2307/248849 management science. R. L. Shultz and D. P. Slevin.
New York, Elsevier, North-Holland: 153-182.
Nickell, G. S., & Seado, P. C. (1986). The Im-
pact Of Attitudes And Experience On Small Sedera, D., & Gable, G. (2004). A Factor and
Business. American Journal of Small Business, Structural Equation Analysis of the Enterprise
10(1), 37–48. Systems Success Measurement Model. Inter-
national Conference of Information Systems,
Pan, S. L., Newell, S., et al. (2001). Knowledge
Washington, D.C.
Integration As A Key Problem In An ERP Imple-
mentation. Proceedings of the 22nd International Sedera, D., Gable, G., et al. (2003). ERP Success:
Conference on Information Systems, New Orleans, Does Organization Size Matter? Proceedings of
Louisiana, Association for Information Systems. the 7th Pacific Asia Conference on Information
Systems, Association for Information Systems.
Piturro, M. (1999). How midsize companies are
buying ERP. Journal of Accountancy, 188(3), Shang, S., & Seddon, P. B. (2002). Assessing
41–48. And Managing The Benefits Of Enterprise
Systems: The Business Manager’s Perspective.
Proudlock, M. J., & Phelps, B. (1999). IT adoption
Information Systems Journal, 12(4), 271–299.
strategies: Best practice guidelines for professional
doi:10.1046/j.1365-2575.2002.00132.x
SMEs. Journal of Small Business and Enter-
prise Development, 6(4), 240–252. doi:10.1108/ Sirinivasan, A. (1985). Alternative Measures
EUM0000000006678 Of System Effectiveness: Associations And
Implications. MIS Quarterly, 9(3), 243–253.
doi:10.2307/248951
970
Size Matters!
Soh, C. P. P., & Yap, C. S. (1992). Impact of KEy tErMs AND DEFINItIONs
consultants on computerisation success in small
businesses. Information & Management, 22, Enterprise System: Customizable, standard
309–319. doi:10.1016/0378-7206(92)90077-S software solutions that have the potential to link
and automate all aspects of the business, incor-
Thatcher, M. E., & Oliver, J. R. (2001). The impact porating core processes and main administrative
of technology investments on a firm’s produc- functions into a single information and technology
tion efficiency, product quality, and productivity. architecture.
Journal of Management Information Systems, Individual-Impact: A measure of the extent
18(2), 17–45. to which [the IS] has influenced the capabilities
Thong, J. Y. L. (2001). Resource constraints and and effectiveness, on behalf of the organization,
information systems implementation in Singa- of key-users.
porean small business. Omega, 29(2), 143–156. Information-Quality: A measure of the qual-
doi:10.1016/S0305-0483(00)00035-9 ity of [the IS] outputs: namely, the quality of the
information the system produces in reports and
Turner, J. S. (1992). Personal DSS success in on-screen.
small business. Information & Management, 22, Organizational-Impact: A measure of the
301–308. doi:10.1016/0378-7206(92)90076-R extent to which [the IS] has promoted improve-
Welsh, J. A., & White, J. F. (1981). A amall busi- ment in organizational results and capabilities.
ness is not a little big business. Harvard Business Public Sector: The public sector is the part of
Review, 59(4), 18–32. economic, administrative and Governance process
that deals with the delivery of goods and services
Whisler, T. (1970). The Impact Of Computers On by and for the government.
Organizations. New York, NY, Praeger Publishers. SAP: SAP [used to denote SAP R/3 software]
is a market leading Enterprise System software.
System-Quality: A measure of the perfor-
mance of [the IS] from a technical and design
perspective.
971
972
Chapter 4.5
Web Services as XML Data
Sources in Enterprise
Information Integration
Ákos Hajnal
Computer and Automation Research Institute, Hungary
Tamás Kifor
Gergely Lukácsy
Budapest University of Technology and Economics, Hungary
László Z. Varga
AbstrAct data sources in enterprise information integration

by showing how the web service wrapper compo-
More and more systems provide data through nent integrates XML data of Web services in the
web service interfaces and these data have to be application domain of digital libraries.
integrated with the legacy relational databases of
the enterprise. The integration is usually done with
enterprise information integration systems which INtrODUctION
provide a uniform query language to all informa-
tion sources, therefore the XML data sources of Traditional Enterprise Information Integration
Web services having a procedural access interface focuses mainly on the integration of different
have to be matched with relational data sources relational data sources, however recent enterprise
having a database interface. In this chapter the information systems follow the service oriented
authors provide a solution to this problem by de- architecture pattern and are based on web services
scribing the Web service wrapper component of the technology1. In addition, more and more informa-
SINTAGMA Enterprise Information Integration tion and service providers on the internet provide
system. They demonstrate Web services as XML web service interface to their system. The integra-
tion of these new information sources requires that
the Enterprise Information Integration system has
DOI: 10.4018/978-1-60566-330-2.ch005
an interface towards web services.
Web Services as XML Data Sources in Enterprise Information Integration
This chapter describes a solution to this prob- rELAtED WOrK

lem using the SINTAGMA Enterprise Information
Integration System2 and extending this system with There are several completed and ongoing re-
a Web Service Wrapper component (which is the search projects in using logic-based approaches
main contribution of this chapter). SINTAGMA is for Enterprise Application Integration (EAI) and
a data centric, monolithic information integration Enterprise Information Integration (EII) as well.
system supporting semi-automatic integration of The generic EAI research stresses the impor-
relational sources using tools and methods based tance of the Service Oriented Architecture, and the
on logic and logic programming (see Benkő et al. provision of new capabilities within the framework
2003) based on the SILK tool-set which is the of Semantic Web Services. Examples for such
result of the SILK (System Integration via Logic research projects include DIP (see Vasiliu et al.
& Knowledge) EU project. 2004) and INFRAWEBS (see Grigorova 2006).
In order to prepare for the challenge of in- We have also approached the EAI issue from the
tegrating XML data provided by web services, agent technology point of view (see Varga et al.
we extended the original SINTAGMA system 2005 and Varga et al. 2004). These attempts aim at
in two directions. First, the architecture of the the semantic integration of Web Services, in most
SINTAGMA system was changed significantly cases using Description Logic based ontologies,
to be made up of loosely coupled components agent and Semantic Web technologies. The goal
rather than a monolithic structure. Second, the of these projects is to support the whole range of
functionality has become richer as, among oth- EAI capabilities like service discovery, security
ers, the system now deals with Web Services as and high reliability.
information sources. The component responsible Most of the logic-based EII tools use descrip-
for this is the Web Service Wrapper which is the tion logics and take a similar approach as we did
main contribution of this chapter. in SINTAGMA, that is, they create a descrip-
Mixing relational data sources and web ser- tion logic model as a view over the information
vices during an information integration scenario sources to be integrated. The basic framework
can be very useful as demonstrated by a use case of this solution is described e.g. by Calvanese et
by Lukácsy et al. 2007 and poses the challenge al. 1998. The disadvantage is that these types of
of representing procedural information as rela- applications deal with relational sources only and
tional data. are therefore not applicable to process modeling.
This chapter is structured as follows. First we This chapter unifies the procedural EAI ap-
put the problem in the context of related work, then proach and the relational EII approach by integrat-
we describe the main ideas behind the SINTAGMA ing relational and functional XML information
system in a nutshell, then we provide an overview sources within the SINTAGMA system. The
of the basic web service concepts and the modelling advantage of this approach is that the integration
language of SINTAGMA, then we present how team does not have to implement web service
we model and query web services, with samples. interface to relational databases nor relational
Finally, we demonstrate web service integration database interface to web services, because the
in a digital library application and summarize our SINTAGMA system automatically integrates
application experiences and conclusions. the different sources. In addition to the integra-
tion, the SINTAGMA system includes several
optimizations when answering queries on the
integrated system.
973
The integration of web services with the rela- Another problem was that the results provided by
tional data sources includes two important tasks: XQUERY require further transformation before it
modeling the web services in the SINTAGMA is returned to SINTAGMA. For these reasons we
system and querying the XML data returned by decided to implement a query engine as described
the web service. in this chapter.
Modeling web services in SINTAGMA is a
reverse engineering task that seems to be straight-
forward, however it is necessary. Most tools tHE sINtAGMA APPrOAcH
available for modeling web services represent
the opposite approach: they create WSDL from The main idea of our approach is to collect and
UML. Although there exist tools for modeling manage meta-information on the sources to be
WSDL in UML (e.g. http://wsdl2xmi.tigris.org/) integrated. These pieces of information are stored
or modeling XSD in UML (e.g. supported by an in the model warehouse of the system in a form of
XML editor), we did not find a tool that combines UML-like models, constraints and mappings. This
the two in the appropriate way from our point of way we can represent structural as well as non-
view. WSDL modeling tools focus on the structure structural information, such as class invariants,
of the WSDL, but do not provide the model of the implications, etc. All of our modeling constructs
message schema contained (or imported) within have well defined semantics.
the WSDL. XSD modeling tools do not provide The process of querying these models is called
information about WSDL specific information mediation. Mediation decomposes complex
such as SOAP protocols and network locations. integrated queries to simple queries answerable
Another problem is that, although models in SIN- by individual information sources and, having
TAGMA are similar to UML, models generated obtained data from these, composes the results
by the available tools cannot be used directly be- into an integrated form. For mediation, we need
cause SINTAGMA has an own modeling language mappings between the separate sources and the
(called SILan) and not all UML components/ integrated model. These mappings are called
features are supported by SINTAGMA. These abstractions because often they provide a more
are the reasons why the new modeling procedure abstract view of the notions present in the lower
(described in this chapter) is needed. level models.
There are tools for querying XML, the most We handle models of different kinds. From one
well-known tools are XPATH and XQUERY. We point of view we can speak of unified models and
studied the possibility of transforming SQL-like local models. Unified models are created from
queries supported by SINTAGMA to XQUERY other ones in the process of integration, while
statements. However we found that the SQL-like the local models represent particular informa-
query language and XQUERY are essentially dif- tion sources. More importantly, we distinguish
ferent. XQUERY is based on the XML instance, between application and conceptual models. The
and not on the schema: it is possible to query application models represent the structure of an
XML fragments in the XML instance given by existing or potential system and because of this
XPATH expressions, but not XML fragments they are fairly elaborate and precise. Conceptual
corresponding to specific complex type defini- models, however, represent mental models of
tion (class instances in our terms). The problem user groups, therefore they are more vague than
is that, if the schema is recursive, to query the application models.
instances of a complex type would require (theo- Access to heterogeneous information sources
retically) infinite number of XPATH expressions. is supported by wrappers. Wrappers hide the syn-
974
tactic differences between the sources of different WSDL is structured as follows. Within the
kind (e.g. RDBMS, XML, Web services, etc.) by root <definitions> element, child element <types>
presenting them to upper layers uniformly as UML encapsulates the XSD definitions (XSD schema)
models. Wrappers also support queries over these of different data types and structures used in
models as they are capable of directly accessing message contents. It is followed by a series of
the types of data sources they are responsible for. <message> declarations that refer (in the <part>
elements) to the types defined previously. <port-
Type> element(s) wrap a sequence of <opera-
WEb sErVIcE INtEGrAtION tion> elements representing abstract operations,
each having <input> and <output> (and optional
In the following we briefly overview the most <fault>) elements that refer to the defined mes-
important concepts of web services and introduce sages. <binding> element(s) specify the transport
the modeling language of SINTAGMA called protocol and message formats for the set of opera-
SILan. Then we describe in detail how different tions listed in <portType>. Finally, the <service>
web services can be represented by models and element contains one or more <port> elements,
queried in the system. Finally, we discuss sample each linked to a <binding>, and a network loca-
web service models. tion in the <soap:address> child element. (We use
the soap namespace prefix to indicate elements
Web services belonging to the SOAP URI. WSDL namespace
prefixes are omitted. XSD elements will be pre-
Web services aim to provide some document or fixed by xs.)
procedure-oriented functionality over the network In this chapter, we consider web services
that can be accessed in a standardized way, typi- conforming to Basic Profile Version 1.1 of Web
cally using SOAP (Simple Object Access Proto- Service Interoperability Organization6 . For
col3) message exchange over HTTP. simplicity, we assume document-literal style
SOAP is an XML based protocol for exchang- messaging protocol, one targetNamespace in the
ing structured and typed information between XSD type definition, and one service in the WSDL
peers. SOAP messages consist of an <Envelope> with one port and (potentially) several operations.
element followed by child element <Body>. Body In document-literal style, message contents are
entries will be referred to as the message content entirely defined by the XSD schema within the
throughout this chapter. WSDL. We note that none of the above constraints
The interface of a web service is described in are theoretical limitations of our approach, and are
WSDL (Web Services Description Language4), typically met by web services in practice.
which is based on XML. In WSDL, a set of opera-
tions with input and output messages are described Modeling in sINtAGMA
abstractly to define a network endpoint. These
endpoints are then bound to concrete protocol Different data sources are modeled uniformly in
and message serialization format. A web service SINTAGMA using the modeling language of the
is defined as a collection of ports that are bound system called SILan. This language is based on
to the network endpoints defined previously. UML (Unified Modeling Language, see Fowler
The location of ports and protocol bindings are & Scott 1998) and Description Logics (see Hor-
specified by SOAP extensibility elements in the rocks 2002), and the syntax resembles IDL7, the
WSDL. Messages are defined using the XSD Interface Description Language of CORBA.
(XML Schema Definition5) type system.
975
The main constructs are classes and associa- complex type definition. Simple type definitions
tions, since these are the carriers of information. A restrict built-in XML types (or other simple types)
class denotes a set of entities called the instances by giving enumerations, minimum, maximum
of the class. Similarly, an n-ary association denotes values, etc. Complex type definitions combine a
a set of n-ary tuples of class instances called links. set of element declarations, or declare XML at-
In a binary association one of the connections tributes, respectively. To each element declaration
can be declared composite, which means that cardinality can be assigned to specify optional or
the instance at the composite end is part of the multiple occurrences for the element. Simple and
instance at the other end (and is not part of any complex type definitions can also be extended,
other instance). Composition associations are also restricted by other type definitions. We note that
referred to as compositions for short. Connections there are many features of XSD omitted here for
of associations can be declared as input, which clarity considerations.
means that the association can only be queried if The SINTAGMA model is obtained by
all the input ends are available. Associations also transforming the XSD description to a SILan
have multiplicity that is used to define cardinality representation. A unique SILan class is assigned
constraints, e.g. one-to-one, one-to-many rela- to each complex type with unique name. Simple
tions. Classes and associations have unique name type element declarations (having a built-in XML
within a model. type or simple type definition) within a complex
Classes can have attributes which are defined type are added as SILan attributes of the class.
as functions mapping the class to a subset of values The name of the attribute is given the name at-
allowed by the type of the attribute. Attributes tribute of the element declaration, and the SILan
have unique names within a class and are one of type is derived from the built-in XML type ac-
the SINTAGMA supported types. cording to a predefined mapping. Complex type
Invariants can be specified for classes and element declarations within a complex type are
associations. Invariants give statements about represented by composition associations between
instances of classes (and links of associations) that classes assigned to different complex types.
hold for each of them. Invariants are based on the Compositions are named uniquely, connection
language OCL (Object Constraint Language, see end aliases are given by the name attribute of the
Clark & Warmer 2002.). element declarations (used at navigation), and oc-
currence indicators (minOccurs, maxOccurs) are
Modeling Web services converted to the appropriate multiplicity of the
composition (e.g. 1..1, 1..*). Simple type element
Modeling web services in SINTAGMA basically declarations with multiple occurrences cannot be
means the construction of a SILan representation represented as simple class attributes in SILan.
of data structures and data types used in com- Therefore separate classes are created wrapping
munication with the web service. The schemes simple types that are then connected to the original
of different web service messages are defined by container class by composition association with
an XSD language description in the <schema> the appropriate multiplicity. Optional attributes
element in the WSDL (or imported here). cannot be expressed in SILan. Their values are
The schema typically consists of a set of simply set to null at query (see next section) if
element declarations, simple and complex type they are absent, instead of creating compositions
definitions. Element declarations declare named with optional multiplicity. The default String
XML elements with type corresponding to a type is assigned to XML types that cannot be
built-in XML type (e.g. int, string), simple or represented precisely in SILan (e.g. date). These
976
types will hold the string content of the related erations which are represented by associations.
XML element. Simple type restrictions are added Web service invocation, however, requires ad-
as attribute invariants, complex type extensions ditional details of the WSDL that are stored as
are indicated by inheritance relations between the metadata in the model (used by the Web Service
corresponding classes. Wrapper component). One is the network location
Message schemes modeled above are then of the web service that is obtained from the
associated with web service operations: an as- <soap:address> element of the port. The other is
sociation is created between classes representing the namespace of the XML messages used in the
input and output messages for each operation in communication with the web service that is
the WSDL. Associations are named uniquely, given by the tartgetNamespace attribute of the
and end connections corresponding to input mes- <schema> element.
sages are labeled as <<input>> in SILan (angle In practice, web services can be more compli-
quotes notation corresponds to UML’s stereotype cated. It may occur that a web service uses several
notation). Connection aliases of associations are schemes and namespaces that require introducing
given the element names wrapping input and namespace metadata into different classes instead
output XML messages of the operation (referred of using a single, global namespace in the model.
in <part> element in WSDL). A WSDL can declare several ports combining web
An example WSDL fragment of a simple add service operations at different network locations.
web service is shown in Figure 1 together with In this case, the network location(s) need to be
the created SILan model and the corresponding assigned to the associations representing opera-
UML class diagram. tions instead of the model.
The constructed model contains every single When a web service uses rpc protocol
data that can be passed to or returned by the web (<soap:binding>), <part> elements that are de-
service in terms of classes, compositions and at- clared at input and output messages are wrapped
tributes, as well as the different web service op- first in input and output classes, which are then
Figure 1. An example WSDL fragment of a simple add web service together with the created SILan
model and the corresponding UML class diagram (© 2008 Á. Hajnal, T. Kifor, Luckácsy, L.Z. Varga,
Used with Permission)
977
Box 1.
connected by the operation association. In the
case of document style binding no such problem SELECT addOperation.AddResponse.result,
occurs, since these classes are already created at addOperation.AddResponse.Details.time
FROM addOperation
processing the schema. A single <part> element WHERE addOperation.AddOperation.op1=1 AND
is allowed in message definitions that refer to addOperation.AddOperation.op2=2
them. Web services not conforming to WS-I Basic
Profile, using encoded messaging style, WSDL
arrays, non-SOAP protocols, etc., need further Operation’s input parameters are given by
workaround, which is omitted here for clarity constraints in the WHERE part of the query.
considerations. Constraints use the ‘=’ operator, in the form of
Class.field=value, which have value assignment
Querying Web services semantics with respect to the input parameters.
through sINtAGMA The ‘.’ operator is used to refer to a class attribute
in SILan, but it is also used to navigate along
SILan query language is an object oriented query associations or compositions. For example, Class1.
language designed to formulate queries over association1.field1 denotes attribute field1 in the
SINTAGMA models. The syntax is similar to class referred by association1 in Class1. Naviga-
SQL used at relational databases: a SILan query tion is used to assign values to input parameters
is composed of SELECT, FROM, WHERE parts. starting from the association representing the
The SELECT keyword is followed by a comma operation. In the case of several input parameters,
separated list of class attributes (columns) of inter- the list of assignments is separated by the AND
est, FROM part enumerates classes (tables) whose logical operator. This way, arbitrary complex web
instances (rows) we search among, and WHERE service inputs can be formulated. For example,
part specifies constraints that must be satisfied by input parameters op1, op2 of addOperation are
all the instances in the result. On the other hand, given values by the query shown in Box 2.
SILan is an object oriented language that relies Queries for models representing web services
on UML modeling, and SILan also supports OCL are executed by the Web Service Wrapper com-
expressions in queries by which we can specify ponent. The passed query is parsed first, then the
navigations through objects. appropriate SOAP message is constructed and
In contrast to relational databases functional sent to the web service provider. Starting from
data sources require input to populate the “da- the association in the query and collecting all the
tabase” with data before the query can actually constraints for the input side (navigations towards
be executed. In the case of web services, input the input end) an XML tree is constructed that
includes the name of the web service operation combines all the web service inputs. Navigations
and the input parameters of the operation. are represented by wrapper XML elements, and
When models representing web services are attribute constraints are represented by simple
queried in SINTAGMA, the web service operation
must be given in the FROM part as the associa-
tion representing the operation. For example, the Box 2.
web service operation called addOperation of the
example in Figure 1 is queried by the construct SELECT addOperation.AddResponse.result,
addOperation.AddResponse.Details.time
below (relevant parts are highlighted in boldface FROM addOperation
characters) (see Box 1). WHERE addOperation.AddOperation.op1=1 AND
addOperation.AddOperation.op2=2
978
XML elements with content corresponding to the content of the XML element. If the child element
constant value. Navigations and constraints that corresponds to a composition (complex type), the
refer to the same instance are unified in the XML child node is processed recursively (considering
tree. The namespace of the XML fragment is set the referred class), and a relation is created from
accordingly to the namespace metadata in the the current row to the new row in another table
model (targetNamespace of the schema) that is representing the child node. Class attributes for
then wrapped in an appropriate SOAP envelope. which no appropriate child element can be found
The input SOAP message composed for the are set to null. The textual content of XML ele-
query of the addOperation is shown below: ments are converted to the proper attribute type
at filling field values. The input SOAP message
<SOAP-ENV:Envelope ...> content sent to the web service provider previously
<SOAP-ENV:Body> is also loaded into the internal database in the same
<AddOperation xmlns=”http:// way. An example answer SOAP message and the
add.com”> associated classes, attributes are shown in Box 3.
<op1>1</op1> The query specifies the set of classes and at-
<op2>2</op2> tributes of interest in the FROM and SELECT
</AddOperation> parts. Note that in SILan it is allowed to query
<SOAP-ENV:Body> associations as well as compositions in the FROM
</SOAP-ENV:Envelope> part, and give navigations to attributes in the
SELECT part. The WHERE part declares con-
When the SOAP message is sent to the internet straints for the instances by constant constraints,
location of the web service (stored as metadata where class attributes are compared to constant
in the model), the requested operation will be values using relational operators, or by association
executed by the web service provider. Results are constraints that must hold between instances.
sent back to the Web Service Wrapper component The Web Service Wrapper component, in the
as another SOAP message, and, unless SOAP fault knowledge of the temporary internal database,
(e.g. missing input parameters) or no response can execute the query similarly to an SQL engine.
errors occur, a temporary internal “database” is Basically, the result is constructed by taking the
populated with data. Cartesian product of the instances of the relevant
The internal database is set up by mapping classes (listed in the FROM part). Constraints in
the content of the answer SOAP message to the the WHERE part are checked for each n-tuple
model. First, the XML document root is added of instances, and if all of them are satisfied, the
as instance to the class representing operation’s
output, then child nodes are processed recursively
considering the model schema: XML sub-elements Box 3.
are added as attributes of the current instance, if
they are simple, or added as new instances in the <SOAP-ENV:Envelope ...>
<SOAP-ENV:Body>
corresponding classes, respectively, if they are <AddResponse> ◄ instance in class addResponse
complex. Intuitively, it means that a new row is <result>3</result> ◄ attribute result of the addResponse
instance
created for the root element in the table of op- <Details> ◄ instance in class details
eration output, and field values are obtained by <time>0.01s</time> ◄ attribute time of the details instance
</Details>
iterating through all child elements. If the name of </AddResponse>
the child element corresponds to a class attribute <SOAP-ENV:Body>
</SOAP-ENV:Envelope>
(simple type), the value of the field is given by the
979
selected attributes (SELECT part) are added to Queries for the model are executed transpar-
the result. The result of the query for the add web ently by the wrapper that communicates with the
service operation contains a single row with field web service using SOAP. Necessary inputs are
result containing integer value 3, and field time obtained from the query, the appropriate request
containing the string representing the execution message is constructed, and sent to the web
time of the operation, e.g. 0.01s. service provider automatically. Result data are
extracted from the answer message, and returned
sample Web service to SINTAGMA.
Models in sINtAGMA An example screenshot of the SINTAGMA
system is shown in Figure 2, where Amazon’s
We have implemented the Web Service Wrapper web service is queried.
component for SINTAGMA and applied it to
several web services ranging from simple ones
(such as Google SOAP Search API providing three DIGItAL LIbrAry DEMONstrAtION
operations) to complex ones (such as Amazon E- APPLIcAtION
Commerce Service providing over 30 operations
with transactions). In the previous section we have seen how a
After entering the URL of the WSDL the model single web service can be modeled and que-
of the web service is built automatically. It can ried in SINTAGMA. In this section we show
be viewed, browsed in a graphical user interface, a digital library application that demonstrates
and queries can be composed for the model. SILan the integration of web services with the help of
abstractions can be created by which web services SINTAGMA. The digital library application is
can participate in integration scenarios. Namely, an OpenURL resolver application developed in
the web service model can be connected to other the SINTAGMA project. OpenURL8 is a NISO
models representing different data sources, for standard9 for identifying documents with different
example other web services or relational databases. types of metadata (for example author, title, ISBN,
Figure 2. Amazon’s web service is queried in SINTAGMA (© 2008 Á. Hajnal, T. Kifor, Luckácsy, L.Z.
Varga, Used with Permission)
980
ISSN) using the URL format. The following is manually and the local application models (lower
a sample OpenURL: http://viola.oszk.hu:8080/ models) for the targets (in our case for Amazon
sokk/OpenURL_Servlet?sid=OSZK:LibriVisio and for the Hungarian National Library) using
n&genre=book&aufirst=Jeno&aulast=Rejto&is the wrappers. Then we have created abstractions
bn=963-13-5374-5&title=Quarantine%20in%20 (relations) between the upper model and the lower
the%20Grand%20Hotel&date=2005 models (one for each data source). The OpenURL
The first part of the URL is a link resolver, resolver application queries the upper model and
in the above example viola.oszk.hu:8080/sokk/ sees it as one single target.
OpenURL_Servlet. The other part contains the If there is a new target we only have to gen-
metadata of the documents, in the above example erate a new lower model using the SINTAGMA
parameters like the first and last name of the au- wrapper and create an abstraction between the
thor, the ISBN code and the title of the book. The new model and the upper model. If the protocol
same OpenURL can be created by several sources or the data model of the target changes we only
(for example in articles containing citations from have to regenerate the lower model and modify
the identified document or in a document meta the old mapping (or create a new one) between
database like The European Library10). In our the lower and the upper model. We do not have to
demonstration application the query OpenURL is modify, recompile and redeploy the source code
created by a web based user interface as shown of the client application.
on Figure 3. The document identified by the First we created the following upper (concep-
OpenURL can be located in several target places. tual) model of the openURL resolver in SILAN
In our demonstration application, as shown on (see Box 4).
Figure 3, the target places are the Amazon book This model is simple because it contains an
store, which is a web service information source OpenURLDescription, which contains the param-
providing XML data services, and the Hungar-
ian National Library, which contains a relational Figure 3. Digital library demonstration applica-
database. tion architecture (© 2008 Á. Hajnal, T. Kifor,
The OpenURL resolver has to search in sev- Luckácsy, L.Z. Varga, Used with Permission)
eral places using different protocols and different
data model for the different possible targets of the
document, therefore our OpenURL resolver ap-
plication uses the different wrappers of SIN-
TAGMA to integrate the different data models of
the different protocols.
SINTAGMA executes the queries for the
targets and collects the result into a single uni-
fied model. The OpenURL resolver queries this
unified model only and does not need to know
the details of the lower level protocols and data
models. As long as the OpenURL model remains
the same, the OpenURL resolver does not need
to be changed even if the protocol or the lower
level data model changes.
We have created an OpenURL unified con-
ceptual model in SINTAGMA (upper model)
981
Box 4.
sponding elements in the Amazon model and the
model OpenURLModel { OpenURL, because there are direct connections
class OpenURL { between the elements, except the author name,
attribute Integer id;
};
where the Amazon author name is the concatena-
class OpenURLDescription { tion of the first and last name of the author in the
attribute Integer openURL;
OpenURL model (see Box 5).
attribute String auFirst; The second abstraction is between the applica-
attribute String auLast; tion model of the Zing SRW service of the Hun-
attribute String title;
attribute String issn; garian National Library and the conceptual
attribute String isbn; model of OpenURL. This abstraction is again a
attribute String pubDate;
}; direct mapping between the corresponding ele-
class SearchResult { ments (see Box 6).
attribute Integer openURLDescription; The last step was to query the conceptual
attribute String source; model of OpenURL from our Open URL re-
attribute String author;
attribute String title;
solver. We used the Java library of the distributed
attribute Integer stock; SINTAGMA system to create the SINTAGMA
attribute String link;
attribute String price;
query. The following is the SINTAGMA query
}; of the sample OpenURL mentioned at the begin-
association AnswerOfQuestion { ning of this section (see Box 7).
connection::OpenURLModel::OpenURLDescription as inp;
connection::OpenURLModel::SearchResult [1..1] as outp Creating models and mappings between the
navigable ; models do not need a programmer, only a knowl-
};
}; edge engineer who knows the business area (library
system in our case). This is possible because
SINTAGMA raised the problem to a higher ab-
eters of the query OpenURL, a SearchResult, straction level and the OpenURL resolver can use
which contains the parameters of the search result,
and an AnswerOfQuestion, which connects the
OpenURL query with the search result. The re-
solver queries this model only and does not know Box 5.
about the different data models below this model.
map bundle Amazon_OpenURLModel between Amazon and
Then we generated the lower model of Ama- OpenURLModel {
zon and Hungarian National Library. We only abstraction nev3 (isop: Amazon::ItemSearch_OPERATION,
isr: Amazon::ItemSearchRequest,
had to pass the WSDL of the applications to the items: Amazon::Items,
SINTAGMA wrapper and it created the models item: Amazon::Item ->
ourl: OpenURLModel::OpenURL) {
automatically. These models are very complex constraint
and not shown here, because the data models of isop.itemSearch.Request = isr and
isop.ItemSearch.Items = items and
the Amazon and the Zing SRW (the web service items.Item = item and
used by Hungarian National Library) web services isop.itemSearch.SubscriptionId = “0W2KPT35S-
FFX0RVEK002” and
are complex. isr.ResponseGroup = “Small” and
The next step was to create two abstractions. isr.SearchIndex = “Books”
implies
The first abstraction is between the application isr.Author = ourl.auLast.concat(“ “.concat(ourl.auFirst)) and
model of Amazon and the conceptual model of ourl.result_isbn = item.ASIN and
OpenURL. This abstraction connects the corre- ourl.origin = “Amazon”;
};
982
Box 6. Box 7.
map bundle Zing_OpenURLModel between Zing and select

OpenURLModel { openURLquery.outp.source, openURLquery.outp.author,
abstraction nev1 (op: Zing::SearchRetrieveOperation_OP- openURLquery.outp.title, openURLquery.outp.stock,
ERATION, openURLquery.outp.link, openURLquery.outp.price
in0: Zing::searchRetrieveRequestType -> from
out: OpenURLModel::OpenURL) { openURLquery: OpenURLModel:: AnswerOfQuestion
constraint where
op.searchRetrieveRequestType = in0 and openURLquery.inp.auFirst.contains(\”Jeno\”) and
in0.maximumRecords = “10” and openURLquery.inp.auLast.contains(\”Rejto\”) and
in0.version = “1.1” and openURLquery.inp.title.contains(\”Quarantine%20in%20
in0.recordSchema = “dc” the%20Grand%
implies 20Hotel\”) and openURLquery.inp.isbn.con-
let d = op.SearchRetrieveOperation.records.record.record- tains(\”963-13-5374-5\”)
Data.toMap in and openURLquery.inp.puDate.contains(\”2005\”)
out.result_title = (String)d.get(“title”) AND
in0.query = out.query;
};
};
code of the querying program does not have to
be changed. Creating the models and mappings
between the models does not need a programmer
always the SINTAGMA query to resolve the but a knowledge engineer who can focus on the
OpenURL expression to any target. business area and logic.
cONcLUsION AcKNOWLEDGMENt
This chapter presented how XML data provided The authors acknowledge the support of the
by web services and relational data can be inte- Hungarian NKFP programme of the SINTAGMA
grated. The main tool to integrate web services project under grant number 2/052/2004. We also
with relational data is the Web Service Wrapper would like to thank all the people participating
component of the SINTAGMA Enterprise In- in this project.
formation Integration system. This component
makes easy the integration of XML data services
with relational databases, because the data model rEFErENcEs
of web services is automatically created by the
Web Service Wrapper of the SINTAGMA system. Benkő, T., Lukácsy, G., Fokt, A., Szeredi, P.,
Based on these application level data models a Kilián, I., & Krauth, P. (2003). Information Integra-
knowledge engineer can create a unified conception through Reasoning on Meta-data. Proceeding
tual model of all data sources, as well as abstract of the workshop, AI moves to IA’’, IJCAI 2003,
mapping between the unified conceptual model Acapulco, Mexico, (pp. 65-77).
and the application level models. Then the concep-
tual model can be queried from the SINTAGMA Calvanese, D., De Giacomo, G., Lenzerini, M.,
system which hides the diversity of different data Nardi, D., & Rosati, R. (1998). Description
sources. The set of data sources can be extended Logic Framework for Information Integration,
easily by generating the application level data Principles of Knowledge Representation and
model for the new data source and creating an Reasoning, (pp. 2-13).
abstraction between the new application model
and the existing conceptual model. The source
983
Clark, T., & Warmer, J. (2002). Object Modeling Vasiliu, L., Harand, S., & Cimpian, E. (2004).
with the OCL: The rationale behind the Object The DIP Project: Enabling Systems & Solutions
Constraint Language. Springer. For Processing Digital Content With Semantic
Web Services. EWIMT 2004 European Workshop
Fowler, M., & Scott, K. (1998). UML Distilled: Ap-
on the Integration of Knowledge, Semantics and
plying the Standrad Object Modeling Language.
Digital Media Technology.
Addison-Wesley.
Grigorova, V. (2006). Semantic Description of
Web Services and Possibilities of BPEL4WS. In-
ADDItIONAL rEADING
formation Theories and Application, 13, 183–187.
Horrocks, I. (2002). Reasoning with Expressive Apps, A., & MacIntyre, R. (2006). Why
Description Logics: Theory and Practice. In Pro- OpenURL? D-Lib Magazine, 12(5). doi:10.1045/
ceeding of the 18th International Conference on may2006-apps
Automated Deduction (CADE 2002) (pp. 1-15). Baader, F., Calvanese, D., McGuinness, D., Nardi,
Lukácsy, G., Benkő, T., & Szeredi, P. (2007). D., & Patel-Schneider, P. (2003). The Description
Towards Automatic Semantic integration. In 3rd Logic Handbook, Theory, Implementation and
International Conference of Interoperability for Applications. Cambridge University Press.
Enterprise Software and Applications (I-ESA Booch, G., Jacobson, I., & Rumbaugh, J. (1999).
2007). The Unified Modeling Language User Guide.
Varga, L. Z., Hajnal, A., & Werner, Z. (2004). Addison-Wesley.
An Agent Based Approach for Migrating Web Cerami, E. (2002). Web Services Essentials.
Services to Semantic Web Services. Lecture Notes O’Reilly & Associates.
in Computer Science Vol. 3192, Springer-Verlag
GmbH, Heidelberg, Germany. In C. Bussler & D. Chawathe, S., Garcia-molina, H., Hammer, J., Irel,
Fensel (Eds.), Artificial Intelligence: Methodol- K., Papakonstantinou, Y., Ullman, J., & Widom,
ogy, Systems, and Applications 11th International J. (1994). The Tsimmis Project: Integration of
Conference, AIMSA 2004, Varna, Bulgaria, Sep- Heterogeneous Information Sources. Proceedings
tember 2-4, 2004, Proceedings, (pp. 371-380). of IPSJ Conference. (pp. 7-18).
ISBN-3-540-22959-0. Hepp, M., Leymann, F., Domingue, J., Wahler, A.,
Varga, L. Z., Hajnal, Á., & Werner, Z. (2005). The & Fensel, D. (2005). Semantic Business Process
WSDL2Agent Tool. In R. Unland, M. Klusch, & Management: A Vision Towards Using Semantic
M. Calisti (Eds.), Software Agent-Based Applica- Web Services for Business Process Management.
tions, Platforms and Development Kits. Whitestein Proceedings of the IEEE International Conference
Series in Software Agent Technologies, (pp. 197- on e-Business Engineering (ICEBE 2005), (pp.
223). Viaduktstrasse 42, CH-4051 Basel, Switzer- 535-540). IEEE Computer Society.
land, Springer Group, ISBN 3-7643-7347-4, 2005. Hodgson, C. (2005). Understanding the OpenURL
Framework. NISO Information Standards Quar-
terly, 17(3), 1–4.
Kline, K., & Kline, D. (2001). SQL in a Nutshell.
O’Reilly & Associates.
984
Lukácsy, G., & Szeredi, P. (2008). Combining De- ENDNOtEs

scription Logics and Object Oriented Models in an
Information Framework. Periodica Polytechnica.
1
Web Services Architecture, W3C Working
Group Note 11 February 2004. http://www.
Polleres, A., Pearce, D., Heymans, S., & Ruckhaus, w3.org/TR/2004/NOTE-ws-arch-20040211
E. (2007). Proceedings of the 2nd International 2
SINTAGMA Enterprise Information Integra-
Workshop on Applications of Logic Program- tion System was developed under the Hun-
ming to the Web, Semantic Web and Semantic garian NKFP programme of the SINTAGMA
Web Services (ALPSWS2007). CEUR Workshop project. Web page: http://www.sintagma.hu
Proceedings, Vol. 287, http://ceur-ws.org/Vol-287. (available in Hungarian language only)
T. Ray, E. (2001). Learning XML. O’Reilly &
3
Simple Object Access Protocol. http://www.
Associates. w3.org/TR/2000/NOTE-SOAP-20000508
4
Web Services Description Language.
Ricardo, J. G., Müller, J. P., Mertins, K., & Zelm, http://www.w3.org/TR/2001/NOTE-ws-
M. (2007). Enterprise Interoperability II: New dl-20010315
Challenges and Approaches, Proceedings of the 5
XML Schema. http://www.w3.org/XML/
3rd International Conference on Interoperability Schema.
for Enterprise Software and Applications (IESA- 6
Web Services Interoperability Organization.
07). Springer Verlag. http://www.ws-i.org
St. Laurent, S., Johnston, J., & Dumbill, E. (2001).
7
Object Management Group: The Common
Programming Web Services with XML-RPC. Object Request Broker: Architecture and
O’Reilly & Associates. Specification, revision 2, July 1995.
8
http://alcme.oclc.org/openurl/docs/pdf/
Studer, R., Grimm, S., & Abecker, A. (Eds.). openurl-01.pdf
(2007). Semantic Web Services. Springer. 9
http://www.niso.org/standards/standard_de-
Van der Vlist, E. (2002). XML Schema. O’Reilly tail.cfm?std_id=783
& Associates.
10
http://www.theeuropeanlibrary.org
Walmsley, P. (2007). XQuery. O’Reilly Media.

Walsh, A. E. (2002). UDDI, SOAP, and WSDL:
The web services specification reference book.
Pearson Education.
Management and Best Processes, edited by Patrick Hung, pp. 82-97, copyright 2009 by Business Science Reference (an
985
986
Chapter 4.6
System-of-Systems
Cost Estimation:
Analysis of Lead System
Integrator Engineering Activities
Jo Ann Lane
University of Southern California, USA
Barry Boehm
University of Southern California, USA
AbstrAct
engineering activities described in Electronic
As organizations strive to expand system capa- Industries Alliance (EIA) 632 (“Processes for
bilities through the development of system-of- Engineering a System”). This research further
systems (SoS) architectures, they want to know analyzed effort and schedule issues on “very
“how much effort” and “how long” to implement large” SoS programs to more clearly identify
the SoS. In order to answer these questions, it is and profile the types of activities performed by
important to first understand the types of activi- the typical LSI and to determine organizational
ties performed in SoS architecture development characteristics that significantly impact overall
and integration and how these vary across differ- success and productivity of the LSI effort. The
ent SoS implementations. This article provides results of this effort have been captured in a
results of research conducted to determine types reduced-parameter version of the constructive
of SoS lead system integrator (LSI) activities and SoS integration cost model (COSOSIMO) that
how these differ from the more traditional system estimates LSI SoS engineering (SoSE) effort.
System-of-Systems Cost Estimation
INtrODUctION other legacy systems, commercial-off-the-shelf

(COTS) products, and new systems into a system
As organizations strive to expand system capabili- of systems, generally with the intent to share
ties through the development of system-of-systems information from related systems and to create
(SoS) architectures, they want to know “how much new, emergent capabilities that are not possible
effort” and “how long” to implement the SoS. with the existing stove-piped systems. With this
Efforts are currently underway at the University development approach, we see new activities
of Southern California (USC) Center for Systems being performed to define the new architecture,
and Software Engineering (CSSE) to develop a identify sources to either supply or develop the
cost model to estimate the effort associated with required components, and then to integrate and
SoS lead system integrator (LSI) activities. The test these high level components. Along with this
research described in this article is in support of “system-of-systems” development approach, we
the development of this cost model, the construc- have seen a new role in the development process
tive SoS integration cost model (COSOSIMO). evolve to perform these activities: that of the LSI.
Research conducted to date in this area has A recent Air Force study (United States Air Force
focused more on technical characteristics of the Scientific Advisory Board, 2005) clearly states
SoS. However, feedback from USC CSSE industry that the SoS Engineering (SoSE) effort and focus
affiliates indicates that the extreme complexity related to LSI activities is considerably different
typically associated with SoS architectures and from the more traditional system development
political issues between participating organiza- projects. According to this report, key areas where
tions have a major impact on the LSI effort. This LSI activities are more complex or different than
is also supported by surveys of system acquisition traditional systems engineering are the system
managers (Blanchette, 2005) and studies of failed architecting, especially in the areas of system
programs (Pressman & Wildavsky, 1973). The interoperability and system “ilities;” acquisition
focus of this current research is to further inves- and management; and anticipation of needs.
tigate effort and schedule issues on “very large” Key to developing a cost model such as CO-
SoS programs and to determine key activities SOSIMO is understanding what a “system-of-sys-
in the development of SoSs and organizational tems” is. Early literature research (Jamshidi, 2005)
characteristics that significantly impact overall showed that the term “system-of-systems” can
success and productivity of the program. mean many things across different organizations.
This article first describes the context for For the purposes of the COSOSIMO cost model
the COSOSIMO cost model, then presents a development, the research team has focused on the
conceptual view of the cost model that has been SoS definitions provided in Maier (1999) and Sage
developed using expert judgment, describes the and Cuppan (2001): an evolutionary net-centric
methodology being used to develop the model, architecture that allows geographically distributed
and summarizes conclusions reached to date. component systems to exchange information and
perform tasks within the framework that they are
not capable of performing on their own outside
cOsOsIMO cONtEXt of the framework. This is often referred to as
“emergent behaviors.” Key issues in developing
We are seeing a growing trend in industry and an SoS are the security of information shared
the government agencies to “quickly” incorpo- between the various component systems, how to
rate new technologies and expand the capabili- get the right information to the right destinations
ties of legacy systems by integrating them with efficiently without overwhelming users with un-
987
Figure 1. COSOSIMO model structure
size Drivers sos

Definition and
cOsOsIMO Integration
cost Drivers Effort
calibration
necessary or obsolete information, and how to Recent LSI research conducted by reviewing
maintain dynamic networks so that component LSI statements of work identifies the following
system “nodes” can enter and leave the SoS. typical LSI activities:
Today, there are fairly mature tools to support
the estimation of the effort and schedule associ- • Concurrent engineering of requirements,
ated with the lower-level SoS component systems architecture, and plans
(Boehm, Valerdi, Lane, & Brown 2005). However, • Identification and evaluation of technologies
none of these models supports the estimation of to be integrated
LSI SoSE activities. COSOSIMO, shown in Figure • Source selection of vendors and suppliers
1, is a parametric model currently under develop- • Management and coordination of supplier
ment to compute just this effort. The goal is to activities
support activities for estimating the LSI effort in a • Validation and feasibility assessment of SoS
way that allows users to develop initial estimates architecture
and then conduct tradeoffs based on architecture • Continual integration and test of SoS-level
and development process alternatives. capabilities
• SoS-level implementation planning, prepara-
tion, and execution
Figure 2. Architecture-based SoS cost estimation
Level 0 sOs
Level 1
s1 s2 sm
Level 2 s11 s12 s1n s21 s22 s2n sm1 sm2 smn
Activity Levels cost Model

SoS Lead System Integrator Effort (SoS scoping, planning, Level 0, and other COSOSIMO
requirements, architecting; source selection; teambuilding, re- levels if lower level
architecting, feasibility assurance with selected suppliers; systems
incremental acquisition management; SoS integration and test; components are
transition planning, preparation, and execution; and continuous also SoSs
change, risk, and opportunity management)
Development of SoS Software-Intensive Infrastructure and Level 0 COCOMO II
Integration Tools
System Engineering for SoS Components Levels 1-n COSYSMO
Software Development for Software -Intensive Components Levels 1-n COCOMO II
COTS Assessment and Integration for COTS-based Components Levels 1-n COCOTS
988
• On-going change management at the SoS affiliates have identified key size drivers and cost
level and across the SoS-related integrated drivers for LSI effort and are shown in Table 1.
product teams to support the evolution of Because there are concerns about the avail-
requirements, interfaces and technology. ability of effort data from a sufficient number of
SoS programs to support model calibration and
With the addition of this new cost model to the validation, current efforts are focusing on defining
constructive cost model (COCOMO) suite of cost a “reduced parameter set” cost model or ways to
models, one can easily develop more comprehen- estimate parts of the LSI effort using fewer, but
sive estimates for the total SoS development, as more specific, parameters. The following para-
shown in Figure 2. graphs present the results of this recent research.
Further observations of LSI organizations in-
dicate that the LSI activities can be grouped into
LsI EFFOrt EstIMAtION three areas: 1) planning, requirements manage-
APPrOAcH ment, and architecting (PRA), 2) source selection
and supplier oversight (SS), and 3) SoS integration
As mentioned above, key to an LSI effort estima- and testing (I&T). There are typically different
tion model is having a clear understanding of the parts of the LSI organization that are responsible
SoSE activities performed by the organization as for these three areas. Figure 3 illustrates, con-
well as which activities require the most effort. In ceptually, how efforts for these three areas are
addition, it is important to understand how these distributed across the SoS development life cycle
SoSE activities differ from the more traditional phases of inception, elaboration, construction,
systems engineering activities. Analysis presented and transition for a given increment or evolution
in Lane (2005) describes how the typical LSI of SoS development.
SoSE activities differ from the more traditional Planning, requirements, and architecting begin
system engineering activities identified in EIA early in the life cycle. As the requirements are
632 (Electronic Industries Alliance, 1999) and the refined and the SoS architecture is defined and
Software Engineering Institute (SEI) Capability matured, source selection activities can begin to
Maturity Model Integration (CMMI) (Software identify component system vendors and to issue
Engineering Institute, 2001). Subsequently, Del- contracts to incorporate the necessary SoS-en-
phi surveys conducted with USC CSSE industry abling capabilities. With a mature SoS architecture
Table 1. COSOSIMO cost model parameters
Size Drivers Cost Drivers

• Requirements understanding
• Architecture maturity
• Level of service requirements
• # SoS-related requirements
• Stakeholder team cohesion
• # SoS interface protocols
• SoS team capability
• # independent component
• Maturity of LSI processes
system organizations
• Tool support
• # SoS scenarios
• Cost/schedule compatibility
• # unique component systems
• SoS Risk Resolution
• Component system maturity and stability
• Component system readiness
989
Figure 3. Conceptual LSI effort profile
and the identification of a set of component systems needs and desires. As the effort ramps down in
for the current increment, the integration team can the transition phase, efforts are typically ramping
begin the integration and test planning activities. up for the next increment or evolution.
Once an area ramps up, it continues through the By decomposing the COSOSIMO cost model
transition phase at some nominal level to ensure into three components that correspond to the three
as smooth a transition as possible and to capture primary areas of LSI SoSE effort, the parameter set
lessons learned to support activities and plans for each COSOSIMO component can be reduced
for the next increment. Boehm and Lane (2006) from the full set and the applicable cost drivers
describe how some of these activities directly made more specific to the target area. Table 2 shows
support the current plan-driven SoS development the resulting set of size and cost drivers for each
effort while others are more agile, forward looking, of the three primary areas. This approach allows
trying to anticipate and resolve problems before the model developers to calibrate and validate
they become huge impacts. The goal is to stabi- the model components with fewer parameters
lize development for the current increment while and data sets. It also allows the collection of data
deferring as much change as possible to future sets from organizations that are only responsible
increments. For example, the planning/require- for a part of the LSI SoSE activities. Finally, this
ments/architecture group continues to manage approach to LSI SoSE effort estimation allows
the requirements change traffic that seems to be the cost model to provide estimates for the three
so common in these large systems, only apply- areas, as well as a total estimate—a key request
ing those changes to the current increment that from USC CSSE industry affiliates supporting
are absolutely necessary and deferring the rest this research effort.
to future increments. The architecture team also Detailed definitions and proposed ratings for
monitors current increment activities in order to these parameters may be found in Lane (2006).
make necessary adjustments to the architecture to The following provides a brief description of each
handle cross-cutting technology issues that arise of the COSOSIMO parameters. Note that several
during the component system supplier construc- of the COSOSIMO parameters are similar to those
tion activities. Likewise, the supplier oversight defined for the constructive systems engineering
group continues to monitor the suppliers for risks, cost model (COSYSMO) and are identified in the
cost, and schedule issues that arise out of SoS descriptions below.
conflicts with the component system stakeholder
990
Table 2. COSOSIMO parameters by SoSE area

COSOSIMO
Associated Size Drivers Associated Cost Drivers
Component
• Stakeholder team cohesion
• SoS PRA capability
• # SoS-related requirements
PRA • Maturity of LSI PRA processes
• # SoS interface protocols
• PRA tool support
• Cost/schedule compatibility with
PRA processes
• SoS PRA risk resolution
• # independent component • SoS SS capability
SS system organizations • Maturity of LSI SS processes
• SS tool support
• Cost/schedule compatibility with SS
activities
• SoS SS risk resolution
• SoS I&T capability
• # SoS interface protocols • Maturity of LSI I&T processes
• # SoS scenarios • I&T tool support
I&T • # unique component systems • Cost/schedule compatibility with I&T
activities
• SoS I&T risk resolution
• Component system maturity and
stability
• Component system readiness
cOsOsIMO size Drivers framework. Note: This does NOT include inter-
faces internal to the SoS component systems, but
Number of SoS-Related Requirements1 it does include interfaces external to the SoS and
between the SoS component systems. Also note
This driver represents the number of requirements that this is not a count of total interfaces (in many
for the SoS of interest at the SoS level. Require- SoSs, the total number of interfaces may be very
ments may be functional, performance, feature, dynamic as component systems come and go in
or service-oriented in nature depending on the the SoS environment —in addition, there may be
methodology used for specification. They may multiple instances of a given type of component
also be defined by the customer or contractor. system), but rather a count of distinct protocols
SoS requirements can typically be quantified at the SoS level.
by counting the number of applicable shalls,
wills, shoulds, and mays in the SoS or marketing Number of Independent Component
specification. Note: Some work may be required System Organizations
to decompose requirements to a consistent level
so that they may be counted accurately for the The number of organizations managed by the
appropriate SoS-of-interest. LSI that are providing SoS component systems.
Number of SoS Interface Protocols Number of Operational Scenarios1
The number of distinct net-centric interface This driver represents the number of operational
protocols to be provided/supported by the SoS scenarios that an SoS must satisfy. Such scenarios
991
include both the nominal stimulus-response thread communication latency, interoperability, flexibil-
plus all of the off-nominal threads resulting from ity/adaptability, and reliability. This parameter
bad or missing data, unavailable processes, net- should be evaluated with respect to the scope of
work connections, or other exception-handling the sub-model to which it pertains.
cases. The number of scenarios can typically be
quantified by counting the number of SoS states, Team Cohesion1
modes, and configurations defined in the SoS
concept of operations or by counting the number Represents a multi-attribute parameter, which
of “sea-level” use cases (Cockburn, 2001), includ- includes leadership, shared vision, diversity of
ing off-nominal extensions, developed as part of stakeholders, approval cycles, group dynam-
the operational architecture. ics, integrated product team (IPT) framework,
team dynamics, trust, and amount of change in
Number of Unique Component responsibilities. It further represents the hetero-
Systems geneity in stakeholder community of the end us-
ers, customers, implementers, and development
The number of types of component systems that team. For each sub-model, this parameter should
are planned to operate within the SoS framework. be evaluated with respect to the appropriate LSI
If there are multiple versions of a given type team (e.g., PRA, SS, or I&T).
that have different interfaces, then the different
versions should also be included in the count of Team Capability
component systems.
Represents the anticipated level of team coopera-
cOsOsIMO cost Drivers tion and cohesion, personnel capability, and con-
tinuity, as well as LSI personnel experience with
Requirements Understanding1 the relevant domains, applications, language, and
tools. For each sub-model, this parameter should
This cost driver rates the level of understanding be evaluated with respect to the appropriate LSI
of the SoS requirements by all of the affected team (e.g., PRA, SS, or I&T).
organizations. For the PRA sub-model, it includes
the PRA team as well as the SoS customers and Process Maturity
sponsors, SoS PRA team members, component
system owners, users, and so forth. For the SS A parameter that rates the maturity level and
sub-model, it is the understanding level between completeness of the LSI’s processes and plans.
the LSI and the component system suppliers/ For each sub-model, this parameter should be
vendors. For the I&T sub-model, it is the level of evaluated with respect to the appropriate LSI
understanding between all of the SoS stakeholders team processes (e.g., PRA, SS, or I&T).
with emphasis on the SoS I&T team members.
Tool Support1
Level of Service Requirements1
Indicates the coverage, integration, and maturity
This cost driver rates the difficulty and critical- of the tools in the SoS engineering and manage-
ity of satisfying the ensemble of level of service ment environments. For each sub-model, this
requirements or key performance parameters parameter should be evaluated with respect to
(KPPs), such as security, safety, transaction speed,
992
the tool support available to appropriate LSI team cols in the SoS framework. Two COSOSIMO
(e.g., PRA, SS, or I&T). sub-models use this parameter, and it should be
evaluated in each case with respect to the LSI
Cost/Schedule Compatibility activities covered by the sub-model of interest.
The extent of business or political pressures to Component System Maturity and

reduce the cost and schedule associated with the Stability
LSI’s activities and processes. For each sub-model,
this parameter should be evaluated with respect A multi-attribute parameter that indicates the
to the cost/schedule compatibility for appropriate maturity level of the component systems (num-
LSI team activities (e.g., PRA, SS, or I&T). ber of new component systems versus number
of component systems currently operational in
Risk Resolution other environments), overall compatibility of
the component systems with each other and the
A multi-attribute parameter that represents the SoS interface protocols, the number of major
number of major SoS/LSI risk items, the maturity component system changes being implemented
of the associated risk management and mitigation in parallel with the SoS framework changes, and
plan, compatibility of schedules and budgets, the anticipated change in the component systems
expert availability, tool support, and level of un- during SoS integration activities.
certainty in the risk areas. For each sub-model,
this parameter should be evaluated with respect Component System Readiness
to the risk resolution activities for the associated
LSI team (e.g., PRA, SS, or I&T). This indicates readiness of component systems
for integration. User evaluates level of verification
Architecture Maturity and validation (V&V) that has/will be performed
prior to integration and the level of subsystem
A parameter that represents the level of maturity integration activities that will be performed prior
of the SoS architecture. It includes the level of to integration into the SoS integration lab.
detail of the interface protocols and the level of
understanding of the performance of the proto-
Figure 4. USC CSE cost model development methodology
Analyze existing
literature Concurrency and
Perform
Step 1 Behavioral feedback implied...
analyses
Step 2 Identify relative
significance
Perform expert-judgment
Step 3
Delphi assessment,
formulate a-priori model
Step 4
Gather project
data
Determine Bayesian
Step 5 A-Posteriori model
Step 6 Gather more
data; refine model
Step 7
993
cOsOsIMO cOst MODEL oped. Finally, the expert judgment and actual data
DEVELOPMENt MEtHODOLOGy models are combined using Bayesian techniques.
In this process, more weight is given to expert
The COSOSIMO cost model is being developed judgement when actual data is not consistent or
using the proven cost model development meth- sparse, and more weight is given to actual data
odology developed over the last several years at when the data is fairly consistent and experts do
the USC CSSE. This methodology, described in not strongly agree.
(Boehm, Abts, Brown, Chulani, Clark, & et al., Since technologies and engineering ap-
2002), is illustrated in Figure 4. proaches are constantly evolving, it is important
For COSOSIMO, the literature review has to continue data collection and model analysis
focused on the definitions of SoSs and SoSE; the and update the model when appropriate. Histori-
role and scope of activities typically performed cally, this has led to parameters related to older
by LSIs; and analysis of cost factors used in re- technologies being dropped and new parameters
lated software, systems engineering, and COTS added. In the case of COSOSIMO, it will be im-
integration cost models, as well as related system portant to track the evolution of SoS architectures
dynamics models that investigate candidate SoSE and integration approaches and the development
cost factors. of convergence protocols.
The behavioral analyses determine the poten- For COSOSIMO, each of the sub-models will
tial range of values for the candidate cost drivers go through this development process. Once the
and the relative impact that each has on the overall sub-models are calibrated and validated, they
effort associated with the relevant SOSE activities. may be combined to estimate the total LSI effort
For example, if the stakeholder team cohesion is for a proposed SoS development program. To
very high, what is the impact on the PRA effort? date, several expert judgment surveys have been
Likewise, if the stakeholder team cohesion is conducted and actual data collection is in process.
very low, what is the resulting impact on PRA
effort? The results of the behavioral analyses are
then used to develop a preliminary model form. cONcLUsION
The parameters include a set of one or more size
drivers, a set of exponential scale factors, and a LSI organizations are realizing that if more
set of effort multipliers. Cost drivers that are re- traditional processes are used to architect and
lated to economies/diseconomies of scale as size integrate SoSs, it will take too long and too much
is increased are combined into an exponential effort to find optimal solutions and build them.
factor. Other cost drivers that have a more linear Preliminary analysis of LSI activities show that
behavior with respect to size drivers are combined while many of the LSI activities are similar to
into an effort multiplier. those described in EIA 632 and the SEI’s CMMI,
Next, the model parameters, definitions, LSIs are identifying ways to combine agile
range of values, rating scales, and behaviors are processes with traditional processes to increase
reviewed with industry and research experts us- concurrency, reduce risk, and further compress
ing a wideband Delphi process. The consensus overall schedules. In addition, effort profiles for
of the experts is used to update the preliminary the key LSI activities (the up-front effort associ-
model. In addition to expert judgement, actual ated with SoS abstraction, architecting, source
effort data is collected from successful projects selection, systems acquisition, and supplier and
covering the LSI activities of interest. A second vendor oversight during development, as well as
model, based on actual data fitting, is then devel- the effort associated with the later activities of
994
integration, test, and change management) show Lane, J. (2005). System of systems lead system
that the percentage of time spent on key activities integrators: Where do they spend their time and
differs considerably from the more traditional sys- what makes them more/less efficient. (Tech. Rep.
tem engineering efforts. By capturing the effects No. 2005-508.) University of Southern California
of these differences in organizational structure Center for Systems and Software Engineering,
and system engineering processes in a reduced Los Angeles, CA.
parameter version of COSOSIMO, management
Lane, J. (2006). COSOSIMO Parameter Defini-
will have a tool that will better predict LSI SoSE
tions. (Tech. Rep. No. 2006-606). University
effort and to conduct “what if” comparisons of
of Southern California Center for Systems and
different development strategies.
Software Engineering, Los Angeles, CA.
Maier, M. (1998). Architecting principles for
rEFErENcEs systems-of-systems. Systems Engineering, 1(4),
267-284.
Blanchette, S. (2005). U.S. Army acquisition – The
Pressman, J., & Wildavsky, A. (1973). Implementa-
program executive officer perspective, (Special
tion: How great expectations in Washington are
Report CMU/SEI-2005-SR-002). Pittsburgh, PA:
dashed in Oakland. Oakland, CA: University of
Software Engineering Institute.
California Press.
Boehm, B., Abt, C., Brown, A., Chulani, S., Clark,
Sage, A., and Cuppan, C. (2001). On the systems
& et al. (2000). Software cost estimation with CO-
engineering and management of systems of sys-
COMO II. Upper Saddle River, NJ: Prentice Hall.
tems and federations of systems. Information,
Boehm, B., Valerdi, R., Lane, J., & Brown, A. Knowledge, and Systems Management 2, 325-345.
(2005). COCOMO suite methodology and evolu-
Software Engineering Institute (2001). Capabil-
tion. CrossTalk, 18(4), 20-25.
ity maturity model integration (CMMI) (Special
Boehm, B., & Lane, J. (2006). 21st century report CMU/SEI-2002-TR-001). Pittsburgh, PA:
processes for acquiring 21st century systems of Software Engineering Institute.
systems. CrossTalk, 19(5), 4-9.
United States Air Force Scientific Advisory Board
Cockburn, A. (2001). Writing effective use cases. (2005). Report on system-of-systems engineering
Boston: Addison-Wesley. for Air Force capability development. (Public
Release SAB-TR-05-04). Washington, DC:
Electronic Industries Alliance. (1999). EIA Stan-
HQUSAF/SB.
dard 632: Processes for engineering a system.
Valerdi, R (2005). The constructive systems en-
Jamshidi, M. (2005). System-of-systems engineer-
gineering cost model (COSYSMO). Unpublished
ing - A definition. Proceedings of IEEE System,
doctoral dissertation, University of Southern
Man, and Cybernetics (SMC) Conference. Re-
California, Los Angeles.
trieved January 29, 2005 from http://ieeesmc2005.
unm.edu/SoSE_Defn.htm
995
ENDNOtE
1
Adapted to SoS environment from CO-
SYSMO (Valerdi, 2005).
This work was previously published in Encyclopedia of Emerging Systems Approaches in Information Technologies: Con-
cepts, Theories, and Applications, edited by David Paradice, pp. 204-213, copyright 2010 by Information Science Reference
996
997
Chapter 4.7
Consistency and Modularity
in Mediated Service-Based
Data Integration Solutions
Yaoling Zhu
Dublin City University, Ireland
Claus Pahl
Dublin City University, Ireland
AbstrAct on semantic data models as a mediator service

into a Web service-oriented information system
A major aim of the Web service platform is the architecture. Automation through consistency-
integration of existing software and information oriented semantic data models and flexibility
systems. Data integration is a central aspect in this through modular declarative data transformations
context. Traditional techniques for information and are the key enablers of the approach.
data transformation are, however, not sufficient to
provide flexible and automatable data integration
solutions for Web service-enabled information sys- INtrODUctION
tems. The difficulties arise from a high degree of
complexity in data structures in many applications A major aim of the Web service platform is the
and from the additional problem of heterogeneity integration of existing software and information
of data representation in applications that often systems (Alonso et al., 2004). Information and
cross organisational boundaries. The authors data integration is a central aspect in this context.
present an integration technique that embeds a Traditional techniques based on XML for data
declarative data transformation technique based representation and XSLT for transformations
between XML documents are not sufficient to
DOI: 10.4018/978-1-60566-330-2.ch006 provide a flexible and automatable data integra-
Consistency and Modularity in Mediated Service-Based Data Integration Solutions
tion solution for Web service-enabled information implementations provides the necessary flexibil-
systems. Difficulties arise from the high degree ity and modularity to deal with complexity and
of complexity in data structures in many business consistency problems. Two central questions to
and technology applications and from the problem the data integration problem and its automation
of heterogeneity of data representation in applica- shall be addressed in this investigation:
tions that cross organisational boundaries.
The emergence of the Web services platform • How to construct data model transforma-
and service-oriented architecture (SOA) as an ar- tion rules and how to express these rules in
chitecture paradigm has provided a unified way to a formal, but also accessible and maintain-
expose the data and functionality of an information able way is central.
system (Stal, 2002). The Web services platform • How integration can be facilitated through
has the potential to solve the problems in the data service composition to enable interoper-
integration domain such as heterogeneity and ability through connector and relationship
interoperability (Orriens, Yang and Papazoglou, modelling.
2003; Haller, Cimpian, Mocan, Oren and Bussler,
2005; Zhu et al., 2004). Our contribution is an in- We show how ontology-based semantic data
tegration technology framework for Web-enabled models and a specific declarative data query and
information systems comprising of transformation language called Xcerpt (Bry and
Schaffert, 2002) and its execution environment can
• Firstly, a data integration technique based be combined in order to allow dynamic data trans-
on semantic, ontology-based data models formation and integration. We focus on technical
and the declarative specification of trans- solutions to semantically enhance data modelling
formation rules and and adapt Xcerpt and its support environment
• Secondly, a mediator architecture based on so that it can facilitate the dynamic generation
information services and the construction of Xcerpt query programs (in response to user
of connectors that handle the transforma- requests) from abstract transformation rules.
tions to implement the integration process.
A data integration technique in the form of bAcKGrOUND

a mediator service can dynamically perform
transformations based on a unified semantic data Information integration is the problem of com-
model built on top of individual data models in bining heterogeneous data residing at different
heterogeneous environments (Wiederhold, 1992). sources in order to provide the user with a unified
Abstraction has been used successfully to address view (Lenzerini, 2002). This view is central in
flexibility problems in data processing (Rouvel- any attempt to adapt services and their underlying
lou, Degenaro, Rasmus, Ehnebuske and McKee, data sources to specific client and provider needs.
2000). With recent advances in abstract, declara- One of the main tasks in information integration
tive XML-based data query and transformation is to define the mappings between the individual
languages (Zhu et al., 2004) and Semantic Web data sources and the unified view of these sources
and ontology technology (Daconta, Obrst and and vice versa to enable this required adaptation.
Smith, 2003), the respective results are ready to Figure1 shows two sample schemas, which might
be utilised in the Web application context. The represent the views of client and provider on a
combination of declarative and semantic speci- collection of customers, that require integration.
fication and automated support of architecture
998
The integration itself can be defined using trans- XSLT is the most widely used XML data inte-
formation languages. gration language, but suffers from some limitations
Information integration has the objective of within our context due its is syntactical focus and
bringing together different types of data from operational language.
different sources in order for this data to be ac-
cessed, queried, processed and analysed in a • Semantics: Only the syntactical integra-
uniform manner. Recently, service-based plat- tion of query and construction part of a
forms are being used to provide integration solu- XSLT transformation program is specified,
tions. In the Web services context, data in XML but consistency in terms of the semantics
representation, which is retrieved from individu- can not be guaranteed.
al Web-based data services, needs to be merged • Modularity: XSLT does not support a join
and transformed to meet the integration require- or composition operator on XML docu-
ments. Data schema integration cannot be fully ments that allows several source XML
automated on a syntactic level since the syntactic documents to merged into one before be-
representation of schemas and data does not con- ing transformed.
vey the semantics of different data sources. For • Maintainability: XSLT transformations
instance, a customer can be identified in the con- are difficult to write, maintain, and reuse
figuration management repository by a unique for large-scale information integration. It
customer identifier; or, the same customer may is difficult to separate the source and target
be identified in the problem management reposi- parts of transformation rules as well as the
tory by a combination of a service support identi- filtering constraints due to its operational
fier and its geographical location, see Figure 1. character without a separation of query and
Ontology-based semantic data models can rec- construction concerns.
tify this problem by providing an agreed vocabu-
lary of concepts with associated properties.
Figure 1. Two schema diagrams of the global data model that need to be integrated (© 2008, Claus
Pahl. Used with permission).
999
Due to these drawbacks, we propose semantic are represented as concepts in the ontology. The
data models and a declarative query and transfor- concept Customer is defined in terms of its proper-
mation approach providing more expressive power ties – data type-like properties such as a name or
and the ability to automatically generate query an identifier and also object type properties such
and transformation programs as connectors for as a collection of services used by a customer.
services-based data integration in Web-enabled Three concept descriptions, using the existential
information systems. A range of characteristics of quantifier “∃” here, express that a customer is
XML query and transformation languages beyond linked to an identification through a supportID
XSLT, which have been studied and compared property, to a name using the custName property,
(Jhingran, Mattos and Pirahesh, 2002; Lenzerini, and to services using Services. In some cases,
2002; Peltier, Bezivin, and Guillaume, 2002), led these properties refer to other composite concepts,
us to choose the fully declarative language Xcerpt sometimes they refer to atomic concepts that act
(Bry and Schaffert, 2002) as our transformation as type names here. Technically, the existential
platform (Zhu, 2007). quantification means that there exits for instance
a name that is a customer name.
DAtA trANsFOrMAtION AND Customer =

cONNEctOr ArcHItEctUrE ∃ supportID . Identification
∧
Mappings between data schemas of different par- ∃ custName . Name ∧
ticipants might or might not represent the same ∃ usedServices . Service
semantical information. The Semantic Web and in Service =
particular ontology-based data domain and service ∃ custID . ID ∧
models (Daconta et al., 2003) can provide input for ∃ servSystem . System
improvements of current integration approaches System =
in terms of data modelling and transformation ∃ hasPart . Machine
validation by providing a notion of consistency,
based on which an automated transformation ap- The ontology represents syntactical and se-
proach can become reliable (Reynaud, Sirot and mantical properties of a common overarching data
Vodislav, 2001, Haller et al., 2005). We define model, which is agreed upon by all participants
consistency here as the preservation of semantics such as service (or data) provider and consumer.
in transformations. This model is actually a domain ontology, cap-
turing central concepts of a domain and defining
Information Architecture them semantically. This means that all individual
XML data models can be mapped onto this com-
Ontologies are knowledge representation frame- mon semantic model. These mappings can then
works that represent knowledge about a domain in be used to automatically generate transforma-
terms of concepts and properties of these concepts. tions between different concrete participant data
We use a description logic notation here, which is models. The overall information architecture is
the formal foundation of many ontology languages summarised in Figure 2.
such as OWL (Daconta et al., 2003). Description Although there is a standardised OWL-based
logic provides us with a concise notation here to equivalent for our description logic ontology, for
express a semantic data model. The elements of practical reasons a corresponding semantically
the XML data models of each of the participants equivalent XML representation is needed. The
1000
Figure 2. Information architecture overview (© 2008, Claus Pahl. Used with permission)
corresponding global XML schema representation between different participant data representa-
for the customer element is: tions can be derived. This construction needs to
address a number of specific objectives regarding
<!ELEMENT Customer (Service, the transformation rules:
System) >
<!ATTLIST Customer • Modularity of transformation rules is need-
supportID ID ed for the flexible generation and configu-
custName Name > ration of transformation rules by allowing
these rules to be specific to particular data
Here, the principle of this mapping becomes elements,
clear: ontology concepts are mapped to XML • Consistency needs to be addressed for the
elements and specific predefined atomic con- reliable generation and configuration of
cepts serve to represent simple properties that are transformation rules by allowing seman-
mapped to XML attributes. We have focused on tics-preserving rules to be constructed
the core elements of ontologies and XML data here automatically.
to highlight the principles. Description elements
of XML such as different types of attributes or Based on a data-oriented domain ontology and
option and iteration in element definition can also two given local data models (source and target,
be captured through a refined property language. expressed as XML schemas) that are mapped
In particular the Web Ontology Language OWL onto the ontology, the rule construction process
provides such constructs (W3C, 2004). is based on three steps:
transformation rule construction 1. Define one transformation rule per concept

in the ontology that is represented in the
The ontology provides a semantically defined target data model.
global data model from which transformations
1001
2. Identify semantically equivalent concepts Xcerpt background

of the selected concepts in the source data
model. We describe Xcerpt principles and the rationale for
3. For each identified concept: choosing it and demonstrate how such a declarative
a. determine required attributes – these are language and its environment need to be adapted
end nodes of the ontological structure, for their deployment in a dynamic, mediated ser-
b. copy semantically equivalent counter- vice context. Xcerpt is a query language designed
parts from the source model. for querying and transforming traditional XML
and HTML data, as well as Semantic Web data
A necessary prerequisite is that all concepts in the form of RDF and OWL. One of the design
of the source model are actually supported by the principles is to strictly separate the matching
target data model. Otherwise, the transformation part and the construction part in a transforma-
definition cannot be completed. tion specification, see Figure 3. Xcerpt follows
The transformation rules based on the sample a pattern-based approach to querying XML data.
ontology for the given customer example will Figure 3 shows a transformation example for
be presented later on once the transformation a customer array based on Figure 1. The structure
language is introduced. These could be formu- of this specification is based on a construction
lated such that data integration problem depicted part (CONSTRUCT) and a source query part
in Figure 1 is formally defined. The mappings (FROM). An output customer in CustomerArray
between participant data models and the data is constructed based on the elements of an item
ontology define semantically equivalent repre- in an arrayOfCustomer by using a pattern match-
sentation of common agreed ontology elements ing approach, identifying relevant attributes in
in the data models. Consequently, the presented the source and referring to them in the construct-
rule construction process is consistent in that it ed output through variables such as Name or
preserves the semantics in transformations. CompanyID. During transformation, these hold
The concrete target of this construction is the concrete values of the selected (matched)
the chosen declarative transformation language elements.
Xcerpt. The construction process has been ex-
pressed herein abstract terms – a complete speci-
Figure 3. Declarative query and transformation
fication in terms of transformation languages such
specification of a customer array element in Xcerpt
as QVT or even Xcerpt itself would have been
(© 2008, Claus Pahl. Used with permission)
too verbose for this context. Declarativeness and
modularity provide the required flexibility for our
solution, in addition to consistency that has been
addressed through the semantic ontology-based
data models. The construction of transformation
rules is actually only the first step in the provision
of XML data integration. These transformations
can be constructed prior to the customer query
construction and stored in rule repositories.
1002
Xcerpt distinguishes two types of specifica- and transformation programs that integrate dif-
tions: ferent data services:
• Goal-based query programs, identified by • In order to promote modularity and code

the keyword GOAL, are executable query reuse, individual integration rules should
programs that refer to input and output re- not be designed to perform complex trans-
sources and that describe data extraction formation tasks – rather a composition of
and construction. individual rules is preferable. The compo-
• Abstract transformation rules, identified by sition of rules through rule chaining de-
the keyword CONSTRUCT as in Figure 3, mand the query part of a service connector
are function-like transformation specifica- to be built ahead of the construction part.
tions with no output resource associated. • The data representation of the global data
model changes as element names change or
Xcerpt extends the pattern-based approach, elements are being removed – these should
which is also used in other query and transforma- not affect the query and integration part of
tion languages, in following ways: the rules. Only an additional construction
part is needed to enable versioning of the
• Firstly, query patterns can be formulated global data model.
as incomplete specifications in three di-
mensions. Incomplete query specifications Modularity and incomplete query specifica-
can be represented in depth, which allows tions turn out to be essential features that are
XML data to be selected at any arbitrary required from a query and transformation language
depth; in breadth, which allows querying in our context. In order to achieve the composi-
neighbouring nodes by using wildcards, tionality of modular rules, a layered approach
and in order. Incomplete query specifica- shall be taken:
tions allow patterns to be specified more
flexibly without losing accuracy. • Ground rules are responsible for populat-
• Secondly, the simulation unification coming XML data in the form of Xcerpt data
putes answer substitutions for the variables terms by reading XML documents from
in the query pattern against underlying individual service providers. These ground
XML terms. rules are tightly coupled to individual data
Web services. These rules instruct the con-
Xcerpt provides a runtime environment with nector where to retrieve elements of data
an execution engine at its core (Schaffert, 2004). objects.
The central problem is to embed this type of envi- • The Xcerpt data terms are consumed sub-
ronment, which can also be found for other query sequently by non-ground queries based on
and transformation languages, into a dynamic, intermediate composite rules. These rules
mediated service setting. are responsible for integrating ground rules
to render data types in the global XML
connector construction schema. However, these rules still do not
and Query composition produce output.
• Finally, the composite rules are responsi-
We have adapted Xcerpt to support the construc- ble for rendering the data objects defined in
tion of service connectors, i.e. executable query the interfaces of the mediator Web services
1003
based on customer requests. The compos- represents a customer as identified in a servicing

ite rules are views on top of ground and system. In the example, rule CustomerArray is a
intermediate representations according to composite rule, based on the Customer and Service
the global schema. Therefore, the exported rules, that could be used to answer a user query
data from a mediator Web service is the directly. The resource identifiers in form of vari-
goal of the corresponding connector (a ables and the interfaces for the data representation
query program). will be supplied to the connector generator. Rule
mappings in the connector generator determine
Xcerpt is a document-centric language, de- which queries are constructed from the repository
signed to query and transform XML documents. for execution.
Therefore, ground rules, which read individual
data elements from the resources, are associ-
ated to at least one resource identifier. This is a tHE MEDIAtED sErVIcE
bottom-up approach in terms of data population INtEGrAtION ArcHItEctUrE
because data is assigned from the bottom level of
the rules upward until it reaches the ultimate goal We propose a mediated service-based architecture
of a hierarchically structured rule. These rules are for the integration of XML data in Web service-
defined through an integration goal (the top-level based information systems. The major aims of
query program) and structured into sub-rules down the proposed mediated software architecture for
to ground rules. the integration and mediation of XML data in the
These layered rules are saved in a repository. context of Web services are threefold: improved
When needed, a rule will be picked and a back- modifiability through declarative rule-based query
ward rule chaining technique for rule composition programs, improved reusability of declarative
enables data objects to be populated to answer integration rules through automated connector
transformation requests. Rule chaining means that construction, and improved flexibility through
resulting variable bindings from a transforma- dynamic generation of consistent, i.e. semantics-
tion rule that is used within a query program are preserving connectors.
chained with those of the query program itself.
Rule chaining is used to build recursive query service-based Mediator
programs. Consistent connectors can then be Architectures
constructed on the fly based on input data such
as the data services and the layered rules. A declarative, rule-based approach can be applied
We apply backward goal-based rule chaining to the data transformation problem (Orriens et
to execute complex queries based on composite al., 2003, Peltier et al., 2002). The difficulty lies
rules. Figure 4 shows an example of this pattern in embedding a declarative transformation ap-
matching approach that separates a possibly partial proach into a service-based architecture in which
query into resource and construction parts. The clients, mediators, and data provider services are
transformation rule maps the supportIdentifier composed (Garcia-Molina et al., 1997). A data
element of the customer example from Figure 1. integration engine can be built in the Web service
Figure 4 is a composite rule based on the Sup- business process execution language WS-BPEL.
portIdentifier construction rule at a lower level. In (Rosenberg and Dustdar, 2005), a business
Figure 5 demonstrates the transformation that rule engine-based approach is introduced to
produces the resulting XML data for the Customer separate the business logic from the executable
service. The output from the Customer mediator WS-BPEL process, which demonstrates that one
1004
Figure 4. Transformation specification in Xcerpt based on goal chaining with one goal-based query
program and two supporting transformation rules (© 2008, Claus Pahl. Used with permission)
Figure 5. The composite rules for customer transformation in Xcerpt (© 2008, Claus Pahl. Used with
permission)
1005
of our objectives can be achieved (Rouvellou et a query/transformation program that com-

al., 2000). These rules, stored in a repository, can piles and translates the incoming data into
be used to dynamically create executable query the required output format.
and transformation programs using a consistency- • The process execution engine is respon-
guaranteeing connector or integration service sible for the integration of XML data and
as the mediator. These integration services are mediation between clients, data providers
the cornerstones of a mediator architecture that and the connector component. The execu-
processes composite client queries that possibly tion engine is implemented in WS-BPEL
involve different data sources provided by different and shall access the Xcerpt runtime engine,
Web services (Wiederhold, 1992). Mediators in which executes the generated query/trans-
an architecture harmonise and present the infor- formation program.
mation available in heterogeneous data sources
(Stern and Davies, 2003). This harmonisation The connector construction component is
comes in the form of an identification of semantic responsible for converting the client query, dy-
similarities in data while masking their syntactic namically create a transformation program based
differences. Figures 1 and 2 have illustrated an on stored declarative transformation rules, and to
example whose foundations we have defined in pass all XML data and programs to the execution
terms of an ontology in order to guarantee con- engine. The system architecture is explained in
sistency for transformations. Figure 6 with a few sample information services
Zhu et al. (2004) and Widom (1995) argue from an application service provider scenario –
that traditional data integration approaches such Customer Data, E-business System, and Request
as federated schema systems and data ware- Analysis Service.
houses fail to meet the requirements of constantly Exposing data sources as services is only the
changing and adaptive environments. With the first step towards building a SOA solution. With-
support of Web service technology, however, it out a service integrator, the data user needs to
is possible to encapsulate integration logic in a understand each of the data models and relation-
separate component as a mediator Web service ships of service providers. The mediator archi-
between heterogeneous data service providers tecture has the following components:
and consumers. Therefore, we build a connector
construction component as a separate integration • Query service. The query service is re-
service, based on (Szyperski, 2002; Haller et al. sponsible for handling inbound requests
2005, Zhu et al., 2004, Rosenberg and Dustdar from the application consumer side and
2005). We develop an architecture where broker or transferring outbound results back. The
mediator functionality is provided by a connector WS-BPEL process engine handles the in-
generator and a transformation engine: ternal messaging of the architecture. The
query service decomposes input mes-
• The connector construction is responsible sages into a set of pre-defined WS-BPEL
for providing connectors based on transfor- processes.
mation rules to integrate and mediate XML • Mediator (BPEL) engine. A media-
documents. The connector construction tor engine is itself a WS-BPEL process.
generates, based on schema information Mediators deliver data according to a
and transformation rules, an executable global schema. The schema may consist
service process that gathers information of various data entities for large enterprise
from the required resources and generates integration solutions.
1006
Figure 6. Component view of a mediator Web service with interactions (© 2008, Claus Pahl. Used with
permission)
• Connector generation service. This com- by composing each Xcerpt query with the cor-
ponent is responsible for generating con- responding transformation rules. In an Xcerpt
nectors for transforming messages both en- query program, there is only one goal query,
tering the WS-BPEL engine from service which will be processed first. The goal query is
clients and leaving the WS-BPEL engine made up of composite transformations rules that
from data provider services according to in turn are made up of ground rules that read XML
the global data model. data from external resources. The process begins
by expanding each composite query according to
The active components, provided as informa- the definitional data mappings that are stored in
tion services, are complemented by two reposi- a rule repository. The rule chaining mechanism
tories: in Xcerpt needs the goal query and all supporting
queries in one query program at runtime.
• Transformation rule repository. The re- The Xcerpt runtime engine reads XML-based
pository allows the reuse of rules and can resources and populates them into data terms
support multiple versions of service pro- before the query terms can start to evaluate them.
viders and mediator services. The drawback is that all resources identifiers have
• Schema repository. The repository stores to be specified inside a query program rather than
the WSDL metadata and the XML schema be passed into a query program as parameters.
information for the Web service provid- Consequently, we adapted the Xcerpt approach to
ers and the mediator Web service. The processing transformation requests in an informa-
schema information is used to validate the tion integration solution. The resource identifiers
XML documents at runtime before they are not hard-coded in ground rules in our setting
are integrated and returned to the client in order to achieve the desired loose coupling to
applications. achieve flexibility and reusability. These resource
identifiers are invisible to the connector construc-
connector Generation tion service. Xcerpt does not support automatic
query program construction by default, although
The construction of a service connector means it provides the necessary backward rule chaining
to generate an executable Xcerpt query program technique to evaluate a chain of queries.
1007
We have developed a wrapper mechanism to in sufficient detail in the context of transformation

pass the resource identifiers from the goal level and re-engineering.
down to the ground rules. Therefore, as an exten- The utilisation of the semantic knowledge that
sion to the original Xcerpt approach, a mediator- is available to represent the services that make up
based data integration architecture is needed where the mediator architecture is another promising
the rules are decoupled from the resources and direction that would increase flexibility in terms
the only the generated Xcerpt-based connectors of dynamic composition. The functionality and
are integrated with the client and provider Web quality attributes of Web services can be in terms
services. WS-BPEL code that coordinates the of one of the widely known service ontologies
mediation and transformation process is gener- such as OWL-S or WSMO (Payne and Lassila,
ated by a connector generator for transformations 2004). Abstract service descriptions can be derived
within the mediator service. from the semantic properties of the data they
provide, process, or consume. Some progress
has been made with respect to semantics-based
FUtUrE trENDs service discovery and composition; the interplay
between semantic data integeration and semantic
Integration has currently been investigated from service integration needs a deeper investigation.
a static perspective looking at existing systems Karastoyanova et al. (2007), for instance, discuss
integration. We discuss emerging needs to ad- a middleware architecture to support semantic
dress this as part of software evolution and legacy data mediation based on semantically annotated
systems integration. Another current trend is the services. Their investigation demonstrates how
increasing utilisation of semantic enhancements, your semantic data mediation can be incorporated
such as ontologies and reasoning frameworks, to into a service-based middleware architecture that
support integration. We address briefly attempts supports SOA-based development. However, the
of using service ontologies, which would comple- need to have an overarching semantic informa-
ment the presented ontology-based information tion architecture also becomes apparent, which
architecture. supports our results.
Re-engineering and the integration of legacy
systems is an aspect that goes beyond the integra-
tion context we described – although the applica- cONcLUsION
tion service provider (ASP) context is a typical
example of a field where ASPs currently convert The benefit of information systems on demand
their systems into service-based architectures must be supported by corresponding informa-
(Seltsikas and Currie, 2002). The introduction of tion management services. Many application
data transformation techniques for re-engineering service providers are currently modifying their
activities can improve the process of re-engineer- technical infrastructures to manage and integrate
ing legacy systems and adopting service-oriented information using a Web services-based approach.
architecture to manage the information technology However, the question of handling information
services (Zhang and Yang, 2004). Business rules integration in a flexible and modifiable way in
often change rapidly – requiring the integration of the context of service-based information systems
legacy systems to deliver a new service. How to has not yet been fully explored.
handle the information integration in the context The presented framework utilises semantic
of service management has not yet been explored information integration technologies for XML data
in service-oriented software architectures. The
1008
crucial solutions for the information integration Bry, F., & Schaffert, S. (2002). Towards a De-
problem are drawn from mediated architectures clarative Query and Transformation Language
and data model transformation, allowing the for XML and Semistructured Data: Simulation
XML data from local schemas to be consistently Unification. In Proceedings Intl. Conference on
transformed, merged and adapted according to Logic Programming. LNCS 2401, (pp. 255-270).
declarative, rule-based integration schemas for Heidelberg, Gerrnany: Springer-Verlag.
dynamic and heterogeneous environments. We
Daconta, M. C., Obrst, L. J., & Smith, K. T. (2003).
have proposed a declarative style of transformation
The Semantic Web – a Guide to the Future of
based on a semantic, ontology-based data model,
XML, Web Services, and Knowledge Management.
with implicit source model traversal and target
Indianapolis, USA: Wiley & Sons.
object creation. The development of a flexible
mediator service is crucial for the success of the Garcia-Molina, H., Papakonstantinou, Y., Quass,
service-based information systems architecture D., Rajaraman, A., Sagiv, Y., & Ullman, Y. D.
from the deployment point of view. Our solution (1997). The TSIMMIS approach to mediation:
based on the query and transformation language Data models and languages. Journal of Intel-
Xcerpt is meant to provide a template for other ligent Information Systems, 8(2), 117–132.
similar languages. One of our central objectives doi:10.1023/A:1008683107812
was to introduce an integration solution from a
Haller, A., Cimpian, E., Mocan, A., Oren, E., &
technical perspective.
Bussler, C. (2005). WSMX - a semantic service-
A number of extensions of our approach would
oriented architecture. In Proceedings Intl. Confer-
strongly benefit its flexibility. Essentially, we
ence on Web Services ICWS 200 5, (pp. 321-328).
plan to address the trends outlined in the previ-
ous section. Systems evolution and legacy system Jhingran, A. D., Mattos, D., & Pirahesh, N. H.
integration shall be addressed through a more (2002). Information Integration: A research
transformation systems-oriented perspective on agenda. IBM Systems Journal, 41(4), 55–62.
integration. We are also working on an integration
Karastoyanova, D., Wetzstein, B., van Lessen, T.,
of service ontologies and general data-oriented do-
Wutke, D., Nitzsche, J., & Leymann, F. (2007).
main ontologies for service-oriented architectures.
Semantic Service Bus: Architecture and Imple-
mentation of a Next Generation Middleware. In
Proceedings of the Second International Workshop
rEFErENcEs
on Service Engineering SEIW 2007, (pp. 347-354).
W3C – the World Wide Web Consortium. (2004). Lenzerini, M. (2002). Data integration: A theo-
The Semantic Web Initiative. Retrieved March 9, retical perspective. In Proceedings Principles of
2008 from http://www.w3.org/2001/sw. Database Systems Conference PODS’02, (pp.
Alonso, G., Casati, F., Kuno, H., & Machiraju, V. 233-246).
(2004). Web Services – Concepts, Architectures Orriens, B., Yang, J., & Papazoglou, M. (2003).
and Applications. Berlin, Germany: Springer A Framework for Business Rule Driven Web
Verlag. Service Composition. Jeusfeld, M.A. & Pastor,
O. (Eds.), In Proceedings ER’2003 Workshops,
LNCS 2814, (pp. 52-64). Heidelberg, Germany:
Springer-Verlag.
1009
Payne, T., & Lassila, O. (2004). Semantic Web Szyperski, C. (2002). Component Software: Be-
Services. IEEE Intelligent Systems, 19(4), 14–15. yond Object-Oriented Programming – 2nd Ed.
doi:10.1109/MIS.2004.29 New York, USA: Addison-Wesley.
Peltier, M., Bezivin, J., & Guillaume, G. (2001). Widom, J. (1995). Research problems in data
MTRANS: A general framework, based on warehousing. In Proceedings of 4th International
XSLT, for model transformations. In Proceed- Conference on Information and Knowledge Man-
ings of the Workshop on Transformations in UML agement, (pp. 25-30).
WTUML’01. Retrieved 21 July 2008 from: http://
Wiederhold, G. (1992). Mediators in the archi-
citeseer.ist.psu.edu/581336.html
tecture of future information systems. IEEE
Reynaud, C., Sirot, J. P., & Vodislav, D. (2001). Computer, 25, 38–49.
Semantic Integration of XML Heterogeneous
Zhang, Z., & Yang, H. (2004). Incubating Services
Data Sources. In Proceedings IDEAS Conference
in Legacy Systems for Architectural Migration.
2001, (pp. 199–208).
In Proceedings 11th Asia-Pacific Software Engi-
Rosenberg, F., & Dustdar, S. (2005). Business neering Conference APSEC’04, (pp. 196-203).
Rules Integration in BPEL - A Service-Oriented
Zhu, F., Turner, M., Kotsiopoulos, I., Bennett, K.,
Approach. In Proceedings 7th International
Russell, M., Budgen, D., et al. (2004). Dynamic
IEEE Conference on E-Commerce Technology,
Data Integration Using Web Services. In Pro-
(pp. 476- 479).
ceedings 2nd International Conference on Web
Rouvellou, I., Degenaro, L., Rasmus, K., Eh- Services ICWS’2004, (pp. 262-269).
nebuske, D., & McKee, B. (2000). Extending busi-
Zhu, Y. (2007). Declarative Rule-based Integra-
ness objects with business rules. In Proceedings
tion and Mediation for XML Data in Web Service-
33rd Intl. Conference on Technology of Object-
based Software Architectures. M.Sc. Thesis.
Oriented Languages, (pp. 238-249).
Dublin City University.
Schaffert, S. (2004). Xcerpt: A Rule-Based Query
and Transformation Language for the Web. PhD
Thesis, University of Munich.
ADDItIONAL rEADING
Seltsikas, P., & Currie, W. L. (2002). Evaluating
the application service provider (ASP) business textbooks
model: the challenge of integration. In Proceed-
ings 35th Annual Hawaii International Conference Bass, L. Clements, & P. Kazman, R. (2003).
2002. 2801 – 2809. Software Architecture in Practice. 2nd Edition.
Boston, USA: Addison-Wesley.
Stal, M. (2002). Web Services: Beyond Compo-
nent-based Computing. Communications of the Krafzig, D., Banke, K., & Slama, D. (2004).
ACM, 45(10), 71–76. doi:10.1145/570907.570934 Enterprise SOA: Service-Oriented Architecture
Best Practices. Upper Saddle River, USA: Pren-
Stern, A., & Davis, J. (2004). Extending the Web
tice Hall.
services model to IT services. In Proceedings
IEEE International Conference on Web Services, Mahmoud, Q. H. (2004). Middleware for Com-
(pp. 824-825). munications: Concepts, Designs and Case Studies.
Indianapolis, USA: John Wiley and Sons.
1010
Articles Li, S.-H., Huang, S.-M., Yen, D. C., & Chang,

C.-C. (2007). Migrating Legacy Information
Abiteboul, S., Benjelloun, O., & Milo, T. (2002). Systems to Web Services Architecture. Journal
Web services and data integration. In Proceedings of Database Management, 18(4), 1–25.
of the Third International Conference on Web
Milanovic, N., & Malek, M. (2004). Current
Information Systems Engineering. (pp. 3-6).
solutions for Web service composition. IEEE
Bengtsson, P., Lassing, N., Bosch, J., & Vliet, H. Internet Computing, 8(6), 51–59. doi:10.1109/
(2004). Architecture-Level Modifiability Analysis MIC.2004.58
(ALMA). Journal of Systems and Software, 69(1),
Milo, T., & Zohar, S. (1998). Using Schema Match-
129–147. doi:10.1016/S0164-1212(03)00080-3
ing to simplify heterogeneous Data Translation.
Bing, Q., Hongji, Y., Chu, W. C., & Xu, B. (2003): Proceeding of the Int’l VLDB Conference. (pp.
Bridging legacy systems to model driven archi- 122-133).
tecture. In Proceedings 27th Annual International
Oquendo, F. (2006). π-Method: a model-driven
Computer Software and Applications Conference
formal method for architecture-centric software
COMPSAC 2003. (pp. 304- 309).
engineering. SIGSOFT Software Engineering
Bolzer, M. (2005). Towards Data-Integration on Notes, 31(3), 1–13. doi:10.1145/1127878.1127885
the Semantic Web: Querying RDF with Xcerpt.
Pahl, C. (2007). An Ontology for Software
Master Thesis. University of Munich.
Component Description and Matching. Interna-
Calvanese, D., Giacomo, G., Lenzerini, M., & tional Journal on Software Tools for Technology
Nardi, D. (2001). Data Integration in Data Ware- Transfer, 9(2), 169–178. doi:10.1007/s10009-
housing. International Journal of Cooperative In- 006-0015-9
formation Systems., 10(3), 237–271. doi:10.1142/
Pahl, C. (2007). Semantic Model-Driven Ar-
S0218843001000345
chitecting of Service-based Software Systems.
Djuric, D. (2004). MDA-based Ontology Infra- Information and Software Technology, 49(8),
structure. Computer Science and Information Sys- 838–850. doi:10.1016/j.infsof.2006.09.007
tems, 1(1), 91–116. doi:10.2298/CSIS0401091D
Rahm, E., & Bernstein, A. (2001). A Survey of
Hasselbring, W. (2000). Information System Approaches to Automatic Schema Matching. The
Integration. Communications of the ACM, 43(6), VLDB Journal, 10(4), 334–350. doi:10.1007/
32–36. doi:10.1145/336460.336472 s007780100057
Hasselbring, W. (2002). Web data integration for Selic, B. (2003). The Pragmatics of Model-Driven
e-commerce applications. IEEE MultiMedia, 9(1), Development. IEEE Software, 20(5), 19–25.
16–25. doi:10.1109/93.978351 doi:10.1109/MS.2003.1231146
Lehti, P., & Fankhauser, P. (2004). XML data in- Sheth, A. P., & Larson, A. (1990). Feder-
tegration with OWL: experiences and challenges. ated database systems for managing distrib-
In Proceedings 2004 International Symposium uted, heterogeneous, and autonomous databases.
on Applications and the Internet. (pp. 160-167). ACM Computing Surveys, 22(3), 183–236.
doi:10.1145/96602.96604
Levy, A. (1998). The information manifold
approach to data integration. IEEE Intelligent
Systems, 13, 12–16.
1011
Velegrakis, Y., Miller, R., & Mylopoulos, J. (2005). Yang, Y., Peng, X., & Zhao, W. (2007). An Au-
Representing and Querying Data Transformations. tomatic Connector Generation Method for Dy-
In Proceedings of the 21st International Confer- namic Architecture. In Proceedings International
ence on Data Engineering ICDE’05. (pp. 81-92). Computer Software and Applications Conference
COMPSAC 2007. (pp. 409-414).
Willcocks, P., & Lacify, C. (1998). The sourc-
ing and outsourcing of IS: Shock of the New? P.
Willcocks and C. Lacity (Eds.), Strategic Sourc- standards
ing of Information Technology: Perspective and
Practices. Chichester, UK: Wiley. Object Management Group. (2003). Model-
Driven Architecture MDA Guide V1.0.1. OMG.
Management and Best Processes, edited by Patrick Hung, pp. 98-113, copyright 2009 by Business Science Reference
1012
1013
Chapter 4.8
Data Warehouse and Business
Intelligence Systems in
the Context of E-HRM
Martin Burgard
Saarland University, Germany
Franca Piazza
Saarland University, Germany
INtrODUctION functionality of these systems and to illustrate

the application possibilities and benefits of these
The increased use of information technology leads systems by selected application examples in the
to the generation of huge amounts of data which context of e-HRM.
have to be stored and analyzed by appropriate
systems. Data warehouse systems allow the stor-
age of these data in a special multidimensional bAcKGrOUND
data base. Based on a data warehouse, business
intelligence systems provide different analysis In the past the importance of data warehouse and
methods such as online analytical processing business intelligence systems has continuously
(OLAP) and data mining to analyze these data. increased and the rate of companies using a data
Although these systems are already widely used warehouse and/or a business intelligence system is
and the usage is still growing, their application in rather high (e.g., Watson, Annino, Wixom, Avery,
the area of electronic human resource management & Rutherford, 2001). An increasing number of
(e-HRM) is rather scarce. Therefore, the objective case study publications (e.g., Marks & Frolick,
of this chapter is to depict the components and 2001; Watson, Wixom, Hoffer, Anderson-Lehman
& Reynolds, 2006) and general literature for prac-
DOI: 10.4018/978-1-59904-883-3.ch034 titioners (e.g., Humphries, Hawkins, & Dy, 1999)
Data Warehouse and Business Intelligence Systems in the Context of E-HRM
are further indicators showing the ever-growing The data warehouse stores the data in multidi-
importance of these systems. On the other hand, mensional data structures, so-called cubes, in
publications concerning these system categories order to provide optimized analysis possibilities
in the context of e-HRM, except for short discus- (e.g., Inmon, 2005). The administration system
sions to some isolated topics as online recruiting supports the data modeling, the ETL-scheduling,
(Lin & Stasinskaya, 2002), enterprise resource the user administration, and the monitoring. As
planning (Ashbaugh & Miranda, 2002), or human some data become obsolete and do not have to be
resource information systems (Kovach, Hughes, accessible for the ongoing management analysis
Fagan, & Magitti, 2002) are rather scarce. Data they can be stored in the archiving system. Fur-
warehouse and business intelligence systems are thermore the archiving system allows backup of
commonly used in sales or marketing depart- the data. To manage the huge amount of data
ments. In contrast, their use in HR departments stored in a data warehouse, information about the
is relatively low (Watson et al., 2001). However, data, so-called metadata, such as calculation rules,
the adoption of these systems in the context of e- content description, or usage documentation, is
HRM offers new potentials to the management of necessary. The depicted components constitute
human resources. In the following their technical the data warehouse system which is the basis for
and functional aspects are depicted. analysis activities performed by using a business
A data warehouse is defined as a “subject- intelligence system.
oriented, integrated, non-volatile and time-variant Business intelligence systems subsume differ-
collection of data in support of management’s ent technologies and methods to access and ana-
decisions” (Inmon, 2005, p. 29). So the main task lyze the data stored in a data warehouse (Turban,
of the data warehouse is to integrate the data from Aronson, & Liang, 2005). The core components
a variety of different source systems existing inside of business intelligence systems are OLAP and
and outside a company in a single data base and data mining.
to store the data in a multidimensional structure
which is optimized to support the management’s
analysis activities. In doing so the operative sys- Figure 1. Reference architecture of data warehouse
tems are no longer charged with the reporting and business intelligence systems
requests of the management which resulted in
poor system performance.
The data warehouse is the core component of
the data warehouse system which further consists
of several components (see Figure 1): the extrac-
tion, transformation, and loading system (ETL-
system), the administration system, the archiving
system, and the metadata repository.
To integrate the data in the data warehouse the
ETL-system enables the extraction of data from
different source systems. Furthermore the ETL-
system transforms the data to eliminate syntactic
and semantic defects and harmonizes the structure
and value of the data. After the transformation
the relevant data is loaded into the data warehouse.
1014
Although there is no explicit definition of Besides OLAP, data mining constitutes another
OLAP in literature the common understanding core component of business intelligence systems.
of OLAP refers to the possibilities to consolidate, Data mining aims at revealing unknown patterns
view, and analyze data according to multiple di- from large databases (Fayyad, Piatetsky-Shapiro,
mensions (Codd, Codd, & Salley, 1993). The fast & Smyth, 1996). A variety of methods from dif-
analysis of shared multidimensional information ferent areas such as statistics, machine learning,
(FASMI) concept characterizes OLAP by means and artificial intelligence are subsumed under the
of five attributes: fast, analysis, shared, multidi- concept of data mining. All of these methods aim
mensional, and information (Pendse & Creeth, at generating hypotheses out of the data without
1995). The attribute fast refers to the requirement any precedent assumption of coherences in the
that the data request has to be fulfilled quickly and data. This constitutes the main difference to OLAP.
the user performing the analysis does not have to OLAP is based on the user’s ability to overlook
wait long for a system response. Following the the data which describe a specific situation. But
FASMI concept an average timeframe of five especially, in very complex decision situations,
seconds characterizes the response time of OLAP- where the coherences of data are unknown and
requests. Analysis, as a further attribute of OLAP, hence can not be analyzed by user driven inter-
denotes the ability to cope with business logic and active queries, data mining can provide further
to facilitate the user’s analysis activities. The data information. The most important functions of
can be analyzed interactively by the user without data mining are classification, segmentation, and
any programming knowledge. The attribute shared association analysis.
refers to the need to implement the necessary Classification consists of examining the stored
security and integrity requirements to be fulfilled data and assigning them to one of a predefined set
by OLAP. Customarily several users try to analyze of classes. Based on preclassified data, a model
the same data at the same time. Hence an appro- is automatically generated. This model allows
priate update locking and authorization concept assigning of unclassified data to a certain class.
has to be assured. Furthermore, multidimension- Applied methods are, for example, rule induction,
ality constitutes a characteristic of OLAP. OLAP discriminant analysis, or multilayer perception
provides a multidimensional conceptual view of (Berry & Linoff, 2004; Cho & Ngai, 2003).
the data which means that relevant measures can Segmentation is the task of grouping hetero-
be analyzed by multiple different dimensions. geneous data into more homogenous subgroups,
In addition, OLAP provides the possibilities to so-called clusters. In contrast to classification,
analyze the data on different hierarchical levels. segmentation does not rely on predefined classes
Measures such as headcount, for instance, can be but rather generates clusters during the segmenta-
analyzed on a detailed disaggregated level (e.g., tion process. It is up to the user to determine the
headcount of a certain department) and on an meaning of the resulting clusters. Applied segmen-
aggregated level (e.g., headcount of the whole tation methods are, for example, cluster analysis
company). The operations for such an interactive or self-organizing maps (Berry & Linoff, 2004).
hierarchical analysis are called drill-down and The association analysis aims at finding fre-
roll-up. Finally information as the fifth attribute quently appearing combinations of data. It can
refers to the allocation of the requested data in a be distinguished into two types. While the static
transparent way. The FASMI-concept integrates association analysis identifies simultaneously
especially the user requirements and delivers an existent combinations, the sequential association
appropriate characterization of OLAP. analysis includes the temporal aspect and discov-
1015
ers the successive appearance of data (Agrawal, more importance for the HR management (e.g.,
Imelienski, & Swami, 1993; Agrawal & Srikant, Srinivasa & Saurabh, 2001). In the following an
1995). exemplary application of an HR relevant analysis
using OLAP is depicted.
For the HR management a detailed report on
tHE ADOPtION OF DAtA the number of sick persons for instance might
WArEHOUsE AND bUsINEss be important. The relevant personnel data are
INtELLIGENcE systEMs IN stored in a cube (see Figure 2) containing the
tHE cONtEXt OF E-HrM dimensions division, region, and time, and the
measure number of sick persons. As OLAP is not
Although the use of data warehouse and business limited to three dimensions, further dimensions,
intelligence systems in the context of e-HRM such as line manager, qualification, and so forth,
is rather scarce, usage offers a high potential to are conceivable. According to the depicted char-
support and improve HRM. These systems can be acteristics of OLAP an HR manager can navigate
adopted in all business areas where comprehen- by using different operations; for instance, with
sive, sophisticated, and data intensive analyses the operation slicing it is possible to filter the
are required to support decisions. Therefore, number of sick persons in 2006 for all divisions
their adoption is appropriate for all HR tasks in all regions. The operation dicing enables one
such as recruitment, appraisal, development, and to see the number of sick persons for 2004 and
compensation. 2005 for the finance and production divisions for
Following Figure 1, a data warehouse contain- America and Europe, for example.
ing only personnel relevant data can be referred Furthermore, OLAP supports a hierarchical
to as HR data warehouse. Thereby, the main analysis. In this context it might be of interest to
reason to adopt a data warehouse system in the analyze the number of sick persons of the finance
context of e-HRM is the lack of consistency of division in America in 2006 in more detail. Using
HR information available in existing separate HR drill-down, a detailed report on the number of
information systems and hence the impossibility sick persons for every quarter in 2006 can be
of appropriate analysis and reporting (Van Wessel, generated (see Figure 3). Furthermore drill-down
Riebers, & de Vries, 2006). This inconsistency is is not restricted to one level so HR managers can
caused by a fragmented system-landscape sup- also navigate to months, weeks, and days to receive
porting the HR tasks. HR applications often have more specific information. In return the roll-up
been developed in different functional areas (e.g., operation allows for the analyzing of data on a
payroll, benefits, staffing, time and attendance) or more aggregated level.
geographic regions (e.g., North America, Europe, These examples illustrate the analysis possi-
Asia Pacific). An HR data warehouse allows the bilities of OLAP which can be employed in any
integration of relevant personnel data from the HR task such as recruitment, appraisal, compen-
internal and external HR data sources and enables sation, and development. OLAP is based on the
comprehensive analysis opportunities by employ- user’s ability to survey the data and to navigate
ing business intelligence systems. In consideration correspondently. While a HR data warehouse
of the proceeding outsourcing of HR activities stores HR data and the actual coherences of the
and the loss of personal relationships between data are not always obvious, data mining extends
the employee and the HR department (Lepak & the analysis possibilities to the context of complex,
Snell, 1998), the ability to generate information data-rich situations. Data mining reveals unknown
about the employees out of HR data will gain patterns out of the data which can be used to sup-
1016
Figure 2. Illustration of an OLAP cube and examples for slicing and dicing
port HR relevant decisions. In the following ex- automatically by the classification algorithm and
emplary applications of data mining in selected is based on the stored attributes describing an
HR contexts are depicted. applicant. The unknown pattern revealed by data
The data mining function classification can mining consists on the relationship between ap-
be used to support and accelerate the applicant plication data and productivity. This will enable
selection process for example. Using the clas- an automatic preselection of potential employees
sification methods applicants can be allocated to and hence support the HR manager to select em-
predefined classes indicating the productivity of ployees with assumable high productivity. A lot
an employee. The classes’ high, medium, and low of further application possibilities are conceivable
sales premiums might serve as indicators show- such as the analysis of the termination behavior
ing the productivity of an employee, for example of employees. It might be of interest to identify
(Cho & Ngai, 2003). The allocation is performed the employees, who tend to sign off, and develop
Figure 3. Example for the OLAP operations drill-down and roll-up
1017
corresponding retention measures. Based on his- can be compared to the career planning and lead
torical data concerning terminations, classification to adjustments if significant differences appear.
methods provide the possibility to automatically The depiction of the selected possibilities to
classify all employees into the classes of poten- adopt data mining in the context of e-HRM could
tially terminating and nonresigning employees. reveal that the main potentials and benefits depend
This information can be used by the HR depart- on the decision support in complex and data-rich
ment to develop target-oriented retention measures decision situations to enable the creation of more
(Min & Emam, 2003). target-oriented personnel measures.
Segmentation, as a further function of data
mining, generates homogeneous clusters of em-
ployees which can be used, for example, to create FUtUrE trENDs
a personnel portfolio. Hence, the segmentation
reveals additional information about the employ- Considering the growing amount of HR data and
ees concerning their affiliation to a certain cluster. the corresponding necessity to store and analyze
So it might be conceivable to generate clusters these data, the application of data warehouse
based on the attributes qualification, gender, and business intelligence systems enable the HR
age, and position of an employee which leads to management to handle these data and perform
homogeneous clusters of employees concerning comprehensive HR planning and controlling.
the named attributes such as the cluster of the old, These systems will develop standard analytical
high qualified managers and the cluster of the systems in all business areas, as well in the HR
young, high qualified specialists. Based on this area. In this article some application possibilities
information the development of cluster-specific and their benefits were introduced, nevertheless
training measures, for instance, is possible. research is necessary to systematically evaluate
Finally, the association function can deliver the application of these systems in the context
useful information for the HR department. The of e-HRM. Research should cover recruitment,
association analysis reveals frequently appearing compensation, appraisal, and development as
combinations of data. Based on the personnel data core tasks of HRM and develop reference data
stored in a HR data warehouse the association models and application scenarios to improve the
analysis might reveal, for example, frequently decision-making process of HR managers. These
appearing combinations of a particular merit- systems are very complex and cost-intensive,
appraisal and a certain supervisor. This information hence their implementation and employment do
can be used for further examinations concerning not automatically lead to benefits for the HRM.
the bias of this supervisor. Furthermore, the in- Reference data models and application scenarios
tegration of the temporal aspect into the analysis facilitate the use of these systems and enable the
can be reached by applying the sequential asso- exhaustion of the potential benefits introduced
ciation analysis. So it might be of interest to the in this article.
HR manager to reveal which successive steps are
actually necessary to reach a certain position in
the company. The result of the sequential analysis cONcLUsION
is a frequently used career path to reach this po-
sition (e.g., the employee attends a management New technologies such as data warehouse and
seminar then receives a high merit-rating, then business intelligence systems enable consolidated
works in a foreign country, and finally gets the storage and innovative analysis of personnel
promotion). These revealed existing career paths data. In this article the technical and functional
1018
aspects of data warehouse and business intel- Cho, V., & Ngai, E. (2003). Data mining for
ligence systems were introduced. Based on the selection of insurance sales agents. Expert Sys-
general functionalities of these systems their tems: International Journal of Knowledge Engi-
application and possible benefits in the context neering and Neural Networks, 20(3), 123–132.
of e-HRM were depicted. It could be shown that doi:10.1111/1468-0394.00235
both systems are applicable in e-HRM. The HR
Codd, E.-F., Codd, S.-B., & Salley, C.-T. (1993).
data warehouse system serves as an integrated data
Providing OLAP to user analysts: An IT-mandate.
storage which contains relevant personnel data
White paper Codd associates. Retrieved Septem-
from multiple widespread HR-applications exist-
ber 6, 2006, from http://www.dev.hyperion.com/
ing in a company. The analysis methods provided
resource_library/white_papers/providing_olap_
by the business intelligence system extend the
to_user_ analysts.pdf
customarily applied methods of HR planning and
controlling. Thereby OLAP allows the interactive Fayyad, U. M., Piatetsky-Shapiro, G., & Smyth, P.
multidimensional analysis of personnel data and (1996). From data mining to knowledge discovery:
data mining reveals hidden coherences in the data an overview. In U. Fayyad, G. Piatetsky-Shapiro,
which can be used to optimize decision-making P. Smyth, & R. Uthurusamy (Eds.), Advances in
processes of HR managers. knowledge discovery and data mining (pp. 1-34).
Camebridge MA: The MIT Press.
Humphries, M., Hawkins, M., & Dy, M. (1999).
rEFErENcEs
Data warehousing: Architecture and implementa-
Agrawal, R., Imelienski, T., & Swami, A. (1993). tion. Upper Saddle River, NJ: Prentice Hall.
Mining association rules between sets of items in Inmon, W.-H. (2005). Building the data ware-
large databases. In P. Buneman & S. Jajodia (Eds.), house. Indianapolis: Wiley Publishing.
Proceedings of the ACM SIGMOD Conference on
Management of Data (pp. 207-216). Washington, Kovach, K., Hughes, A., Fagan, P., & Maggitti, P.
D.C.: ACM Press. (2002). Administrative and strategic advantages
of HRIS. Employment Relations Today, 29(2),
Agrawal, R., & Srikant, R. (1995). Mining se- 43–48. doi:10.1002/ert.10039
quential patterns. In P. S. Yu & A. Chen (Eds.),
Proceedings of the International Conference on Lepak, D., & Snell, S. A. (1998). Virtual HR: Stra-
Very Large Data Bases (VLDB) (pp. 3-14). Taipei, tegic human resource management in the 21st cen-
Taiwan: IEEE Computer Society. tury. Human Resource Management Review, 8(3),
215–234. doi:10.1016/S1053-4822(98)90003-1
Ashbaugh, S., & Miranda, R. (2002). Technology
for human resources management: Seven ques- Lin, B., & Stasinskaya, V. (2002). Data ware-
tions and answers. Public Personnel Management, housing management issues in online recruiting.
31(1), 7–20. Human Systems Management, 21(1), 1–8.
Berry, M., & Linoff, G. S. (2004). Data mining Marks, W., & Frolick, M. (2001). Building cus-
techniques for marketing, sales and customer tomer data warehouses for a marketing and service
relationship. Indianapolis: Wiley. environment: A case-study. Information Systems
Management, 18(3), 51–56. doi:10.1201/1078/4
3196.18.3.20010601/31290.7
1019
Min, H., & Emam, A. (2003). Developing the KEy tErMs AND DEFINItIONs
profiles of truck drivers for their successful re-
cruitment and retention: A data mining approach. Business Intelligence System: Subsumes dif-
International Journal of Physical Distribution ferent technologies and methods such as OLAP
& Logistics Management, 33(2), 149–162. and data mining to analyze the data stored in a
doi:10.1108/09600030310469153 data warehouse.
Data Mining: Subsumes a variety of methods
Pendse, N., & Creeth, R. (1995). The OLAP report. to extract unknown patterns out of a large amount
Retrieved September 6, 2006, from http://www. of data. Data mining methods originate from the
olapreport.com area of machine learning, statistics, and artificial
Srinivasa, R., & Saurabh, S. (2001). Business intelligence. The main tasks of data mining are
intelligence and logistics (white paper). Retrieved classification, segmentation, and association
September 18, 2006, from http://www.dmreview. analysis.
com/whitepaper/wid328.pdf Data Warehouse: A subject-oriented, inte-
grated, time-variant, and nonvolatile collection
Turban, E., Aronson, J., & Liang, P. (2005). Deci- of data. The data are usually stored in multidi-
sion support systems and intelligent systems (7th mensional cubes, an optimized way to provide
ed.). Saddle River, NJ: Prentice Hall. data for analyze purposes.
Van Wessel, R., Ribbers, P., & de Vries, H. (2006). Data Warehouse System: Consists of several
Effects on IS standardization on business process components. The core component is a data ware-
performance: A case in HR IS company standard- house as a database and further components are
ization. In R. H. Spraque (Ed.), Proceedings on the ETL-system, administration system, archiving
39th Hawaii International Conference on System system, and metadata repository.
Sciences. Los Alamitos, CA: IEEE Computer HR Data Warehouse: A specific data ware-
Society Press. house for HR data. The company’s HR relevant
data is collected, integrated, and stored with the
Watson, H. J., Annino, D. A., Wixom, B. H., aim of producing accurate and timely HR manage-
Avery, K. L., & Rutherford, M. (2001). Current ment information and supporting data analysis.
practices in data warehousing. Information Sys- OLAP Operations: To analyze the data in
tems Management, 18(1), 47–56. doi:10.1201/1 OLAP cubes navigation and hierarchical analysis
078/43194.18.1.20010101/31264.6 are differentiated. Navigation operations are, for
Watson, H.-J., Wixom, B. H., Hoffer, J. A., An- instance, slicing and dicing. Roll-up and drill-
derson-Lehman, R., & Reynolds, A. M. (2006). down support the hierarchical analysis. Thereby,
Real-time business intelligence: Best practices at the roll-up operation allows for analyzing data
continental airlines. Information Systems Manage- on a more aggregated level and drill-down is the
ment, 23(1), 7–18. doi:10.1201/1078.10580530/ reverse operation.
45769.23.1.20061201/91768.2 Online Analytical Processing (OLAP): Re-
fers to the possibilities to consolidate, view, and
analyze data according to multiple dimensions.
The fast analysis of shared multidimensional in-
formation (FASMI) concept characterizes OLAP
by means of five attributes.
This work was previously published in Encyclopedia of Human Resources Information Systems: Challenges in e-HRM,
edited by Teresa Torres-Coronas and Mario Arias-Oliva, pp. 223-229, copyright 2009 by Information Science Reference
1020
1021
Chapter 4.9
Implementation of ERP in
Human Resource Management
Zhang Li
Harbin Institute of Technology, China
Wang Dan
Chang Lei
INtrODUctION ERP by 2000 (Kumar, & Hillegersberg, 2000; Siau,

2004), and it was projected that organizations’ total
In 1999, Peter Drucker said: “A new Information spending on ERP adoptions was an estimated $72.63
Revolution is well under way. It is not a revolution billion in 2002 (Al-Marshari, 2002).
in technology, machinery, techniques, software or Many scholars have recognized the importance
speed. It is a revolution in concepts.” As a result of of people in organizations, and this viewpoint is
information technology (IT) innovation and reorga- the central focus of the human resource manage-
nization, enterprise resource planning (ERP) was ment (HRM) perspective (Pfeffer, 1995). In this
proposed by the Gartner Group in the early 1990s. perspective, HRM has the potential to be one of
It is a successor to manufacturing resource planning the key components of overall enterprise strategy.
(MRP II) and attempts to unify all departmental Additionally, HRM may provide significant com-
systems together into a single, integrated software petitive advantage opportunities when they are used
program that runs off a single database so that the to create a unique (i.e., difficult to imitate) organi-
various departments can more easily share infor- zational culture that institutionalizes organizational
mation and communicate with each other (Koch, competencies throughout the organization (Bowen
2002). Over 60% of the U.S Fortune 500 had adopted & Ostroff, 2004).
Typically, an ERP system supports HRM, opera-
DOI: 10.4018/978-1-60566-026-4.ch292 tion and logistics, finance, and sales and marketing
Implementation of ERP in Human Resource Management
functions (Davenport, 1998) (see Figure 1). But the suggests emerging research opportunities within
early development stage of ERP in enterprises was the domain of the topic.
all along with the center of production and sales
course. Until recently, research has empirically
supported the positive relationship between cor- bAcKGrOUND
porate financial performance and HRM function,
and managers have also realized that HRM can ERP, a term coined by the Gartner Group, is not
deliver organizational excellence and competitive simply a tool that provides singular outputs, but
advantage for enterprises (Boudreau & Ramstad, rather an infrastructure that supports the capa-
1997; Huselid, 1995; Wright, McMahan, Snell, bilities of all other IT-based tools and processes
& Gerhart, 2001). The HRM module was intro- utilized by a firm (Enslow, 1996). Shang and Sed-
duced into ERP, forming a highly integrated and don (2000) classified the different types of ERP
efficient resource system with the other function benefits as: IT infrastructure benefits, operational
modules of ERP. However, there are still many benefits, managerial benefits, strategic benefits,
HRM-related problems that may result in the and organizational benefits. Palaniswamy (2002)
failure of ERP projects arising. So, there have pointed out that the failures of ERP projects were
been regular appeals to scholars for more research not because the software were coded incorrectly,
about the implementation of ERP systems in the rather the companies failed to understand the real
HRM perspective in the last few years (Barrett organizational needs and systems required to solve
& Mayson, 2006). their problems to improve performance. Lynne,
This article introduces the functions of an HRM Axline, Petrie, and Cornelis (2000) analyzed the
module in ERP systems from the fields of human adopters’ problems with ERP including project
resource planning, recruitment management, phase problems, problems with product and
training management, time management, perfor- implementation consultants, shakedown phase
mance management, compensation management, problems, underestimating data quality problems
and business trip arrangement. Then it analyzes and reporting needs, and so on.
five HRM-related problems that may block the Within the managerial literatures, a coherent
enterprises from implementing ERP successfully, approach provides a conceptual basis for as-
and it provides reasonable recommendations. serting that human resource is a key source of
Finally, the article discusses future trends and competitive advantages, since it offers a unique
contribution to value creation, rarity, imperfect
imitability, and non-substitutability of a firm’s
Figure 1. Function modules of an ERP system
strategic resources (Bellini & Canonico, 2007).
Stone (2007) considered the past, present, and
future of HRM theory and research. He concluded
that HRM theory and research has considerable
potential to enhance organizational efficiency
and effectiveness. Ashbaugh and Rowan (2002)
summarized the technology features of a modern
HRM system (see Table 1).
In addition, some scholars have already studied
the relationship or connection of ERP implemen-
tation with HRM. For instance, Ashbaugh and
Rowan (2002) argued that the major difference
1022
Table 1. Technology features of modern HRM

ment, time management, performance manage-
system
ment, compensation management, and business
Integration User friendliness trip arrangement (Ahmad & Schroeder, 2003;
Common relational database Enhanced reporting and Li, 2001; Stone, 2007). Data from all function
Flexible and scalable technol- analysis
ogy Process standardization and
systems will be collected into a central database,
Audit trail and drill down malleability and the database can further supply data needed
capabilities Internet and capabilities
for all function systems by integration.
Robust security Document management and
Workflow imaging
Human Resource Planning
between ERP and its predecessors (e.g., MRP II) Based on the requirements of enterprises, manag-
is the linkage of financial and HRM applications ers can use the HRM module of an ERP system to
through a single database in a software application establish human resource planning conveniently.
that is both rigid and flexible. Wright and Wright The ERP system assists the decision making of
(2002) listed two of the most-cited HRM risks in managers by simulating the performance of hu-
an ERP system: lack of user involvement and inad- man resource planning and comparing the data.
equate training. Hsu, Sylvestre, and Sayed (2006) Additionally, the ERP system is also able to analyze
supplied another often-overlooked HRM factor or forecast the human resource planning costs by
when implementing an ERP system—that is, the integrating relevant information.
result of high stress levels on the staff, particularly
in the finance or accounting departments, which Recruitment Management
are already under stress from the heavy workload
in a legacy system. Li (2001) studied the HRM Recruitment should be taken as a significant
function module in an ERP system. He insisted investment because human resources are the
that the practical HRM system should be built up foundational assets of an enterprise. To keep
to improve incentive mechanism and to strengthen advantages in competition, the human resources
the training of employees while applying ERP. department must have a reasonable recruitment
system to select talents for enterprise. The ERP
system can support recruitment management in
IMPLEMENtAtION OF ErP three ways. First, it optimizes the recruitment
rELAtED tO HrM process to reduce the workload. Second, it offers
scientific management to recruitment costs. Third,
Functions of the HrM it provides useful information for the decision
Module in ErP system making on recruitment management.
We have studied the necessity and essentiality of Training Management

HRM to the implementation of ERP in the preced-
ing part of this article. The adoption of ERP also Training in the use of multiple skills including
greatly impacts HRM by extending its functions to process improvement skills, which can provide
the all-direction management category (see Figure long-term work-life security rather than job secu-
2). The functions of HRM have developed from rity (Schonberger, 1994). The implementation of
simple compensation calculating and personnel ERP can help train employees to acquire techni-
management to the fields of human resource plan- cal, interpersonal, and business skills required to
ning, recruitment management, training manage- become fully participating team members in the
1023
Figure 2. Traditional and extended functions of HRM
early stage of team development (Pasmore & Performance Management

Mlot, 1994). In other team development stages,
by giving support to the human resource depart- Performance evaluation might consider the fol-
ment to make an appropriate training plan, the lowing issues: How are the facilitative and op-
ERP system can also help train team members to erational activities allocated to individuals in an
accept new skills, improved management regula- organization, and how does the facilitative content
tions, and so on. of a task vary in an organization (Nilakant, 1994).
The human resource department can establish an
Time Management evaluation index system according to these issues.
By integrating the performance management
Time management may support the planning, system with the time management system, the
controlling, and management processes of HRM. It ERP system will record data in a central database
means to arrange the time table for the enterprises and keep relevant data timely for each evaluation
and staff flexibly according to the local calendar. index. These data will be useful to the decision
The ERP system can record the attendance rate and making of managers on corporate strategy, too.
other relevant information by using a Telematics
Control Unit (TCU). For example, data related to
the compensation will be further processed in the
compensation management system.
1024
Compensation Management may block the success of ERP projects in enter-

prises (Lynne et al., 2000; Wright & Wright, 2002;
A reasonable compensation system should be able Hsu et al., 2006; Sun, & Xu, 2006).
to apply proper calculation methods in terms of First is the shortage of professionals, especially
different regions, departments, positions, and so the inter-disciplinary talents who are expert at both
forth. The implementation of ERP will achieve IT and management. ERP is not just an advanced
this objective by integrating the compensation technique, it is also an advanced management
management system with other systems (e.g., concept. Therefore, the inter-disciplinary profes-
timing management system, performance manage- sionals are crucial to the success of ERP. This
ment system) so that it can update relevant data problem is especially serious in the small and
in a timely fashion so as to establish a dynamic medium-sized enterprises because of the weakness
compensation calculation system. The human of strength and management level.
resource department can simulate the performance Second is the deficient talent introduction
of the calculation system to forecast compensation mechanism. Employees who take charge of the
information needed and to adjust the structure of ERP project of an enterprise are under high pres-
the compensation management system. This is sure because they take great responsibilities for
an excellent improvement because it decreases the success of the implementation. If an enterprise
management costs as well as problems caused does not have an effective talent introduction
by the intervention of manpower. Compensation mechanism to attract talented people who are
management also includes other functions such needed, it will be hard to start the ERP project
as salary payments, loans for staff, and so forth. at all, not to mention the successful application.
Third is the insufficient education and train-
Business Trip Arrangement ing for employees. The cultivation of talents is a
process that needs much time investment as well
A business trip arrangement system can control as money, while some enterprises do not want
the whole flow of a business trip from application to pay much investment to the education and
to ratification and reimbursement. These data will training of employees due to lack of an in-depth
be further processed in other function modules understanding of ERP.
of ERP (e.g., finance module) through systems Fourth is the poor incentive mechanism for
integration. employees. Above all, the compensation mecha-
nisms of enterprises may not be attractive enough.
Main HrM-related Problems of ErP A competitive compensation mechanism cannot
Implementation in Enterprises only attract applicants, but it also prevents the job
hopping of employees. Still, the ERP projects have
ERP has been broadly applied in enterprises for not been supported enough by superior administra-
nearly 20 years because of the enormous potential tion departments. For example, the application of
economic benefits. However, 53% of the ERP ERP needs the support of all involved departments,
projects in U.S. firms were failed by 1996, and the while managers hesitate to place departmental
success rate of ERP projects in Chinese enterprises backbones on the ERP implementation. It may also
was only less than 20% by 2002 (Edwards, 1999; influence the working enthusiasm of employees
Yang & Zhao, 2003). Hsu et al. (2006) pointed and kill more innovations if superiors interfere
out that HRM factors played a significant role in too much with the implementation.
almost all failed information systems. This article Fifth is the lack of exterior consultation or a
analyzes five main HRM-related problems that supervision system. An enterprise must pass the
1025
scientific verification of experts if it wants to • Enterprises should place great emphasis

adopt ERP. The implementation must be carried on personnel education and training. The
out under the direction of exterior professional human resource department can carry an
organizations and the supervision of a special ideological education that involves both se-
system. Enterprises may neglect the necessity nior managers and common staff to arouse
and importance of these functions. their concerns about ERP implementation.
The frequency and quality of personnel
training should be strengthened as well.
rEcOMMENDAtIONs • Enterprises should enhance exterior con-
sultation as well as the supervision func-
To implement ERP successfully, enterprises must tion for the adoption of ERP. It is a good
pay attention to such tasks of HRM as follows idea to invite exterior experts to enterprises
(Lynne et al., 2000; Ashbaugh & Rowan, 2002; to make practical guidance. The human re-
Sun & Xu, 2006): source department can also organize staff
to visit the enterprises that implement
• Enterprises should realize the importance ERP successfully to draw on experience.
of human resources. An ERP system is a Additionally, the human resource depart-
production of IT, while human resources ment should be in charge of the supervi-
are part of the essential power of the inven- sion function or help your enterprise estab-
tion or improvement of IT. The idea that lish a supervision department, made up of
HRM is just a secondary function must be interior managers and exterior experts.
updated.
• Enterprises should redesign the functions
of HRM. ERP extends the traditional func- FUtUrE trENDs
tions of HRM greatly. The simple com-
pensation calculating or personnel man- Enterprise resource planning is a new concept
agement cannot meet the requirements of introduced by the Gartner Group in 2000 to label
an ERP system. The extended functions the latest extensions to ERP (Classe, 2001). The
of HRM are more favorable for inter-de- new concept is that, having successfully integrated
partmental cooperation, because they can internal business applications such as finance,
conveniently provide the decision making sales, and marketing to increase efficiency and
with information needed. create a total overview of the business, ERP II
• Enterprises should establish an effective can be used to integrate external applications with
talent introduction mechanism and an ef- collaborative commerce arrangements, e-business,
fective incentive mechanism for employees. and the supply chain (Payne, 2002). Traditional
An attractive compensation mechanism ERP is the main component in an ERP II system,
can provide useful market competitive ad- but for the purposes of the collaboration, an ERP
vantage for enterprises to hold talents. On II system is opened to inflow and outflow of infor-
the other hand, enterprises should supply mation (Moller, 2003). On the other hand, during
employees with reasonable freedom to the last decade, a new wave of human resource
make decisions within position responsi- technology known as electronic human resource
bilities. This can enhance the working ini- management (e-HRM) has emerged with the ad-
tiative as well as the sense of responsibility vent of intranet- and Internet-based technologies.
of employees. E-HRM is mainly connecting staff and managers
1026
with the human resource department electronically Ashbaugh, S., & Rowan, M. (2002). Technology
through the human resource portal (Lai, 2006). for human resources management: Seven ques-
The basic expectations are that using e-HRM will tions and answers. Public Personnel Management,
decrease costs, will improve the human resource 31(1), 7–20.
service level, and will give the human resource
Barrett, R., & Mayson, S. (2006). Exploring the
department space to become a strategic partner
intersection of HRM and entrepreneurship: Guest
(Rual, Bondarouk, & Velde, 2007).
editors’ introduction to the special edition on HRM
When we combine the ERP II concept with
and entrepreneurship. Human Resource Manage-
e-HRM technology, we find a valuable issue to
ment Review, 16(4), 443–446. doi:10.1016/j.
study—that is, the research about the implemen-
hrmr.2006.08.001
tation of ERP II based on e-HRM. Additionally,
we can also study other details of this issue ac- Bellini, E., & Canonico, P. (2007). Knowing com-
cording to different countries, cultures, industries, munities in project driven organizations: Analys-
and so forth. ing the strategic impact of socially constructed
HRM practices. International Journal of Project
Management, (September): 29.
cONcLUsION
Boudreau, J. W., & Ramstad, P. M. (1997).
Measuring intellectual capital: Learning from
As IT continues to development, there will be
financial history. Human Resource Manage-
more and more enterprises adopting ERP systems.
ment, 36, 343–356. doi:10.1002/(SICI)1099-
ERP can extend the traditional functions of HRM
050X(199723)36:3<343::AID-HRM6>3.0.CO;2-
greatly and also heighten the importance of HRM
W
in enterprises. Enterprises that implement ERP
must perfect the functions of HRM to raise the Bowen, D. E., & Ostroff, C. (2004). Understand-
success rate, so to enhance the whole management ing HRM-firm performance linkages: The role
level of enterprises. A poignant hope is that this of the strength of the HRM system. Academy of
article will be helpful to both scholars and practi- Management Review, 29, 203–221.
tioners who wish to improve the current situation
Classe, A. (2001). Business-collaborative com-
of ERP implementation.
merce—the emperor’s new package. Accountancy,
(November).
rEFErENcEs Davenport, T. H. (1998). Putting the enterprise into
Ahmad, S., & Schroeder, R. G. (2003). The im- (July-August): 121–131.
pact of human resource management practices on
operational performance: Recognizing country Drucker, F. P. (1999). Management challenges
and industry differences. Journal of Operations for the 21st century (p. 97). New York: Harper-
Management, 21, 19–43. doi:10.1016/S0272- Collins.
6963(02)00056-6 Edwards, J. (1999). Three-tier client—server at
Al-Marshari, M. (2002). Enterprise resource plan- work. New York: John Wiley & Sons.
ning (ERP) systems: A research agenda. Industrial Enslow, B. (1996). Which comes first: ERP or
Management & Data Systems, 102(3), 165–170. supply chain planning projects? Gartner Group
doi:10.1108/02635570210421354 Best Practices and Case Studies.
1027
Hsu, K., Sylvestre, J., & Sayed, E. N. (2006). Pasmore, W.A., & Mlot, S. (1994). Developing self-
Avoiding ERP pitfalls. Journal of Corporate managing work teams: An approach to successful
Accounting & Finance, (May-June): 67–74. integration. Compensation and Benefits Review,
doi:10.1002/jcaf.20217 26(4), 15–23. doi:10.1177/088636879402600403
Huselid, M. A. (1995). The impact of human Payne, W. (2002). The time for ERP? Work Study,
resource management practices on turnover, 51(2), 91–93. doi:10.1108/00438020210418827
productivity, and corporate financial performance.
Pfeffer, J. (1995). Producing sustainable competi-
Academy of Management Journal, 38, 635–672.
tive advantage through the effective management
doi:10.2307/256741
of people. The Academy of Management Execu-
Koch, C. (2002). The ABCs of ERP. CIO Maga- tive, 9, 55–69.
zine, (February).
Rual, J. M. H., Bondarouk, V. T., & Velde, M.
Kumar, K., & Hillegersberg, J. V. (2000). ERP (2007). The contribution of e-HRM to HRM
experiences and evolution. Communications of the effectiveness: Results from a quantitative study
ACM, 43(4), 24–26. doi:10.1145/332051.332063 in A Dutch ministry. Employee Relations, 29(3),
280–291. doi:10.1108/01425450710741757
Lai, W. H. (2006). Implementing e-HRM:
The readiness of small and medium sized Schonberger, R. J. (1994). Human resource
manufacturing companies in Malaysia. Asia management lessons from a decade of total qual-
Pacific Business Review, 12(4), 465–485. ity management and re-engineering. California
doi:10.1080/13602380600570874 Management Review, 36(4), 109–123.
Li, Y. F. (2001). Thoughts about the ERP human Shang, S., & Seddon, P. B. (2000). A comprehen-
resource management. Journal of Yunnan Uni- sive framework for classifying the benefits of ERP
versity of Finance and Economic, 17(10), 12–16. systems. Proceedings of the 6th Americas Confer-
ence on Information Systems (pp. 1005-1014).
Lynne, M. M., Axline, S., Petrie, D., & Cornelis,
T. (2000). Learning from adopters’ experiences Siau, K. (2004). Enterprise resource planning
with ERP: Problems encountered and success (ERP) implementation methodologies. Journal
achieved. Journal of Information Technology, of Database Management, 15(1), 1–4.
15, 245–265. doi:10.1080/02683960010008944
Stone, D. L. (2007). The status of theory and
Moller, C. (2003). ERP II—next-generation ex- research in human resource management: Where
tended enterprise resource planning. Proceedings have we been and where should we go from here?
of the 7th World Multi-Conference on Systemics, Human Resource Management Review, 17(2),
Cybernetics and Informatics. 93–95. doi:10.1016/j.hrmr.2007.04.005
Nilakant, V. (1994). Transdisciplinary approach to Sun, X., & Xu, W. (2006). Research on ERP and
a theory of performance in organizations. Human enterprise’s human resources informatization.
Systems Management, 13(1), 41–48. Science Technology Information Development &
Economy, 16(7), 229–230.
Palaniswamy, R. (2002). An innovation-diffusion
view of implementation of enterprise resource
planning (ERP) systems and development of a
research model. Information & Management, 40,
87–114. doi:10.1016/S0378-7206(01)00135-5
1028
Wright, P. M., McMahan, G. C., Snell, S. A., the computer systems of different areas of the
& Gerhart, B. (2001). Comparing line and HR business—production, sales, marketing, finance,
executives’ perceptions of HR effectiveness: Ser- human resources, and so forth—and run them off
vices, roles, and contributions. Human Resource a single database. Also defined as an application
Management, 40, 111–123. doi:10.1002/hrm.1002 and deployment strategy to integrate all things
enterprise-centric.
Wright, S., & Wright, A. (2002). Information
Human Resource: The people that staff and
system assurance for enterprise resource planning
operate an organization.
systems: Unique risk considerations. Journal of
Human Resource Management (HRM):
Information Systems, 16, 99–113. doi:10.2308/
The function within an organization that focuses
jis.2002.16.s-1.99
on recruitment of, management of, and provid-
Yang, J.Y., & Zhao, X.W. (2003). Research on ing direction for the people who work in the
HRM in ERP system. Journal of Beijing Institute organization.
of Technology, (August), 73-77. Information Technology (IT): The collection
of technologies that deal specifically with process-
ing, storing, and communicating information, in-
cluding all types of computer and communications
systems as well as reprographics methodologies.
Manufacturing Resource Planning (MRP
Electronic Human Resource Management
II): A method for the effective planning of all
(e-HRM): The planning, implementation, and
resources of a manufacturing company, includ-
application of information technology for both
ing functions of business planning, production
networking and supporting at least two individual
planning and scheduling, capacity requirement
or collective actors in their shared performing of
planning, job costing, financial management,
HR activities.
forecasting, and so forth.
Enterprise Resource Planning (ERP): An
approach to the provision of business support
software that enables companies to combine
This work was previously published in Encyclopedia of Information Science and Technology, Second Edition, edited by Mehdi
Khosrow-Pour, pp. 1856-1862, copyright 2009 by Information Science Reference (an imprint of IGI Global).
1029
1030
Chapter 4.10
A Study of Information
Requirement Determination
Process of an Executive
Information System
Chad Lin
Curtin University of Technology, Australia
Koong Lin
Tainan National University of the Arts, Taiwan
INtrODUctION understood and least formalized yet most critical

phase of the information systems development
An executive information system (EIS) pro- (ISD) process. This phase is so crucial that many
vides senior management with easy access to information systems researchers argue that IRD
information relevant to their needs. It can spread is the single most important stage during an EIS
horizontally across and vertically down to other project development process, and if the IRD is
organizational managers and provide three ma- inaccurate and incomplete, the resultant system
jor types of benefits: information, management will also be inaccurate and incomplete.
support, and organizational support (Salmeron, Hence, understanding the issues that influence
2002). According to Salmeron, one key EIS suc- the IRD process of EIS is of critical importance to
cess factor is the fulfillment of users’ information organizations (Poon & Wagner, 2001). However,
needs. However, the user information require- little is known about the issues that influence IRD
ments determination (IRD) process during the processes during the implementation of an EIS
implementation of an EIS remains a problematic project (Khalil, 2005). Therefore, this article aims
exercise for most organizations (Walter, Jiang, to examine key issues surrounding the IRD pro-
& Klein, 2003). This is because IRD is the least cess during the implementation of an EIS project
in a large Australian public-sector organization.
DOI: 10.4018/978-1-59904-843-7.ch092 The article first reviews relevant literature with
A Study of Information Requirement Determination Process of an Executive Information System
respect to IRD and EIS. Key findings and issues or inaccurate requirement specifications, lack of
identified from the case study are also presented. user involvement, lack of flexibility of computer-
The chapter examines these findings and issues based information systems, poor communication,
in light of these organizations’ IRD practices, different worldviews of the systems analysts, and
and concludes by providing some lessons for EIS other factors (Guinan, Cooprider, & Faraj, 1998;
project implementation. Kirsch & Haney, 2006). Each of these will be
discussed briefly in the subsections that follow.
bAcKGrOUND Incomplete and/or Inaccurate

requirements specifications
IRD is a critical phase of ISD. IRD is primar-
ily concerned with specific applications such This can often lead an organization to address the
as EIS. IRD has generated a lot of interest and wrong problem or identify incorrect information
debate among researchers and practitioners as a needs. Dissatisfaction of the stakeholders with
potential means for improving the success rates their IS derives from the problem of specifications
of ISD projects such as EIS (Havelka, 2002; Wu not being stated accurately and/or completely
& Shen, 2006). The IRD process, which Browne (Davidson, 2002; Khalil, 2005). This can also
and Ramesh (2002, p. 625) defined as “a set of arise from users having totally unrealistic expec-
activities used by a systems analyst when assessing tations of the final EIS. Therefore, incomplete
the functionality required in a proposed system,” and inaccurate requirements specifications can
has become increasingly important in obtaining often result in identifying the wrong information
the correct and complete set of user requirements. needs or addressing the incorrect IRD problem.
A number of tools and techniques have been This may ultimately lead to EIS project failures.
proposed to support the IRD process during the According to Browne and Ramesh (2002),
EIS project: prototyping, joint application devel- the following challenges should be recognized
opment (JAD), rapid application development by both analysts and users when they are dealing
(RAD), data flow diagrams (DFDs), and entity among themselves:
relationship diagrams (ERDs; Duggan & Thachen-
kary, 2004; Spina & Rolando, 2002). However, • There can never be a complete, correct set
despite the existence of all these techniques and of user information requirements.
tools, the history of ISD has been littered with • Requirements are not stable over time, but
numerous reports of the complete failure of EIS are in a constant process of evolution.
projects (Khalil, 2005). The common causes of • The facilitation skills of systems analysts
these failures stem largely from difficulties in deal- are crucial to the effective management of
ing with the information requirements (Browne the IRD process.
& Ramesh, 2002; Davis, 1987). In many cases, • Systems analysts work in highly political
budget blowouts and missed deadlines occur. Too contexts.
often, initial design and programming is followed
by a reassessment of needs, redesign, and then Lack of User Involvement
more programming (Urquhart, 2001). Many EIS
project failures have little to do with technical One of the major factors contributing to the failures
or programming issues. The source of many of of EIS projects is the lack of user involvement.
these problems lies with one or a combination By failing to be involved during the system de-
of the following major factors: incomplete and/ velopment stages, users might feel frustrated and
1031
disillusioned when they perceive new technolo- Poor communication

gies such as EIS as the threatening creations of
outsiders (Robertson & Robertson, 1999). This Poor communication between users and analysts
usually results in resistance and conflicts between is also a major factor contributing to the failure
the project sponsors, the systems analysts, and the of EIS (Urquhart, 2001). Communication skills
users (Davidson, 2002). of systems analysts have a significant impact on
Lack of user involvement often results in successful and complete information requirements
distrust between the users, the systems analysts, of EIS. Some of the most important reasons for
and the project sponsors. Users feel unable to communication difficulties are as follows (Doug-
specify what they want because they do not know las, 2003; Guinan et al., 1998; Urquhart):
what is possible while the systems analysts try to
explain what is possible but describe it in ways • The different perspectives of the different
not understood by the users (Browne & Rogich, stakeholders involved in a system study
2001; I. Wu & Shen, 2006). This usually not • Uncertainty on the part of the users of the
only reduces job satisfaction on both sides but impact the final system will have on their
also leads to less-than-adequate systems design individual roles in the organization
(Alvarez, 2002). • The observation that the user operates with
informal systems and that the formal pro-
Lack of Flexibility of computer- cedure of the existing systems has been
based Information systems overtaken by less formal, unauthorized
procedures
Computer-based information systems (e.g., EIS) • The problem facing both users and systems
often lack the flexibility to meet changing user analysts that new systems almost certainly
information requirements and have little interac- include technological innovations
tion with the manual systems (Salmeron, 2002;
I. Wu & Shen, 2006). These are often due to the Worldview of the systems Analysts
way computers have to be programmed, in which
any change that involves a change to the program The education and practice of systems analysts can
requires a detailed sequence of steps to be taken, also be the source of the problems when dealing
which can be time consuming and disruptive. with IRD processes since few systems analysts are
Some changes, even changes that appear trivial equipped to deal with the essentially social nature
to the nonexpert user, cannot be incorporated in of IS. The systems analysts tend to think that they
the system without a substantial redesign of the are the experts who analyze the problem, define it,
computerized parts of the system (Lauesen & and provide the solution (Berry, 2002). Many of
Vinter, 2001; Sutcliffe, 2000). Moreover, since the problems of ISD projects such as EIS can be
the organizations and the people within them are attributed to organizational behavioral problems.
dynamic and constantly changing all the time, a These behavioral problems are the result of bad
computer-based information system that takes designs. These bad designs are attributed to the
too long to finish will not be able to meet users’ way systems analysts view organizations, their
needs and hence will become a major stumbling users, and the function of ISD.
block to the success of the EIS.
1032
Other Factors these included two main sponsors of the EIS

project, an external systems analyst, and six key
There are also some other significant factors that users of the EIS.
might affect the success of ISD projects. These The interviews focused on the EIS project de-
include an inaccurate assessment of the scope of velopment process, different stakeholders’ views
the problem and broader organizational issues, of the EIS, the IRD process, and the evaluation
poor budget control, a delay in the development process of the EIS. Each interview lasted between
of applications, difficulty in making changes, hid- 1 to 2 hours. All interviews were taped and the
den backlog, program and software bugs, systems transcripts were sent to the interviewees for vali-
that cost much more to develop and maintain than dation. In cases where there were differences in
expected, and development processes that are not opinion between participants, either follow-up
dynamic (Alvarez, 2002; Browne & Ramesh, interviews were conducted or e-mails were sent
2002; Havelka, Sutton, & Arnold, 2001). to clarify their positions.
Other data collected included some of the actual
project proposals and detailed requirements speci-
rEsEArcH MEtHODOLOGy fications for the EIS project, planning documents,
and some meeting minutes. More than 300 pages
The objective of this research is to examine key of transcripts were coded and analyzed. The data
issues of the user-requirement determination collection at this organization continued until a
process during the EIS project development pro- point of theoretical saturation, which is when the
cess. An in-depth case study was carried out in value of an additional interview was considered
one large Australian public-sector organization to be negligible (Eisenhardt, 1989). Qualitative
involved in the implementation of an EIS project. content analysis was then used to analyze the data
The organization was responsible for providing gathered (Miles & Huberman, 1994). The analysis
an important education service within Australia. of the materials was also conducted in a cyclical
It had an annual turnover of A$500 million and manner and the issues identified were double-
about 3,000 employees. In order to meet the checked by the researchers and other experts. The
necessary educational quality requirements and guidelines (i.e., multiple interpretations) set out
guidelines set out by the Australian government, by Klein and Myers (1999) for conducting and
the organization had decided to implement an EIS evaluating interpretive field studies in information
to assist it in making proper decisions. The objec- systems were followed to improve the quality of
tives of the EIS were to (a) support organizational the research.
reporting in the areas of program and planning
review, annual reporting, and benchmarking and
best practices, (b) support the organization in its rEsEArcH FINDINGs
undertaking of quality-related activities, and (c)
identify deficiencies in data sources. A number of issues emerged from the analysis of
Initially, the researchers had attended six ses- the data and some of the key issues surrounding
sions of the IRD process between the external the IRD process of the EIS project are presented
systems analyst and the key users. On completion below in some detail. Related information from
of all these sessions, the researchers refined and the observation and document review has been
modified the interview questions, which were integrated into the discussion to further support
drafted before these sessions. Then, 16 interviews the findings.
were conducted with nine key participants, and
1033
theme 1: Problems in the users. The users felt that their ability to participate
Using IrD Methodology in the IRD process could be enhanced by having
such a group session.
The interview data suggest that there was a gen- Instead, the IRD process for this EIS project
eral agreement among the users that no ISD/IRD was, as perceived by the users, merely a question-
methodology, tool, or problem-solving methodol- and-answer exercise. Although the users were
ogy had been used by the external systems analyst given the opportunity to raise any questions and
during the IRD process with users for the EIS concerns about the existing systems as well as the
project. Instead, only an interview was carried forthcoming EIS, the problem was that there was
out by the external systems analyst to gather the no prior information given to the users before the
required information from the users. For example, IRD sessions. The users felt that they were not
one user said, “It felt very much like questions to given any time and information to prepare for the
which I was responding because it’s an interview meetings with the systems analyst. The problem
like that....I didn’t feel that I was settling into was compounded by the lack of follow-up by
something that I was participating in. So it’s very the systems analyst. The users did not take part
much like a question and answer.” The user had in the rest of the EIS project and were critical of
expected some sort of methodology to be used the project sponsors and the systems analyst for
by the systems analyst during the IRD sessions. not consulting them about the project. The re-
The researchers’ observation had supported their searchers were told privately by one of the project
claims. Some of the users suggested that the use sponsors that the systems analyst was instructed
of a proven methodology and diagram would be not to involve the users further in other phases
valuable for the IRD process. of the project. The project sponsors were getting
However, the sponsors and systems analyst impatient with some of their users regarding their
claimed that some sort of methodologies had been information requirements.
used during the IRD sessions, although this had
not been observed by the researchers. For example, theme 3: Lack of User satisfaction
the systems analyst said, “I worked loosely to
various methodologies, sort of used in the past, Most users were unhappy with the IRD process
in particular, Arthur Andersen’s Method One and of this EIS project and were not impressed by
APT. But they tended to direct more on experience the performance of the project sponsors and, in
and referencing the documents.” Furthermore, particular, the systems analyst. For example, one
the systems analyst went as far as saying that the user was very critical of the project sponsors and
use of diagrams such as DFDs and ERDs would the systems analyst and said, “I think what they
confuse the users. Most of the users interviewed need to do is to give the user an understanding of
by the researchers had rejected this claim. what they have envisaged the EIS system should
be able to do and where it fits….Also, they should
theme 2: Lack of User Involvement articulate in a way that someone who is not a
systems person can understand.” None of the
All users indicated that their contributions to the users were given enough information and time
IRD sessions had been hampered by the lack of to prepare for the IRD process. For example, one
information. In addition, rather than having several user complained and said, “If people are going
IRD sessions with the systems analyst, most users to be involved [in the IRD process], they need to
suggested that a group session would be far more know why...” The problem had been compounded
effective as it tended to create synergy among the by the instruction by the project sponsors not to
1034
spend too much time listening to the requirements past records, chances are we will. And when that
of the users, and also the fact that the scope of the happens, everyone will pass the buck. The MIS
project was unclear. type of people will say, ‘but I’ve fulfilled what
you have told us.’” All users felt that this had
theme 4: Lack of Project scope been repeated to some extent in this EIS project. A
good example of this was the lack of information
Both users and the systems analyst complained given to the users by the systems analyst before
about the lack of scope and information for this the IRD session.
EIS project. Some of the ideas put forward by the The EIS project had also appeared to be plagued
users included the following: (a) A group session with politics. Many users interviewed were un-
should be deployed to elicit users’ requirements happy with the way that the project sponsor had
and needs, (b) more research should be conducted been given a significant role in this EIS project.
by the systems analyst before the IRD process, On the other hand, the project sponsors also re-
and (c) more information about the purpose of vealed that they were getting impatient with some
the visits by the systems analyst should be given of the users within the organization. The project
beforehand. sponsors had admitted to the researchers that they
As mentioned previously, the reason for not did not get along with some of the users. To make
giving the proper information to the users before the matter worse, the systems analyst also agreed
the meetings could be due to the fact that the with the view expressed by some of the users that
instruction given by the project sponsors to the this EIS project was likely to fail as a result of
systems analyst was to finish the IRD phase as the prevailing culture and politics existing within
soon as possible. For example, the systems analyst the organization. Both the systems analyst and the
said, “Problems that I had with this particular case users had seen many ISD project failures before,
is not being so much with gathering of informa- both within and outside the organization.
tion requirements from users....The problem I had Overall, most of the key issues identified from
with IRD is perhaps, not being able to maintain a this study are largely consistent with the literature.
limited scope.” The systems analyst was having However, the research has further identified that
difficulty in maintaining a limited scope of the lack of user satisfaction and the organizational
EIS project and hence was not able to tell the us- culture and politics also have a major impact on
ers exactly what the project was going to be like. the success of the implementation of EIS projects.
theme 5: culture and Politics

FUtUrE trENDs
Several users pointed out that the culture and
politics within the organization forced many During the last decade, the names of information
employees to be disillusioned about the whole systems have changed from executive informa-
process as they felt that they could not make any tion systems to business intelligence (BI) systems
difference. For example, one user complained (J. Wu, 2000). BI is a significant trend of EIS as
about the culture and politics that existed within the technology has significantly evolved from
the organization that were the cause for users not internally developed graphical user interfaces
being consulted about the implementation of new to packaged applications that provide users with
projects such EIS. This had often led to project easy access to data for analysis. BI is defined as
failures. For example, he said, “Now I hope we the process of monitoring and analyzing busi-
don’t end up with yet another project failure. On ness transactions by using business intelligence
1035
to align business operations with the tactical and ments for the EIS, the lack of user involvement,
strategic goals of the organization. In addition, and user dissatisfaction. It was also surprising to
BI encompasses software for extraction, trans- hear from the systems analyst himself and most
formation, and loading (ETL); data warehousing; users that they were not very optimistic that this
multidimensional or online analytical processing EIS project would succeed due to a long history
(OLAP); data analysis; and data mining. How- of ISD project failures within the organization.
ever, there are still some challenges to overcome A contribution of this short article is that it has
before BI can be used and implemented more further identified that a lack of user satisfaction
widely. These include recognizing BI projects as and issues regarding the organizational culture
cross-organizational business initiatives, engaging and politics have a major impact on the success
business sponsors, and developing an automated of the implementation of EIS projects.
Web intelligence system to extract actionable orga-
nizational knowledge by leveraging Web content.
rEFErENcEs
cONcLUsION Alvarez, R. (2002). Confessions of an information

worker: A critical analysis of information require-
This case study illustrates the dynamic relation- ments discourse. Information and Organization,
ships between project sponsors, users, and the 12, 85–107. doi:10.1016/S1471-7727(01)00012-4
systems analyst during the IRD process of an Berry, D. M. (2002). The importance of ignorance
EIS project. Most of the users’ complaints were in requirements engineering: An earlier sighting
centered on the difficulties in giving accurate and and a revisitation. Journal of Systems and Software,
complete requirements to the systems analyst 60, 83–85. doi:10.1016/S0164-1212(01)00103-0
during the IRD process. Their difficulties not
only stemmed from the inability of the users to Browne, G. J., & Ramesh, V. (2002). Improving
specify what they wanted, but were also affected information requirements determination: A cogni-
by the attitude of the systems analyst and project tive perspective. Information & Management, 39,
sponsors of the EIS project toward the opinions 625–645. doi:10.1016/S0378-7206(02)00014-9
of the users. The results also indicated that there
Browne, G. J., & Rogich, M. B. (2001). An em-
were discrepancies between what the systems
pirical investigation of user requirements elicita-
analyst said about what he did (espoused theory)
tion: Comparing the effectiveness of prompting
and what he actually did (theory in use) during
techniques. Journal of Management Information
the IRD process. For example, the systems analyst
Systems, 17(4), 223–249.
had insisted that some sort of formal methodology
was used to elicit user requirements when in fact Davidson, E. J. (2002). Technology frames and
there was none. framing: A social-cognitive investigation of re-
Moreover, this research has found that there quirements determination. MIS Quarterly, 26(4),
were significant differences in opinion between 329–358. doi:10.2307/4132312
the users and the systems analyst. For example,
Davis, G. B. (1987). Strategies for information
although there was a high degree of agreement
requirements determination. In R. D. Galliers
about the lack of project scope and the existence of
(Ed.), Information analysis: Selected readings
issues in culture and politics, there were significant
(chap. 13). Sydney, Australia: Addison-Wesley.
disagreements about the deployment of the IRD
methodology for gathering information require-
1036
Douglas, H. (2003). A user-oriented model of Lauesen, S., & Vinter, O. (2001). Preventing
factors that affect information requirements deter- requirement defects: An experiment in process
mination process quality. Information Resources improvement. Requirements Engineering, 6(1),
Management Journal, 16(4), 15–32. 37–50. doi:10.1007/PL00010355
Duggan, E. W., & Thachenkary, C. S. (2004). Miles, M. B., & Huberman, A. M. (1994). Qualita-
Supporting the JAD facilitator with the nominal tive data analysis: An expanded sourcebook. CA:
group technique. Journal of Organizational and Sage Publications.
End User Computing, 16(2), 1–19.
Poon, P. P., & Wagner, C. (2001). Critical success
Eisenhardt, K. M. (1989). Building theories from factors revisited: Success and failure cases of
case study research. Academy of Management information systems for senior executives. Deci-
Review, 14(4), 532–550. doi:10.2307/258557 sion Support Systems, 30, 173–418. doi:10.1016/
S0167-9236(00)00069-5
Guinan, P. J., Cooprider, J. G., & Faraj, S. (1998).
Enabling software development team performance Robertson, S., & Robertson, J. (1999). Master-
during requirements definition: A behavioral ing the requirements process. Harlow, England:
versus technical approach. Information Systems ACM Press.
Research, 9(2), 101–125. doi:10.1287/isre.9.2.101
Salmeron, J. L. (2002). EIS evolution in large
Havelka, D. (2002). Requirements determination: Spanish business. Information & Management,
An information systems specialist perspective of 40, 41–50. doi:10.1016/S0378-7206(01)00130-6
process quality. Requirements Engineering, 6(4),
Spina, M. J., & Rolando, J. A. (2002). JAD on
220–236. doi:10.1007/PL00010361
a shoestring budget: CrossTalk. The Journal of
Havelka, D., Sutton, S. G., & Arnold, V. (2001). Defense Software Engineering, 15(7), 26–28.
Information systems quality assurance: The effect
Sutcliffe, A. G. (2000). Requirements analysis
of users’ experiences on quality factor percep-
for socio-technical system design. Information
tions. Review of Business Information Systems,
Systems, 25(3), 213–233. doi:10.1016/S0306-
5(2), 49–62.
4379(00)00016-8
Khalil, O. E. M. (2005). EIS information: Use
Urquhart, C. (2001). Analysts and clients in orga-
and quality determination. Information Resources
nizational context: A conversational perspective.
Management Journal, 18(2), 68–93.
The Journal of Strategic Information Systems, 10,
Kirsch, L. J., & Haney, M. H. (2006). Requirements 243–262. doi:10.1016/S0963-8687(01)00046-4
determination for common systems: Turning a
Walter, B. A., Jiang, J. J., & Klein, G. (2003). Stra-
global vision into a local reality. The Journal
tegic information and strategic decision making:
of Strategic Information Systems, 15, 79–104.
The EIS/CEO interface in smaller manufactur-
doi:10.1016/j.jsis.2005.08.002
ing companies. Information & Management, 40,
Klein, H. K., & Myers, M. D. (1999). A set of 487–495. doi:10.1016/S0378-7206(02)00063-0
principles for conducting and evaluating inter-
Wu, I., & Shen, Y. (2006). A model for explor-
pretive field studies in information systems. MIS
ing the impact of purchasing strategies on user
Quarterly, 23(1), 67–94. doi:10.2307/249410
requirements determination of e-SRM. Informa-
tion & Management, 43, 411–422. doi:10.1016/j.
im.2004.11.004
1037
Wu, J. (2000, February 10). Business intelligence: Executive Information System (EIS): It
What is business intelligence? DM Review, 1-2. provides organizations with a powerful yet simple
tool to view and analyze key factors and perfor-
mance trends in the areas of sales, purchasing,
production, finance, and so forth.
Information Requirements Determination
Business Intelligence (BI): It is the process (IRD): It is a set of activities used by a systems
of monitoring and analyzing business transaction analyst when assessing the functionality required
processes to ensure that they are optimized to meet in a proposed system.
the business goals of the organization. Joint Application Development (JAD): It
Data Mining: It is an information extraction is a process originally developed for designing
activity whose goal is to search large volumes a computer-based system. It brings together
of data for patterns and discover hidden facts business users and IT professionals in a highly
contained in databases. focused workshop.
Data Warehouse: It is a relational database Rapid Application Development (RAD): It
that is designed for query and analysis, and usu- is a methodology for compressing the analysis,
ally contains historical data that are derived from design, build, and test phases into a series of short,
transaction data. iterative development cycles.
This work was previously published in Encyclopedia of Decision Making and Decision Support Technologies, edited
by Frederic Adam and Patrick Humphreys, pp. 807-813, copyright 2008 by Information Science Reference (an imprint
of IGI Global).
1038
1039
Chapter 4.11
Towards Identifying the
Most Important Attributes
of ERP Implementations
Piotr Soja
Cracow University of Economics, Poland
Dariusz Put
Cracow University of Economics, Poland
AbstrAct characteristics of ERP implementation which af-

fect project success. This study builds on data gath-
Enterprise resource planning (ERP) systems have ered using a questionnaire directed toward people
been implemented in various and diverse orga- playing leading roles in ERP implementations in
nizations. The size of companies, their industry, a few dozen companies. Twelve attributes were
the environment, and the number of implemented identified and divided into three sets representing:
modules are examples of their heterogeneity. In effort, effect, and the synthetic measure of success
consequence, a single procedure which leads to calculated on the basis of the obtained data. Two
the success of implementation does not appear to agglomeration methods were employed to identify
exist. Therefore, there have been many implemen- exemplar and anti-exemplar groups and objects.
tations that have failed during, and also after, the These elements were thoroughly analyzed, which
implementation process. As a result, a considerable led to identifying the most and the least desired
amount of research has been trying to identify is- attributes of an ERP implementation project.
sues influencing ultimate project success and also The findings are discussed and related with the
to recognize the best implementation projects. The results of prior research. Finally, implications for
aim of this work is to identify the most important practitioners and concluding remarks summarise
the chapter.
DOI: 10.4018/978-1-60566-146-9.ch007
Towards Identifying the Most Important Attributes of ERP Implementations
INtrODUctION works suggest that benefits gained by large and

small sized organisations seem to be similar (e.g.,
The implementation of an ERP system is a great Shang & Seddon, 2000; Soja, 2005) and other
challenge for a company making the effort of studies advocate that benefits differ by company
introducing such a system into its organisation. size (Mabert et al., 2003).
The implementation project is usually connected Prior studies also suggest other criteria of ERP
with sizeable expenses for computer software and projects grouping that might influence imple-
hardware, as well as for the implementation ser- mentations’ conditions. These criteria include the
vices provided by a system solution supplier (e.g., extent of ERP package modification (Soh & Sia,
Sarkis & Gunasekaran, 2003). The implementation 2005), implementation scope and duration time
effects could be very diverse, beginning from the (Soja, 2005, 2006). The results imply that the
considerable enhancement of enterprise activity implementations’ conditions are diverse depend-
and increase of its profitability, to the rejection of ing on project type defined by dividing criteria.
the system introduced (e.g., Holland et al., 1999; Moreover, the project type can have an impact
McNurlin & Sprague, 2002). The companies in- on the effects achieved by a company as a result
troducing ERP packages into their organisations of ERP implementation. In particular, the project
differ quite significantly. The implementation duration seems to have an important influence on
endeavours called ERP projects comprise both achieved results (Soja, 2005).
simple installations of single modules of a system The multitude of potential factors influencing
and complex solutions dealing with the installation ERP projects is illustrated by the complex divi-
of many system modules in numerous units of a sion presented by Parr and Shanks (2000). They
company (Parr & Shanks, 2000). suggest the following categories for the division
Therefore, ERP implementation projects form of projects: implementation physical scope (single
a very diverse population and in order to compare or multiple site), extent of organisational changes,
particular implementations, one has to keep this level of system modification, module implementa-
diversity in mind so that such a comparison is tion strategy, and allocated resources in terms of
reasonable (e.g., Stensrud & Myrtveit, 2003). time and budget. Taking into consideration the
Thus, it seems appropriate to group purpose- above-mentioned criteria of a division, there are a
fully implementation projects into homogenous great many implementation types. Therefore, Parr
collections, where the comparison of projects is and Shanks distinguish three main categories of
feasible and sensible. Only in this situation can ERP implementations: comprehensive, averagely
we talk about a “model” implementation project complicated (middle-road) and simple (vanilla).
and examine the project discovered in order to Overall, it seems that it is hard to find a gener-
reveal the most needed characteristics. ally accepted division of ERP projects into groups,
Among the methods of projects grouping which would constitute homogenous collections
suggested by prior studies, there are those em- of similar implementations. Prior studies suggest
ploying company size (e.g., Bernroider & Koch, various criteria of ERP projects grouping and
2001; Buonanno et al., 2005; Everdingen et al., these divisions take into consideration merely
2000; Loh & Koh, 2004) and those relying on a the variables defining the efforts made in order
criterion of the number of user licenses (Sedera et to implement a system, but they completely omit
al., 2003). While previous research indicates that the issue of achieved effects. Meanwhile, incorpo-
company size is an important criterion influenc- rating the parameters describing implementation
ing ERP project conditions, the results regarding results could lead to interesting conclusions.
the benefits achieved are mixed. Some research
1040
The goal of this paper is an attempt to dis- eral parties participating in ES adoption: vendors,
cover the most desired attributes of a model ERP consultants and IS/IT managers. They should pay
implementation project. The article is based on attention to improve not only the quality of ERP
research conducted among a few dozen companies products, but also user’s knowledge and involve-
introducing an ERP system into their organisations ment. They should also be aware of the importance
in Poland. In order to achieve the paper’s goal, of choosing suitable consultants and suppliers
the statistical methods of element grouping were (Wu & Wang, 2007). Furthermore, emphasizing
employed, which allowed us to extract the groups the role of knowledge transfer, McGinnis and
of homogenous projects that were then ordered on Huang (2007) introduced the idea of constituting
the basis of the measure of achieved success. This a knowledge-sharing community which may play
procedure allowed us to distinguish the projects the crucial role of a platform that can be used to
having the most desirable characteristics, as well provide a common frame of reference to all ERP
as those with the least desirable attributes. activities. The authors also highlight the concept
that the actual implementation does not finish the
whole adoption endeavour, which should be still
LItErAtUrE rEVIEW monitored and handled after the system rollout.
The quality of the actual implementation process
The idea of discovering issues determining the influences the course of the post-implementation
success of enterprise system (ES) implementation phase. In particular, Nicolaou (2004) suggests five
projects attracted the attention of a considerable critical dimensions which affect the whole adop-
number of researchers. There are a great many tion project, including the post-implementation
research approaches which investigate numer- stage. These issues relate to the review of overall
ous issues and employ various understandings project scope and planning, the review of driving
of project success. The works differ in various principles for project development, the effective-
aspects, i.e. the employed methodologies and ness of misfit resolution strategies, the evaluation
chosen variables. Some scholars examine the of attained benefits, and the evaluation of learning.
actual implementation process, others focus on the The enterprise system adoption projects differ
post-implementation phase, there are also those greatly as regards the scope of the implementa-
who treat ES adoption as a continuous endeavour tion. In general, the most complicated full-scope
without a clearly defined end point. The inves- and highly-integrated enterprise system adoption
tigated issues vary from technological, through projects are perceived as more likely to bring
organisational to those connected with people, the best benefits for the company. This issue is
and from operational to strategic considerations. illustrated by Ranganathan and Brown (2006),
The following section summarises major find- who conclude that a company’s announcement of
ings achieved by prior research connected with ERP adoption positively influences stock market
the attributes and characteristics of successful ES returns. The authors suggest that the greater the
adoption projects. project’s scope (in terms of number of locations
Among many issues analysed by research- and the extent of introduced functionality) the
ers, knowledge seems to play a paramount role stronger the influence.
in enterprise system adoption. In particular, the Researchers highlight the need for identify-
researchers emphasize the need of adequate ing adequate characteristics of enterprise system
knowledge transfer from external consultants in order to achieve its successful adoption. The
to clients in an adopting organisation (Ko et al., enumerated features which appear to be the most
2005). This process should be dealt with by sev- important encompass system quality (Kositanurit
1041
et al., 2006) together with other quality dimen- process improvement and ought to have dual
sions related to information and service (Chien focus: operational and strategic. The authors
& Tsaur, 2007), and perceived usefulness of the demonstrate that firms which had only opera-
system (Amoako-Gyampah, 2007). In particular, tional focus achieved lower performance than
on the basis of research conducted among 571 those having dual or strategic focus. In short,
respondents and employing Technology Accep- the discussion above illustrates that enterprise
tance Model, Amoako-Gyampah (2007) suggests system adoption should be first and foremost an
that users’ intention to use an ERP might depend organisational and business project. It should be
more on how useful they perceived the system treated as a business-led endeavour, in contrast to
than on their level of involvement or how easy it an IT related initiative (e.g., Law & Ngai, 2007;
will be to use the system. Furthermore, the author Nicolaou, 2004; Tinham, 2006).
advocates that perceived usefulness influences The ES adoption project needs constant
the users’ involvement and beliefs concerning monitoring as regards employed resources, both
the necessity of changes. However, on the other internal and external. As far as internal resources
hand, Amoako-Gyampah points out the role of are considered, Peslak (2006), on the basis of the
legacy systems and related people’s habits and opinions of over 200 top financial executives,
expertise, which may negatively influence their advocates that cost and time were found to be
attitudes towards ERP. major determinants of project success and should
Identifying the desired features of the enterprise be carefully measured and monitored. Further,
system is a necessary condition for successful the author also points out that the use of external
adoption. However, equally important is the ap- consultants for implementation should be carefully
propriate implementation process, during which controlled since it was found to adversely impact
one of the most crucial issues is to achieve an cost performance.
alignment between the system’s capabilities and The enterprise system adoption projects form
the company’s needs. Sharma and Vyas (2007) a very diverse group differing in several aspects.
emphasize this issue by talking about the synergy This issue is discussed by Stensrud and Myrtveit
between technology and management, which is (2003), who examined 30 ERP projects using
advocated as an important element influencing Data Envelopment Analysis Variable Returns to
the success of ERP adoption. Scale (DEA VRS) to measure the productivity of
Enterprise system adoption is inevitably con- software projects and employed the method to
nected with some extent of organisational changes identify outstanding projects of ERP systems that
which are carried out in the adopting company. may serve as role models. As a result, they suggest
The more radical approach to organisational that the average efficiency among investigated
changes is called business process reengineering projects is approximately 50%. Furthermore,
(BPR), while the less radical method bears the the authors notice that there were significant
name of business process improvement (BPI) differences in productivity between projects in
(Law & Ngai, 2007). Schniederjans and Kim various industries. Therefore, one should exhibit
(2003), on the basis of a survey conducted among caution when benchmarking and comparing proj-
115 US electronic manufacturing firms, advocate ects across industries. The authors suggest that
that business process change should precede the performance assessments should include both
enterprise system implementation. Furthermore, productivity and quality indicators, and it should
Law and Ngai (2007), drawing for the experience also take into account other external factors such
of 96 companies, conclude that during enterprise as schedule constraints.
system adoption firms should undertake business
1042
As regards the difficulties with the comparison and logistics activities. The idea of limited benefits
of different ES adoption projects, the impact of achieved by small and medium-size companies
organisational size on ES project conditions was is reflected in the research of Sun et al. (2005),
studied by several researchers. The majority of who conclude that as the implementation schedule
authors defined organisational size taking into increases, the cost increases accordingly, while
consideration the number of employees; other the achievement increases to some point, beyond
studies understood the size of a company in terms which there is no significant achievement benefit.
of the level of revenues (Mabert et al., 2003) or
defined the size of an ES project as a number of
installed licences of the system (Sedera et al., rEsEArcH MEtHODOLOGy
2003). While investigating ES projects, the schol-
ars employed research approaches based on case This study is based on exploratory research con-
studies, interviews and surveys. Their respondents ducted among enterprises introducing an ERP
were mainly adopters; however, some studies also package into their organisations. A field study
enquired system supplier representatives (e.g., was adopted as a general research approach, and
Mabert et al., 2003). a questionnaire was employed as a data-gathering
The results of prior works illustrate that in the technique (e.g., Boudreau et al., 2001). The
case of small firms, the most important issues research question posed in this study could be
comprise available human resources and system expressed as follows: What are the most desired
fit into company’s organisation (Bernroider & attributes of an ERP implementation project?
Koch, 2001; Buonanno et al., 2005; Mabert et The research questionnaire was comprised
al., 2003; Muscatello et al., 2003; Raymond & of questions with a mixture of scale, multiple
Uwizeyemungu, 2007), which result in a shorter choice and open questions. The purpose of these
implementation time, lower costs, and lack of the queries was to provide demographic data and
need for significant organisational changes (Adam details necessary to assess project conditions and
& O’Doherty, 2000; Bernroider & Koch, 2001; implementation effects. The list of respondent
Mabert et al., 2003). The paramount significance enterprises was prepared on the basis of reports
of human resources in the case of small and analysing the ERP market in Poland and databases
medium-size companies is expressed by Sun et containing companies’ address data. The resulting
al. (2005), who claim that a great emphasis during list contains firms that introduced an ERP pack-
ERP implementation should be placed on people. age into their organisations with a broad scope,
It is worth noting that the issues connected estimated on the basis of available data.
with organisational changes are not perceived The questionnaire was directed toward the
unambiguously by researchers: some point out people playing leading roles (the project leader, if
the lack of willingness to perform organisational possible) in the implementation. With the help of
changes in the case of small firms, others claim the questionnaire, data has been gathered regarding
that small companies are more likely to change the conditions of implementation projects, efforts
their processes to fit the system (Mabert et al., incurred, as well as results achieved. The collected
2003). The cited research works also suggest that data contain various pieces of information regard-
benefits realized by small and large companies ing both the implementation process and achieved
differ: larger firms achieve greater benefits, with a results. Part of the data contains objective items,
special emphasis on a financial indicators improve- while the other (subjective) include respondents’
ment, while smaller companies accomplish first individual evaluations. The achieved collection of
and foremost the improvement of manufacturing projects is varied; hence, an attempt to identify the
1043
model object, having the most desired attributes In the research conducted, two agglomeration
which led to the completion of implementation methods were employed: hierarchical Ward’s
goals, is not an easy task. method (e.g., El-Hamdouchi & Willett, 1986),
The group of objects was characterised by which is the most commonly used agglomerative
12 attributes, which were divided into 3 distinct method employed for forming clusters (Everitt,
subsets. In the first subset, there were input indi- 1993) and non-hierarchical k-Means method. The
cators of an implementation process—let us call aim of Ward’s method is to join objects together
them “effort indicators”. The second group was into ever increasing sizes of clusters using a
comprised of variables relating to the implementa- measure of similarity of distance. K-means, on
tion results—called “effect indicators” (see Table the other hand, is a simple non-parametric clus-
1). The third, one-element subset, contained the tering method that minimises the within-cluster
calculated variable being a synthetic measure of variability and maximises the between cluster
implementation success, which was calculated variability. The k-means method requires that
on the basis of data gathered from the enterprises the number of clusters is specified beforehand
(Soja, 2004). In the next stage of the research, (National Statistics 2001 area classification,
this measure was used to establish the hierarchy 2001). One possibility for obtaining the number
of the groups. is to run Ward’s method and use the outcome as
The success synthetic measure, based on the initial configuration for k-means.
understanding of success in the information sys- Since in the k-Means method a researcher
tems domain (e.g., Lyytinen, 1988), employs 5 has to arbitrarily provide the number of clusters,
partial measures: (1) the actual scope of an imple- the two-phased approach is common in cluster
mentation with respect to the planned implemen- analysis research. In the first stage, the hierarchical
tation, (2) the actual duration with respect to the method is applied in order to determine a prelimi-
assumed duration, (3) financial budget with regard nary number of clusters (e.g., Ward’s method),
to the planned budget, (4) users’ level of satisfac- and in the second step the actual classification of
tion from the system introduced, and (5) the ex- objects using the k-Means method takes place (e.g.,
istence and achievement of project goals (Soja, Everitt et al., 2001). This approach was adopted
2006). during this research and is illustrated in Figure 1.
The procedure is aimed at the separation of
object groups which are similar to each other but
differing to a greater extent from the objects be-
Table 1. Implementation effort and effect variables longing to the remaining groups (e.g., Kaufman
& Rousseeuw, 1990). Firstly, the standardization
Effort variables Effect variables
of variables was carried out, which allowed us to
• Company size measured • Actual duration time of an remove the excessive influence of variables hav-
by number of employees implementation (AD)
(Size) • Measure of financial budget ing a wide range of values from the outcome of
• Planned duration time spending with regard to the the research. In the next step, the Ward’s hierar-
of an implementation planned budget (Bud)
(PD) • Implemented scope (Scope) chical grouping method was applied. On the basis
• Zero-one variable bear- • User satisfaction indicator (US) of a distance diagram obtained, two decisions
ing information whether • Level of achievement of project
the MRP Explosion goals (Goal) were made:
module was implemented • Subjective measure of positive
(MRP) effects of an implementation (PE)
• Number of implement- • Subjective measure of negative 1. Some objects were excluded from further
ed modules except MRP effects of an implementation (NE) processing. The objects most dissimilar to
Explosion (Mod)
other items or those forming small, two-
1044
Figure 1. Research model
element groups, were treated as accidental cluster, the average value of success measure
measurement. Their exclusion allowed us, was calculated using synthetic success measures
at the next stage, to receive more homog- evaluated for objects belonging to a particular
enous clusters, containing objects the most group, and on the basis of this value, hierarchy of
similar to each other and laying closer to the the group was determined. The cluster having the
hypothetical centre of a cluster. greatest average value of success measures was
2. The k value was selected for the applied k- recognised as containing objects with the most
Means method. The greater the k, the more desired characteristics. Simultaneously, the cluster
the clusters. These clusters tend to be smaller with the least average value of success measures
and contain more similar objects. A small k was recognised as having objects with the least
means, on the other hand, fewer groups and desirable attributes.
more diverse objects within each subset. In Within each of these two extreme groups, one
order to determine the k value, the distance object having the smallest distance from the hypo-
diagram achieved with the use of Ward’s thetical centre of a cluster was distinguished. The
method was employed. After excluding object coming from “the best” group was regarded
objects dissimilar to other items, the visual as exemplar (a model implementation), while the
analysis of number clusters in the diagram object extracted from the group having the least
was performed. average value of success measure was perceived
as anti-exemplar (an anti-model implementation).
During the next stage, the k-Means method Since the calculation was performed three times,
was applied. The calculations were performed three exemplars and three anti-exemplars were
three times: (1) for effort indicators only, (2) for extracted, characterising the most needed and,
effect indicators only, and (3) for all eleven in- also, the least desirable attributes of variables
dicators of effort and effect together. As a result, describing efforts, effects as well as efforts and
the separated groups of similar objects were ex- effects jointly (in some cases two objects were
tracted, together with the distance of each object distinguished, since both were equidistant from
from the hypothetical centre of a cluster. For each the centre).
1045
Table 2. Companies by industry

The detailed analysis of the attributes of ex-
emplar and anti-exemplar objects, as well as basic Branch / Industry n %
statistics calculated for clusters distinguished as Machinery Manufacturing 12 19%
the best and the worst, allowed us to draw conclu- Food Manufacturing 12 19%
sions as regards to the most needed and the least Chemical Products Manufacturing 11 17%
desired parameters of an implementation project.
Metal Products Manufacturing 8 13%
Thus, the research question can be answered, i.e.
Trade 6 9%
the most desired attributed of an ERP project can
Electrical Equipment Manufacturing 5 8%
be elicited. These conclusions could be a sug-
Power Industry 5 8%
gestion for people responsible for running ERP
Construction 2 3%
implementation projects, so that they pay atten-
Finance 2 3%
tion to certain facts, which contribute to project
Other 1 2%
success or failure.
rEsEArcH DAtA 1000 workers, and constitutes more than one fifth
of the companies researched. The second larg-
During the research, 223 enterprises were con- est group is made up by the biggest companies
tacted and 68 (30%) answers were obtained from employing over 1000 workers, which represents
enterprises representing the whole country and 20% of enterprises evaluated. Certainly, the least
various industries. From among the questionnaires numerous group is formed by small companies,
received, 64 were accepted for further analysis. employing not more than 100 workers.
All enterprises investigated in this study represent The implementation projects researched make
companies which introduced an ERP system into up quite a diverse group when project duration
their organisations. The companies classified by time is taken into consideration. Among the com-
industry type are described as shown in Table panies examined, there are projects lasting not
2, where the number of companies belonging to more than a couple of months, as well as imple-
particular industries was provided. As can be easily mentations with a duration time longer than 3
seen, the vast majority of companies comprise of years. Table 4 illustrates the number of projects
manufacturing enterprises (75%). as regards planned and actual duration time.
For the purpose of analysis, this study ad- The examined projects are also diverse as
opted the criterion defining enterprise size as the regards to the implementation scope defined by
number of employees. The understanding of
“small” and “large” companies is derived from
the European Community’s definition for small Table 3. Companies by number of workers
and medium-sized companies (e.g., The Commis- Number of workers n %
sion of the European Community, 1996).
20 to 50 3 5%
The investigated enterprises differ significantly
51 to 100 3 5%
in their size regarding the number of employees,
101 to 200 10 16%
which can be seen in Table 3. It contains, in sub-
201 to 300 10 16%
sequent rows, the number of companies (column
301 to 500 11 17%
n) employing a number of workers which falls
501 to 1000 14 21%
within a specified range. The largest group is
over 1000 13 20%
formed by companies employing from 501 to
1046
Table 4. Projects by duration time Table 5. Projects by implemented modules
Number of companies by project Module n %

Duration time duration
Finance 61 95%
planned actual
Inventory 59 92%
up to 6 months 11 9
Sales 55 86%
6 to 12 months 19 18
Purchasing 54 84%
1 to 1,5 year 18 14
Shop Floor Control 37 58%
1,5 to 2 years 4 9
MRP Explosion 29 45%
2 to 3 years 9 7
3 and more years 3 7
Table 6. Projects by number of implemented
modules (without MRP Explosion)
the number of installed modules of an ERP system.
Number of modules (without MRP n %
The following modules were taken into consid- Explosion)
eration: Finance, Purchasing, Inventory, Sales, 1 4 6%
Shop Floor Control and MRP Explosion. The last 2 1 2%
module is treated with special attention, because 3 6 9%
its implementation is exceptionally difficult and 4 23 36%
usually requires previous implementation and
5 30 47%
established use of several key modules of a system.
Table 5 contains the number of companies imple-
menting subsequent modules of a system, and Table 7. Number of ERP packages implemented
Table 6 includes numbers of companies by the
total number of modules introduced, with the # ERP Package
exception of the module MRP Explosion. 17 R/3
Finally, it seems interesting to present the range 7 IFS Apps
of implemented ERP packages, which is visible 6 MMRP (MicroMRP)

in Table 7. It contains ERP system names and the 5 MFG/Pro
number of companies adopting a particular pack- 3 Baan IV, Exact, Scala
age (column #). As can be seen, the projects re- 2
Adaptix*, Digitland Enterprise*, MAX (ICL), Movex,
Tetra cs/3
searched form a very varied collection, introduc-
ing 26 various packages. The world’s leader, SAP ASW, Concorde, Fourth Shift, JDEdwards, Komadres*,
1 Manager II*, Mapics, One World, Oracle Apps, Plati-
R/3, is clearly the most popular solution and was num ERA, Prodis, Promis S/4, System 21, Triton
introduced in 25% of the projects researched. * package developed in Poland
Then came IFS Applications, then MicroMRP,
and MFG/Pro. It is interesting to note that their
usage is three times lower than SAP’s. Also, it is rEsULts
interesting that the vast majority of researched
companies introduced localised foreign solutions– Pictures 1–3 contain distance diagrams achieved
only 6 firms implemented packages developed by application of the Ward’s method, consecu-
and known in Poland (9%). tively using effort variables, effect variables, and
all variables. On the basis of diagram analysis,
the selected observations were excluded from
1047
further processing – they were different in each The outcome of the division of object collec-
case. The longer the vertical line on the diagram tion with the help of k-Means method is pre-
(see Figure 2, Figure 3, and Figure 4), the less the sented in Table 9. The clusters are ordered from
observation is similar to the others. Those which the largest to the smallest average value of success
were represented by the longest lines or consti- measure. This means that the group having a
tuted small two-element groups where excluded. smaller identifier contains objects with more
Table 8 contains the list of objects together with desired properties from an implementation effi-
the ultimate cardinality of the object sets used in ciency point of view. Along with each object
further research. This table also includes the k identifier, the distance from the hypothetical
value determined on the basis of the analysis of centre of an appropriate cluster was placed. The
object clusters classified for further processing.
Figure 2. Distance diagram obtained with the use of Ward’s method applied for 4 effort variables
Figure 3. Distance diagram obtained with the use of Ward’s method applied for 7 effect variables
1048
Figure 4. Distance diagram obtained with the use of Ward’s method applied for all 11 variables
Table 8. List of objects excluded from further processing on the basis of Ward’s distance diagram analysis
(see Figures 2, 3, 4)
Measurement for 4 effort Measurement for 7 effect Measurement for all 11

variables variables variables
Excluded objects A37, A63, A25, A49, A03, A18, A53, A41, A58, A18, A54, A63, A53, A63, A22, A05, A58, A18,
A21, A02 A05, A28, A31, A37, A55, A02, A34, A39, A51, A35, A37, A29,
A26, A67, A48, A07 A41, A03, A20, A28, A54, A14
Number of objects
selected for further 56 48 46
analysis
Chosen k value 7 8 8
table also contains an average success measure well planned: actual implementation time is
determined for each cluster. similar to the planned one (although in all cases
slightly exceeded), and budget is only insignifi-
Exemplar and Anti-Exemplar cantly exceeded. Correspondingly, the planned
Objects’ characteristics scope of an implementation was 100 percent
completed. Predictably, the level of goals achieve-
On the basis of the data obtained by employing ment is very high (equal to 3 to 5, where 5 is a
k-Means method, the exemplar and anti-exemplar maximum value), and the satisfaction level of
objects were chosen. The achieved results were users is estimated at the level of 4 (maximum 5)
put together in Table 10. in all cases. Furthermore, the great advantage of
The data in Table 10 allow us to draw certain subjective positive effects indicated by respon-
conclusions regarding the characteristics of ex- dents over negative effects demonstrates user
emplar and anti-exemplar objects. In all three satisfaction, and, indirectly, project success.
cases, exemplars are characterised by quite a long The objects distinguished as anti-exemplars are
implementation time. The whole undertaking is characterised mainly by a short planned imple-
1049
Table 9. Hierarchy of clusters obtained with the use of k-Means method, determined on the basis of
average success measure
Group For 4 effort variables For 7 effect variables For all 11 variables
ID
Object identifiers Average Object identifiers Average Object identifiers Average
(distance from cluster success (distance from cluster success (distance from cluster success
centre in parentheses) measure centre in parentheses) measure centre in parentheses) measure
1 A27(.328), A31(.068) .8339 A01(.209), A03(.134) .8528 A27(.232), A31(.251) .8339
A42(.395), A48(.542) A35(.166), A43(.172) A42(.347), A48(.432)
A52(.277), A57(.277) A44(.122), A56(.200) A52(.298), A57(.234)
A60(.082) A57(.217) A60(.179)
2 A22(.000), A35(.117) .8013 A08(.255), A14(.127) .7979 A01(.191), A10(.261) .8059
A39(.000), A51(.117) A33(.079), A38(.281) A30(.166), A44(.232)
A52(.133), A62(.166) A50(.233)
3 A08(.216), A11(.247) .7553 A21(.215), A25(.131) .7950 A08(.303), A13(.220) .7983
A13(.104), A15(.158) A27(.156), A46(.248) A16(.252), A23(.301)
A16(.250), A20(.126) A60(.038) A43(.282), A49(.285)
A23(.065), A43(.096) A56(.162), A65(.203)
A45(.219), A53(.142) A66(.153), A68(.201)
A56(.038), A59(.294)
A65(.038), A66(.038)
A68(.158)
4 A01(.188), A04(.188) .7536 A10(.294), A11(.230) .7853 A07(.243), A09(.187) .7812
A30(.154), A46(.383) A19(.265), A20(.285) A17(.219), A26(.206)
A50(.195), A55(.394) A40(.224), A47(.254)
A58(.238) A50(.167), A51(.142)
5 A05(.282), A07(.101) .7402 A13(.161), A16(.255) .7681 A33(.192), A38(.285) .7679
A09(.142), A10(.225) A17(.086), A36(.174) A46(.279), A59(.288)
A14(.303), A19(.142) A65(.168), A66(.123) A62(.176)
A26(.142), A29(.318)
A44(.033), A64(.396)
6 A06(.094), A17(.059) .6525 A09(.102), A23(.253) .7671 A02(.272), A04(.432) .6902
A40(.071), A41(.093), A30(.215), A39(.160) A21(.286), A25(.310)
A47(.108) A42(.156), A49(.177) A55(.228)
A59(.196), A68(.167)
7 A28(.140), A33(.244) .6363 A04(.396), A06(.222) .6499 A06(.265), A11(.263) .6746
A34(.257), A36(.235) A22(.331), A29(.249) A15(.284), A19(.279)
A38(.292), A54(.133) A34(.310), A64(.286) A40(.274), A45(.449)
A62(.177), A67(.140) A47(.304), A64(.463)
8 A15(.122), A45(.122) .6327 A36(.112), A67(.112) .6023
Note: Exemplar and anti-exemplar objects are bold
mentation time, together with a relatively large estimated to be lower than in the case of positive
number of implemented modules. In decidedly implementations (below 3 except for 2 cases). The
most cases, the companies did not implement users perceive a somewhat small number of posi-
the MRP module; therefore, they could not be tive and negative effects from the implementation,
treated as the most extensive projects introducing and, what seems to be interesting, their number is
a system in its full functionality. The budget was similar, and in 3 cases there were more positive
exceeded, which could suggest that the project effects than negative ones.
was not properly planned. In these companies, It is worth noting that the R/3 package from
the implementation scope was not entirely re- SAP was implemented in the case of exemplar
alised; therefore, the level of goal completion is projects obtained on the basis of both effort vari-
1050
Table 10. Exemplar and anti-exemplar objects of implementation process
ID Succ Size Industry System Mod MRP PD AD Bud Scope US Goal PE NE

Effort variables
Exemplar A31 .877 >1000 Transport R/3 4 0 24 30 110 98 4 5 4 0
Anti-
A54 .363 >500 Food Concorde 3 0 5 5 200 25 3 2 1 0
exemplar
Effect variables
Exemplar A44 .891 >300 Food R/3 5 1 12 13 110 98 4 5 4 1
A15 .676 >300 Food R/3 4 0 6 6 110 90 2 2 2 2
Anti- A45 .589 >200 Food Exact 4 0 1 1 100 80 1 1 1 2
exemplar
Electrical
A06* .574 >100 MMRP 5 1 8 18 150 90 3 3 2 3
Eq.
All variables
Exemplar A60 .822 >1000 Power IFS 3 0 24 30 100 100 4 3 4 1
A36 .635 >500 Chemical Exact 4 0 3 5 130 80 4 2 3 0
Anti-
A67 .569 >500 Chemical Scala 4 0 6 6 130 80 3 0 3 0
exemplar
A11* .689 >100 Metal Manager II 5 0 12 12 130 80 3 3 2 2
Symbols: ID – company identifier, Succ – average success measure, Other – as in Table 1
* – chosen from the next to last group, because anti-exemplars A15 and A45 as well as A36 and A67 belonged to two-element groups
ables and effects variables. However, on the other Table 11, where the clusters having the objects
hand, an exemplar project which was distinguished with the most desired characteristics are depicted,
on the basis of all variables introduced the IFS and in Table 12 containing data regarding clusters
system. The interesting fact is that in the case of with the worst objects. The first table comprises
effect variables, both exemplar and anti-exemplar two parts (instead of three) because the analysis
projects implemented the R/3 package, and both performed with the use of effort variables and
were from the food industry. Practically all model all variables yielded the same exemplar groups
objects introduced foreign packages, only one “all containing the same objects.
variables’ anti-exemplar” object implemented The analysis of data put together in Tables 10
Polish software (Manager II). The results show and 11 leads to certain general conclusions regard-
that there is no single ERP package connected ing the implementation process in those pattern
with exceptional implementation performance. companies. The objects included in exemplar
clusters represent various industries, while the
Exemplar and Anti-Exemplar anti-exemplar groups mainly comprise companies
clusters’ characteristics from chemical (6 companies out of 12) and food
(4 companies) industries. Planned duration time
In order to verify the observations achieved on is longer in exemplar clusters than in anti-exem-
the basis of exemplar and anti-exemplar object plar groups, while the number of implemented
analysis, for all extreme clusters (containing the modules is similar or even bigger in anti-exemplar
best and the worst objects), the basic statistics were groups. This means that, in the case of anti-ex-
estimated (average, minimum and maximum) on emplar projects, the implementation duration time
the basis of attribute values of objects belonging was estimated too optimistically. The implemen-
to particular clusters. The results are presented in tation scope is near 100 percent among exemplars
1051
Table 11. Average, minimum and maximum values of variables estimated for clusters containing objects
with the best parameters
Succ Size Industry System Mod MRP* PD AD Bud Scope US Goal PE NE

Effort variables and all variables. number of objects = 7
Avg .834 >1000 4 power 3.6 22.9 24.4 106 97 3.57 3.9 3.3 0.7
3 IFS
1 machinery
Min .776 >1000 3 R/3 2 0% 14 14 100 90 3 3 2 0
1 food
1 One World
Max .877 >1000 1 transport 5 36 36 110 100 4 5 4 1
Effect variables. number of objects = 7
Avg .853 >500 2 food 3.9 9.6 9.6 103 93 3.57 4.4 4 1
2 MMRP
1 chemical
Min .775 >50 2 R/3 1 3 4 100 80 3 4 3 1
1 machinery 57%
1 Exact
1 metal
Max .909 >1000 1 IFS 5 18 18 110 100 4 5 5 1
1 power
* % of companies in group implementing MRP module
Table 12. Average, minimum and maximum values of variables estimated for clusters containing objects
with the worst parameters
Succ Size Industry System Mod MRP* PD AD Bud Scope US Goal PE NE

Effort variables. number of objects = 8
Avg .636 >500 2 Adaptix 3.1 8.4 11.4 124 82 3.38 1.8 2.8 0.8
1 Concorde
Min .363 >500 4 chemical 1 3 5 100 25 3 0 1 0
1 Exact
2 food
1 IFS 0%
1 machinery
1 Oracle
Max .843 >500 1 power 4 14 18 200 100 4 4 4 4
1 Scala
1 R/3
Effect variables. number of objects = 2
Avg .633 >200 4 3.5 3.5 105 85 1.5 1.5 1.5 2
1 Exact
Min .589 >200 2 food 4 0% 1 1 100 80 1 1 1 2
1 R/3
Max .676 >300 4 6 6 110 90 2 2 2 2
All variables. number of objects = 2
Avg .602 >500 4 4.5 5.5 130 80 3.5 1 3 0
1 Exact
Min .569 >500 2 chemical 4 0% 3 5 130 80 3 0 3 0
1 Scala
Max .635 >500 4 6 6 130 80 4 2 3 0
* % of companies in group implementing MRP module
and is considerably lower among anti-exemplars. comes; however, to a lesser extent than among
User satisfaction is definitely greater in the case exemplar projects.
of exemplar clusters; furthermore, the apparent The exemplar group extracted on the basis
advantage of subjective positive effects over of both effort and all variables consists of the
negative results was observed. Nevertheless, in largest companies employing more that 1000
the case of anti-exemplar clusters, positive effects people. This suggests that in the largest enterprises
were more often recognised than negative out- the ERP implementation brings about the best
1052
results. However, within this group, on average sent mainly food, but also metal, machin-
less that 4 system modules were introduced, and ery, and other industries. However, on the
none of the projects installed an MRP Explosion other hand, almost all anti-exemplar com-
module. Hence, these implementations can not be panies were manufacturing enterprises.
recognised as the most complicated. On the other • The lowest success level in implementing
hand, the most complicated full-scope implemen- ERP system was reached by companies op-
tations make up the majority of projects reaching erating in food and chemical industries, as
the best effects, i.e. belonging to the exemplar well as those belonging to the group of me-
group obtained on the basis of effect variables. dium enterprises as regards to the number
Namely, 57 percent of the “best effects” exemplar of employees. Therefore, the results sug-
implementations introduced an MRP Explosion gest that company industry can be a signifi-
module and the projects of this kind, on average, cant factor for the project success.
installed 4 other modules. • Implementations tended to be more suc-
The exemplar groups contain projects imple- cessful in large enterprises of 1000 or more
menting only foreign systems, mainly those most employees. On the other hand, implemen-
popular among implementations researched, tations ended with failure, i.e. achieved a
i.e. R/3, IFS, and MMRP. Naturally, this can be very low level of success measure, mainly
partially explained by the frequency of their oc- among medium sized enterprises of 500 to
currence. However, the presence of well known 1000 employees.
foreign packages, which have mature software • Exceptionally good effects were achieved
solutions and implementation methodologies by companies implementing an ERP sys-
rooted in long-term experience, suggests that the tem within its full functionality. Moreover,
system solution reliability is the deciding factor. all anti-exemplar clusters include only par-
On the other hand, practically all exemplar pack- tial projects. Thus, the integrating aspect
ages were present among anti-exemplar projects of a system is seen when it embraces the
or clusters. This suggests that the deciding factor company holistically.
is not only the system itself, but rather the way
it is implemented into a particular organisation. The above-mentioned observations raise the
issue of ERP system fit, i.e., whether a particular
system solution fits a given company, and, also,
DIscUssION OF FINDINGs whether a company really needs such a compli-
cated system. A good illustration of this is the fact
Lessons Learned that chemical companies were present among the
worst performers. However, none of them used
On the basis of the research results, the following the packages renowned for their outstanding
observations can be made. performance in chemical industry, like SAP R/3
(Stefanou, 2001).
• The company type does not seem to be a Furthermore, the results suggest that a par-
factor deciding about ERP implementation ticular system solution is not a factor determining
success. Companies obtaining the best ef- project success. It turned out that the particular
fects as a consequence of ERP implementa- package was present both among outstanding
tion are mainly manufacturing enterprises, projects and worst performers. Instead, the way
though it is difficult to indicate any specific of introducing the system seems to play a vital
industry that they belong to – they repre- role for project outcome.
1053
comparison with Prior research hand, Sun et al. (2005) emphasize that for small
and medium sized companies people related issues
This study’s findings, claiming that the company are of paramount importance.
type does not seem to be a deciding factor in The results illustrating the vital need for an
ERP implementation success, partially support adequate amount of time planned for an imple-
the findings of Ettlie et al. (2005). Namely, the mentation project are consistent with findings
authors concluded that the strategic predictors regarding difficulties during enterprise system
of a successful enterprise system deployment do implementation and impediments to its success.
not depend on a company’s industry, which is The prior studies recognize mainly organisational
defined very broadly as the firm’s core activity: problems connected with time over-runs (Kremers
manufacturing versus service. This definition of & van Dissel, 2000; Soja, 2008; Themistocleous
a company’s industry seems equivalent to the et al., 2001) and the alignment of organisational
understanding of company type employed by this structure with enterprise system (Kim et al., 2005;
study. Nonetheless, this research results imply that Wright & Wright, 2002). This study’s results con-
a company’s actual industry plays a crucial role firm the findings of Peslak (2006) who perceive
and that practitioners have to pay special attention time and budget as the major determinants of
while implementing ERP system in chemistry project success. ES adopting companies should
and food industries companies. These findings use time wisely and adequately plan education and
are consistent with the results of Stensrud and training so that when the system goes live, users are
Myrtveit (2003), who concluded that there were comfortable with it and understand what they are
significant differences in productivity between supposed to be doing, how and why. This should
projects in different industries, and, also, that also be ongoing long after the implementation is
projects conducted in the process industry were complete (Tinham, 2006).
the least efficient. The second impediment most often recog-
As regards the role of company size in enter- nized by prior research, i.e. the alignment of
prise system adoption, as was already mentioned, organisational structure with enterprise system,
there are mixed results presented by various reis illustrated by the issue of system fit and imple-
searchers. This study’s findings also contribute to mentation scope, raised by this study outcome.
this debate and suggest that results achieved by Prior research suggests that the issue of system
large companies are greater than by small firms, fit especially concerns smaller companies, who
which supports the findings of some prior research tend to suffer more from the system misfit (e.g.,
(e.g., Mabert et al., 2003). In particular, they are Soja & Paliwoda-Pękosz, 2007). The idea of fit
consistent with the results of Sun et al. (2005) between the system characteristics and the com-
and their claims that small and medium sized pany’s needs is highlighted by Peslak (2006), who
companies’ achievement increases to some point, advocates that modification to enterprise systems
beyond which there is no significant achievement should be minimized since they negatively affect
benefit. Nonetheless, we must bear in mind that both cost and time performance.
there are other research works suggesting that the The fit between the system features and com-
benefits achieved by small and large companies panies’ needs is also connected with the scope of
are similar (e.g., Shang & Seddon, 2000). Also, an implementation project perceived in terms of
there are research works suggesting that difficul- introduced functionality. This study’s findings
ties experienced during enterprise system adoption illustrate that better results were achieved by
generally do not differ across company size (Soja companies which implemented greater scope of
& Paliwoda-Pękosz, 2007). However, on the other ERP modules. These results are consistent with
1054
the findings of Ranganathan and Brown (2006) portance of a project and the lack of care
who discovered that the announcements of ERP during execution.
adoptions created greater abnormal stock market • It is necessary to be careful in the case of
returns for ERP adoptions with greater functional implementation projects conducted in the
scope than for those with lesser functional scope. food and chemical industry – the proj-
Finally, the outcome of this study is consistent ects in companies representing those two
with the research works illustrating that enterprise branches most often ended with failure.
system implementation should be treated as a This also suggests the need for further re-
business-led project, in contrast to an IT related search on the influence of a company’s in-
initiative (Law & Ngai, 2007; Nicolaou, 2004; dustry on the project as a whole.
Tinham, 2006). In particular, this study’s results • The implementers should pay special at-
reveal that the best projects demonstrated the tention to the choice of a particular system
greatest number of declared goals and the highest solution. They have to ensure the proper
level of goals’ achievement. This suggests that the fit between a system and the adopting
best projects among the sample investigated were organisation.
treated as business-driven endeavours.
cONcLUsION
Implications for Practitioners
The study examines the ERP implementations
Taking the research results into consideration, and, using the statistical methods of elements
a series of suggestions for practitioners dealing grouping, extracts the projects with best and worst
with ERP projects could be formulated. Making parameters. The core contribution of this paper
use of these suggestions can have a positive influ- is that it illustrates the new method of estimating
ence on an implementation project course and its ERP implementation success factors by employ-
final outcome. ing combined methods of clustering analysis. The
study’s results can be useful for practitioners as
• The implementation endeavour has to they suggest some recommendations towards ERP
be well planned – the best results were implementation improvement. These suggestions
achieved by companies where actual dura- emphasise the need for the proper organisation
tion time was similar to the planned time; of the project and the issue of system fit to the
also budget was as planned or only insigni- particular business environment. Furthermore, this
ficantly exceeded. It is necessary to ensure study should benefit the academic community as
adequate time for system implementation; it shows an innovative method of investigating the
haste can be a factor causing problems and issues influencing ERP project outcome. Further
having influence on a weak ultimate effect. research can enhance the process described by
In the research conducted, projects from introducing more variables capturing project’s
the weakest group had an average time effects and efforts, and, also can establish new
planned and usually exceeded this time. categories of projects’ estimation, such as imple-
• Special attention should be paid to partial mentation efficiency. The main limitation of this
scope implementations; according to the study is the sample of respondents. Though the
results obtained, such implementations too number of research participants is quite substantial
often end with failure. This could be con- (64), further analysis should cover more compa-
nected with an underestimation of the im- nies and ensure better distribution of projects.
Particularly, this suggestion applies to package
1055
type, since having comparable samples of adopt- El-Hamdouchi, A., & Willett, E. (1986). Hier-
ers of various packages would allow us to better archic Document Classification Using Ward’s
investigate the issues connected with a system fit Clustering Method. In Proceedings of the 9th
and its performance. The results also suggest the International ACM SIGIR Conference on Research
need for further research on the projects’ condi- and Development in Information Retrieval. New
tions depending on a company’s industry. York: ACM Press, pp. 149-156.
Ettlie, J. E., Perotti, V. J., & Joseph, D. A. (2005).
Strategic predictors of successful enterprise
rEFErENcEs
system deployment. International Journal of
Adam, F., & O’Doherty, P. (2000). Lessons from Operations & Production Management, 25(10),
enterprise resource planning implementation in 953–972. doi:10.1108/01443570510619473
Ireland – towards smaller and shorter ERP projects. Everdingen, Y., Hillegersberg, J., & Waarts,
Journal of Information Technology, 15, 305–316. E. (2000). ERP adoption by European midsize
doi:10.1080/02683960010008953 companies. Communications of the ACM, 43(4),
Amoako-Gyampah, K. (2007). Perceived Useful- 27–31. doi:10.1145/332051.332064
ness, User Involvement and Behavioral Intention: Everitt, B. S. (1993). Cluster Analysis. Edward
an Empirical Study of ERP Implementation. Arnold, London.
Computers in Human Behavior, 23, 1232–1248.
doi:10.1016/j.chb.2004.12.002 Everitt, B. S., Landau, S., & Leese, M. (2001).
Cluster Analysis. London: Edward Arnold
Bernroider, E., & Koch, S. (2001). ERP selection
process in midsize and large organizations. Busi- Holland, C., Light, B., & Gibson, N. (1999). A
ness Process Management Journal, 7(3), 251–257. Critical Success Factors Model for Enterprise Re-
doi:10.1108/14637150110392746 source Planning Implementation. In Proceedings
of the 7th European Conference on Information
Boudreau, M., Gefen, D., & Straub, D. (2001). Systems ECIS, Copenhagen Business School,
Validation in IS Research: A State-of-the-Art Copenhagen, Denmark, 273-287.
Assessment. MIS Quarterly, 25(1), 1–16.
doi:10.2307/3250956 Kaufman, L., & Rousseeuw, P. J. (1990). Find-
ing Groups in Data: An Introduction to Cluster
Buonanno, G., Faverio, P., Pigni, F., Ravarini, A., Analysis. New York: John Wiley & Sons
Sciuto, D., & Tagliavini, M. (2005). Factors affect-
ing ERP system adoption: A comparative analysis Kim, Y., Lee, Z., & Gosain, S. (2005). Impedi-
between SMEs and large companies. Journal ments to successful ERP implementation process.
of Enterprise Information Management, 18(4), Business Process Management Journal, 11(2),
384–426. doi:10.1108/17410390510609572 158–170. doi:10.1108/14637150510591156
Chien, S.-W., & Tsaur, S.-M. (2007). Investigat- Ko, D. G., Kisrch, L. J., & King, W. R. (2005). An-
ing the Success of ERP Systems: Case Studies in tecedents of Knowledge Transfer from Consultant
Three Taiwanese High-Tech Industries. Comput- to Clients in Enterprise System Implementations.
ers in Industry, 58(8-9), 783–793. doi:10.1016/j. MIS Quarterly, 29(1), 59–85.
compind.2007.02.001
1056
Kositanurit, B., Ngwenyama, O., & Osei-Bryson, Muscatello, J. R., Small, M. H., & Chen, I. J.
K.-M. (2006). An Exploration of Factors that (2003). Implementing ERP in small and midsize
Impact Individual Performance in an ERP Envi- manufacturing firms. International Journal of Op-
ronment: An Analysis Using Multiple Analytical erations & Production Management, 23, 850–871.
Techniques. European Journal of Information doi:10.1108/01443570310486329
Systems, 15, 556–568. doi:10.1057/palgrave.
National Statistics 2001 area classification (2001).
ejis.3000654
Area classification for statistical wards. http://
Kremers, M., & van Dissel, H. (2000). ERP System www.statistics.gov.uk/about/methodology_by_
Migrations. Communications of the ACM, 43(4), theme/area_classification/wards/downloads/
53–56. doi:10.1145/332051.332072 area_classification_for_statistical_wards_meth-
ods.pdf, retrieved 2006-01-20.
Law, C. C. H., & Ngai, E. W. T. (2007). ERP
Systems Adoption: An Exploratory Study of the Nicolaou, A. I. (2004). ERP Systems Implemen-
Organizational Factors and Impacts of ERP Suc- tation: Drivers of Post-Implementation Success,
cess. Information & Management, 44, 418–432. Decision Support in an Uncertain and Complex
doi:10.1016/j.im.2007.03.004 World, The IFIP TC8/WG8.3 International Con-
ference, 589-597.
Loh, T. C., & Koh, S. C. L. (2004). Critical ele-
ments for a successful enterprise resource plan- Parr, A., & Shanks, G. (2000). A Taxonomy of ERP
ning implementation in small- and medium-sized Implementation Approaches. In Proceedings of the
enterprises. International Journal of Production 33rd Hawaii International Conference on System
Research, 42(17), 3433–3455. doi:10.1080/0020 Sciences HICSS, Maui, Hawaii, USA, 2424-2433.
7540410001671679
Peslak, A. R. (2006). Enterprise Resource Plan-
Lyytinen, K. (1988). Expectation failure concept ning Success. An Exploratory Study of the
and systems analysts view of information sys- Financial Executive Perspective. Industrial Man-
tems failures: Results of an exploratory study. agement & Data Systems, 106(9), 1288–1303.
Information & Management, 14(1), 45–56. doi:10.1108/02635570610712582
doi:10.1016/0378-7206(88)90066-3
Ranganathan, C., & Brown, C. V. (2006). ERP
Mabert, V. A., Soni, A., & Venkataramanan, M. Investments and the Market Value of Firms.
A. (2003). The impact of organization size on Information Systems Research, 17(2), 145–161.
enterprise resource planning (ERP) implementa- doi:10.1287/isre.1060.0084
tions in the US manufacturing sector. International
Raymond, L., & Uwizeyemungu, S. (2007). A pro-
Journal of Management Science, 31, 235–246.
file of ERP adoption in manufacturing SMEs. Jour-
McGinnis, T. C., & Huang, Z. (2007). Rethinking nal of Enterprise Information Management, 20(4),
ERP Success: A New Perspective from Knowl- 487–502. doi:10.1108/17410390710772731
edge Management and Continuous Improvement.
Sarkis, J., & Gunasekaran, A. (2003). Enterprise
Information & Management, 44(7), 626–634.
resource planning – modeling and analysis . Eu-
doi:10.1016/j.im.2007.05.006
ropean Journal of Operational Research, 146,
McNurlin, B. C., & Sprague, R. H., Jr. (2002). 229–232. doi:10.1016/S0377-2217(02)00545-3
Information Systems Management in Practice.
5th edition, Upper Saddle River.
1057
Schniederjans, M. J., & Kim, G. C. (2003). Imple- Soja, P. (2008). Difficulties in Enterprise System
menting Enterprise Resource Planning Systems Implementation in Emerging Economies: Insights
with Total Quality Control and Business Process from an Exploratory Field Study in Poland. In-
Reengineering . International Journal of Opera- formation Technology for Development. Special
tions & Production Management, 23(4), 418–429. Issue on Information Technology Investments in
doi:10.1108/01443570310467339 Emerging Economies, 14(1), 31–51.
Sedera, D., Gable, G., & Chan, T. (2003). ERP Soja, P., & Paliwoda-Pękosz, G. (2007). Towards
Success: Does Organization Size Matter? In the Causal Structure of Problems in Enterprise
Proceedings of the Pacific Asia Conference on System Adoption. In Proceedings of the 13th
Information Systems (PACIS), 10–13 July, Ad- Americas Conference on Information Systems,
elaide, South Australia, 1075-1088. Keystone/Colorado, USA.
Shang, S., & Seddon, P. B. (2000). A Compre- Stefanou, C. J. (2001). A framework for the
hensive Framework for Classifying Benefits of ex-ante evaluation of ERP software. European
ERP Systems. In Proceedings of the 6th Americas Journal of Information Systems, 10(4), 204–215.
Conference on Information Systems, Long Beach, doi:10.1057/palgrave.ejis.3000407
CA, USA, 1005-1014.
Stensrud, E., & Myrtveit, I. (2003). Identifying
Sharma, A., & Vyas, P. (2007). DSS (Decision High Performance ERP Projects. IEEE Transac-
Support Systems) in Indian Organised Retail tions on Software Engineering, 29(5), 398–416.
Sector, Indian Institute Of Management. doi:10.1109/TSE.2003.1199070
Soh, C., & Sia, S. K. (2005). The challenges of Sun, A. Y. T., Yazdani, A., & Overend, J. D. (2005).
implementing “vanilla” version of enterprise Achievement Assessment for Enterprise Resource
systems. MIS Quarterly Executive, 4(3), 373–384. Planning (ERP) System Implementations Based
on Critical Success Factors (CSFs). International
Soja, P. (2004). Success Factors in ERP Systems
Journal of Production Economics, 98, 189–203.
Implementations. Result of research on the Polish
doi:10.1016/j.ijpe.2004.05.013
ERP market. In Proceedings of the 10th Americas
Conference on Information Systems AMCIS, New The Commission of the European Community.
York, USA, 3914-3922. (1996). (96/280/EC) Commission recommenda-
tion of 3 April 1996 concerning the definition of
Soja, P. (2005). The Impact of ERP Implementa-
small and medium-sized enterprises. In Official
tion on the Enterprise – an Empirical Study. In
Journal No. L 107 30/04/1996, pp.4-9.
Proceedings of the 8th International Conference
on Business Information Systems, Poznan, Poland, Themistocleous, M., Irani, Z., O’Keefe, R. M.,
389-402. & Paul, R. (2001). ERP Problems and Applica-
tion Integration Issues: An Empirical Survey. In
Soja, P. (2006). Success factors in ERP systems
Proceedings of the 34th Hawaii International
implementations: lessons from practice. Journal
Conference on System Sciences.
of Enterprise Information Management, 19(4),
418–433. doi:10.1108/17410390610678331 Tinham, B. (2006). Your Guide to Choosing and
Implementing ERP. Manufacturing Computer
Solutions.
1058
Wright, S., & Wright, A. M. (2002). Informa- Wu, J.-H., & Wang, Y.-M. (2007). Measuring
tion System Assurance for Enterprise Resource ERP Success: The Key-Users’ Viewpoint of the
Planning Systems: Unique Risk Considerations. ERP to Produce a Viable IS in the Organization.
Journal of Information Systems, 16(Supplement), Computers in Human Behavior, 23, 1582–1596.
99–113. doi:10.2308/jis.2002.16.s-1.99 doi:10.1016/j.chb.2005.07.005
This work was previously published in Global Implications of Modern Enterprise Information Systems: Technologies
and Applications, edited by Angappa Gunasekaran, pp. 114-136, copyright 2009 by Information Science Reference (an
1059
1060
Chapter 4.12
Challenges and Solutions
for Complex Business
Process Management
Minhong Wang
The University of Hong Kong, Hong Kong
Kuldeep Kumar
Florida International University, USA
AbstrAct oriented architecture have been applied in modeling

and developing complex systems. However, there is
A business process displays complexity as a result considerable ambiguity involved in differentiating
of multiple interactions of its internal components between these overlapping technologies and their
and interaction between the process and its environ- use in developing BPM systems. No explicit link-
ment. To manage complexity and foster flexibility age has been established between the requirement
of business process management (BPM), we present of complex BPM and the supporting technologies.
the DCAR architecture for developing complex In this study, we use the DCAR architecture as the
BPM systems, which includes decomposition of foundation to identify the BPM requirements for
complex processes (D); coordination of interactive employing technologies in developing BPM sys-
activities (C); awareness of dynamic environments tems. Based on an examination of the both sides
(A); and resource selection and coordination (R). (BPM requirements and supporting technologies),
On the other hand, computing technologies, such as we present a clear picture of business process
object-oriented programming, component-based de- complexity with a systemic approach for develop-
velopment, agent-oriented computing, and service- ing complex BPM systems by using appropriate
computing technologies.
DOI: 10.4018/978-1-60566-669-3.ch001
Challenges and Solutions for Complex Business Process Management
INtrODUctION foster flexibility of complex systems, modularity

is the key to the solution (Baldwin et al., 1997;
Businesses around the world are paying more Simon, 1981). Modularity in BPM requires decom-
attention to process management and process posing a complex BPM system into a number of
automation to improve organizational efficiency interacting components that perform the processes.
and effectiveness. It is increasingly common to Based on the investigation of business process
describe organizations as sets of business pro- complexity and modularity theory, we present
cesses that can be improved by business process the DCAR architecture for developing complex
management (BPM). Most approaches to BPM BPM systems, which include decomposition of
have used information technologies to support or complex processes (D); coordination of interactive
automate business processes, in whole or in part, activities (C); awareness of dynamic environments
by providing computer-based systems support. (A); and resource selection and coordination (R).
These technology-based systems help coordinate On the other hand, various modular computing
and streamline business transactions, reduce op- technologies, such as Object-Oriented Program-
erational costs, and promote real-time visibility ming (OOP); Component-Based Development
in business performance. (CBD); Agent-Oriented Computing (AOC);
Traditional approaches to building and imple- and Service-Oriented Architecture (SOA); have
menting BPM systems use workflow technolo- emerged to model and develop complex systems.
gies to design and control the business process. There has been a proliferation of studies about
Workflow-based systems follow highly structured the application of these modular technologies in
and predefined workflow models, and are well developing BPM systems (Weske, 1998; Kammer
suited to applications with stable inputs, processes, et al., 2000; Jennings et al., 2002; Wang et al.,
and outputs. Contemporary business processes are 2005b; Leymann et al., 2002). As the modular
becoming increasingly complex and dynamic as computing paradigms and technologies become
they seek to cope with a wide range of internal popular, researchers often attempt to employ and
and external interactions and changes. To provide integrate them in creating business process man-
sufficient flexibility and adaptability in BPM, a agement solutions. However, there is considerable
number of researchers have been investigating the ambiguity involved in differentiating between
approaches and techniques for developing BPM these overlapping terminologies and consequently
systems for an increasingly turbulent environment their use for BPM systems development. The
(Casati et al., 1999; Chiu et al., 1999; Weske, 2001; fundamental questions about the use of these
Wang et al., 2002, 2005a; K. Kumar et al., 2006). technologies, i.e., why we need to use them for
Most studies have focused on present process solutions of BPM, how we apply them, and how
structures and provide rapid response to changes we integrate them with other solutions, remain
that lead to temporary and short term fluctuations unexamined. Most research on technology sup-
in the organization’s activities. port for BPM is experience-driven, ad-hoc, and
In this study, we view business process as a often lacks a systematic analysis of the rationale
complex system that adapts to continuously chang- for the technology support (Wang et al., 2008a).
ing and unpredictable environments in order to Little work has examined the root of complexity
survive. A business process displays complexity of business processes, the need for effective ap-
because of multiple interactions of its internal proaches to BPM, and how this need affects the
components and interaction between the process technology solutions for process management
and its environment. To manage complexity and (K. Kumar et al., 2006). In this study, we analyze
1061
the differences and relationships between these business Process Management

overlapping terminologies and techniques, and
match them to BPM requirements. The DCAR Business Process Management (BPM) refers to
architecture we proposed for complex BPM is activities performed by organizations to design
used as the foundation to identify the BPM require- (capture processes and document their design in
ments for employing these modular computing terms of process maps), model (define business
technologies in developing BPM systems. Based processes in a computer language), execute (de-
on an examination of both sides (BPM require- velop software that enables the process), moni-
ments and supporting technologies), we present a tor (track individual processes for performance
clear picture of business process complexity with measurement), and optimize (retrieve process
a systemic approach on how these technologies can performance for improvement) operational busi-
be applied and integrated in developing systems ness processes by using a combination of models,
for complex process management. methods, techniques, and tools (van der Aalst et al.,
2002; Melão et al., 2000). Process design in turn
includes the differentiation or subdivision of the
rEsEArcH bAcKGrOUND process into underlying tasks or activities, process-
configuration (arranging the process tasks into a
business Process logical sequence to produce the process outcome),
and selection and allocation of specific actors
A business process can be simply defined as and resources to particular process-tasks. The
a collection of activities that create value by monitoring and control phases can be an extension
transforming inputs into more valuable outputs to provide feedback into the continuing design,
(Hammer et al., 1993). These activities consist of implementation, and execution cycle. Among vari-
a series of steps performed by actors to produce a ous approaches or methods for BPM, this study
product or service for the customer. In more de- focused on the use of information technologies to
tails, a business process is typically a coordinated support or automate business processes, in whole
and logically sequenced set of work activities or in part, by providing computer-based systems
and associated resources that produce something support. These technology-based systems help
of value to a customer (El Sawy, 2001). Each coordinate and streamline business transactions,
process has an identified customer; it is initiated reduce operational costs, and promote real-time
by a process trigger or a business event (usually visibility in business performance.
a request for product or service arriving from Traditional approaches to building and imple-
the process customer); and it produces a process menting BPM systems use workflow technologies
outcome (the product or a service requested by to design and control the business process (van
the customer) as its deliverable to the process der Aalst et al., 2002). Workflow-based systems
customer. Given the scope and variety of actions follow highly structured and predefined work-
that are needed to produce the product or service, flow models, and are well suited to applications
the process is differentiated (sub-divided) into a with stable inputs, processes, and outputs. Con-
set of tasks or activities. These tasks are assigned temporary business processes are complex and
to and performed by actors (either machines or dynamic. They evolve and change over time as
humans), using resources such as work-stations; a result of complex interactions, resource com-
machines; raw materials; and supplies that are petition, breakdowns and abnormal events, and
available to the actors. other sources of uncertainty. Realizing the need
1062
to provide sufficient flexibility and adaptability gests. Business processes are best viewed as
in BPM, many researchers are investigating the networks, in which a number of actors collabo-
approaches and techniques for developing BPM rate and interact to achieve a business goal. A
systems for an increasingly turbulent environment. business process displays complexity because of
In perusing this, most studies have focused on multiple interactions of its internal components
the capabilities that are based on present process and interaction between the process and its en-
structures of an organization and provide rapid vironment (Melão et al., 2000). In recent years,
response to changes that lead to temporary and business environments have been changing from
short term fluctuations in the organization’s ac- centralized-and-closed to distributed-and-open.
tivities. Efforts to improve flexibility in business Business processes are becoming increasingly
processes can be found in numerous studies on complex and dynamic as they seek to cope with
adaptive workflow/process modeling, workflow/ a wide range of internal and external interactions
process monitoring, and exception management and changes. In this situation, BPM should be
(Casati et al., 1999; Chiu et al., 1999; Weske, 2001; able to manage a number of components and their
Wang et al., 2002, 2005a; K. Kumar et al., 2006). complex interactions in business processes, in par-
On the other hand, while faced with revolutionary ticular in continuously changing and interplayed
changes, an organization needs more flexibility environment (Wang et al., 2006a; K. Kumar et al.,
to rearrange its internal or intra-organizational 2006). Furthermore, attention should be paid to
structures and processes. Efforts can be found situations where dynamic collaboration and soft-
in research on business process analysis and connection between business partners is playing
redesign, and business process reconstruction an increasingly important role (K. Kumar, 2001;
or reengineering, which enable new forms of Wang et al., 2008b). This needs both informa-
working and collaborating within an organization technologies and managerial capabilities to
tion or in cross-border businesses (Hammer et adapt the organization structure and its decision-
al., 1993). With the recent growth of electronic making and communication processes in order
services, a business process can be dynamically to facilitate cross-hierarchical, cross-functional,
established by connecting or composing appropri- cross-product/service, and cross-market capability
ate services based on market demand (A. Kumar development.
et al., 2002; Petrie et al., 2003; K. Kumar et al., In this context, we need to shift from the
2007). A worldwide network of organizations can mechanistic view of workflow paradigm which
be formed through process composition over the focuses on static and structure features of business
Internet (Casati et al., 2001). process. Instead, business processes can be viewed
as complex dynamic systems that adapt to continu-
ously changing and unpredictable environments
cHALLENGEs OF bUsINEss in order to survive. This requires the integration
PrOcEss MANAGEMENt of organizational, managerial, and technological
issues in understanding and managing business
business requirement processes. In addition to a set of logically related
tasks, more aspects need to be taken into account
To business, dealing with changes is a fact of such as environment awareness; knowledge for
everyday life that must be exploited. Real-world process management; flexible resource coordina-
processes are often much messier than what the tion; and so on.
typical input-transformation-output view sug-
1063
technology support MODULArIZAtION FOr

cOMPLEX systEMs
To deal with a complex system like business
process management, software technologies As discussed, business processes display complex-
have been on constant evolution. Structured or ity due to interactions of their internal components
function-oriented analysis was used in 1970s for and interaction of the process with its environment.
functional decomposition of more stable sys- Business processes are complex systems that are
tems, concealing the details of an algorithm in a made up of a number of interacting objects with dy-
function. To deal with constant changes related namic behavior. Alexander and Simon addressed
to data structure, object-oriented programming the theories about how we design complex sys-
(OOP) was addressed to separate data from the tems; to design a complex structure, one powerful
applications, where data and its corresponding technique is to decompose it into semi-independent
operations were encapsulated within an object. and interrelated components, which in turn have
Based on this method, traditional workflow sys- their own components in (Simon, 1981). Though
tems have been developed by taking processes or they did not use the word “modularity”, the con-
workflows out of the applications to improve the cept was central to their thinking. Baldwin and
control and change of business processes (Weske, Clark (1997) addressed modularity as a particular
1998). More recently, to deal with more complex design structure, which refers to development
and frequent changes of business processes, new of a complex product or process from smaller
computing technologies have emerged in BPM, subsystems that can be designed independently.
such as Component-Based Development (CBD) It is possible for us to view all entities – social,
in Kammer et al. (2000); Agent-Oriented Com- biological, technological, or otherwise – as hier-
puting (AOC) in Jennings et al. (2002) and Wang archically nested system. Modularity is the key to
et al. (2005b); and Service-Oriented Architecture managing complexity and fostering flexibility of
(SOA) in Leymann et al. (2002). complex systems (Baldwin et al., 1997; Simon,
However, the fundamental questions about the 1981). It ensures easy maintenance and updates
use of these technologies, i.e., why we need to use of complex systems by reducing the interactions
them for solutions of BPM, how we apply them, among the components. A complex system can
and how we integrate them with other solutions, be managed by separating the high-frequency
remain unexamined. Most research on technology intra-module linkages from the low-frequency
support for BPM is experience-driven, ad-hoc, and inter-module linkages, and limiting the scope
often lacks a systematic analysis of the rationale of interactions between modules by hiding the
for the technology support (Wang et al., 2008a). intra-module relations inside a module box. A
There is only minimal work that examines the module in a complex system is a unit whose struc-
root of complexity of business processes, the need tural elements are powerfully connected among
for effective approaches to BPM, and how this themselves and relatively weakly connected to
need affects the technology solutions for process elements in other units. The guiding principle
management (K. Kumar et al., 2006). for decomposition is that intra-module cohesion
is generally stronger than inter-module coupling.
This fact can be used to distinguish diverse in-
teractions, and deal with them by encapsulating
the intra-module interactions inside a module box
(Wolters, 2002). Using modularization, a complex
1064
structure can be decomposed into sub-functions, that are visible and those that are hidden. Hidden
sub-processes, sub-areas, and in other ways. Al- components lying inside a black box (the task
ternative decompositions correspond to different module) are isolated from other parts and are
ways of dividing the responsibilities. allowed to vary without changing much in other
While pursuing modularization, choosing right parts. This provides flexible process structures by
granularity of the component is an important issue permitting uncertainty as well as accommodating
in decomposing complex systems. The identity of changes.
any unit as the module is not fixed, but determined
by the level of analysis we choose (Schilling,
2003). Granularity is the size of the unit under DcAr ArcHItEctUrE FOr
consideration. It refers to the degree to which a cOMPLEX PrOcEss MANAGEMENt
system can be separated and reorganized. For ex-
ample, a company is generally divided into depart- Based on the above investigation of BPM chal-
ments, the departments into groups, and the groups lenges (3rd section) and the modularity theory for
into employees. Systems of large components managing complexity (4th section), we propose the
are called coarse-grained, and systems of small DCAR architecture for developing BPM systems,
components are called fine-grained. Granularity which includes decomposition of complex pro-
is a relative concept that can be precisely defined cesses (D); coordination of interactive activities
only in a specific context (Elfatatry, 2007). For (C); awareness of dynamic environments (A); and
instance, if a component implements all functions resource selection and coordination (R).
of a banking system, then it can be considered
coarse grained; if it supports just credit-balance Decomposition of complex
checking, it is considered fine-grained. The greater Processes (D)
the granularity, the more the flexibility, but the
greater the overheads of synchronization and According to the modular architecture, a business
communication. To maximize the performance process can be decomposed into tasks, task into
of modularization, a complex system must be sub-tasks, and so on. Interactions between sub-tasks
analyzed and implemented by choosing right within a task are often encapsulated within the task;
granularity of the components. A tradeoff between interactions between tasks are encapsulated within
modularity and integrity should be considered their higher-level process or task. This raises the
to ensure overall system performance (Garud et issue on how we decompose complex processes.
al., 2003). Traditional workflow approaches have selected
With respect to business process management, “task” as the basic module for building process
modularity supports two things. First, modularity management systems. A business process can be
increases the range of “manageable” complexity. decomposed into a number of semi-dependent and
It does this by limiting the scope of interactions interrelated tasks. These tasks are then linked to each
between various components by encapsulating other in a pre-established and usually sequential
some components (such as data, resources, knowl- inter-relationship or dependency. With the extension
edge, and responsibility of a specific task) and of business processes from intra-organizational to
their interactions inside the task module, thereby inter-organizational scope, we need to deal with
reducing the amount and range of interweaving interactions within an organization as well as
that occurs in an interconnected process. Second, interactions across organizations. Moreover, the
modularity accommodates uncertainty. Through complexity of BPM is increased by interweaving
modularity, components are partitioned into those of inter- and intra-organizational interactions.
1065
To manage the complexity, we need to dis- requirements: 1) differentiation, or the division

tinguish between inter- and intra-organizational of work into tasks to be performed by various
interactions and deal with them by isolating one actors; and 2) integration, that is, the coordina-
type from another. We propose “service” as a tion of these tasks to accomplish the goals of the
high level view of the building block of a process, activity (Mintzberg, 1979). After decomposing a
where a process is composed of a set of services. complex process into a number of task compo-
Each service is provided by a corresponding actor nents, we need to coordinate various interactions
(organization, individual, or computer program), between the components at different levels in a
and can be further decomposed into sub-tasks. For network hierarchy. In the context of a hierarchy,
example, a complex supply chain management a component can be involved in vertical interac-
process can be decomposed into “customer order tions with its subordinates and super-ordinates,
service”, “procurement service”, “manufacturing and in horizontal interactions with its peers. A
service”, and “transportation service”; each service component in a complex system, no matter how
can be provided by a corresponding organizational large or small, may interact with a limited set of
actor, and can be further decomposed throughout superiors, inferiors and coordinate peers (Simon,
several layers as shown in Figure 1. 1981).
In a complex system, the components interact,
coordination of Interactive represented as coupling or dependency between
Activities (c) the components. Based on degree of strength of
dependency, the components in a complex system
To manage complex interactions in complex pro- are loosely or tightly coupled (Simon, 1977). Based
cesses, multiple actors, activities, resources, and on the relationship between the components in their
goals need to be coordinated. Every organized interactions, the components in a complex sys-
human activity gives rise to two fundamental tem can be centrally or de-centrally coordinated.
Figure 1. Decomposition of a supply chain process
1066
Take the example of supply chain management in Awareness of turbulent

Figure 1. In a loosely coupled context of “supply business Environment (A)
chain management”, the participating components
include “customer order service”, “procurement As a result of complex interactions, resource
service”, “manufacturing service”, and “trans- competition, abnormal events, and other sources
portation service”. Each service is regarded as an of uncertainty, business processes continuously
autonomous entity, managing its own activities evolve and change over time. A complex process
as well as the interaction with other services. is usually semi-structured or unstructured to the
Decentralized coordination or mutual negotiation extent that there is an absence of routine proce-
can be suggested to govern the supply chain if no dures for dealing with it. In such situations, BPM
formal centralized governance authority exists. In solutions do not depend on providing the computer
a tightly coupled context of “order process”, the system with exact details about how to accomplish
participating components including “customer a process; but provide the system with guidelines
login”, “order input”, “order confirmation”, and to help it determine how to deal with the process.
“order submission” are highly interdependent In other words, problem solving is regarded as an
and mixed, and bundled into a single integrated interaction between the behaving organism and
package. Centralized coordination can be used if the environment under the guidance of a control
there is a central authority granted the power to system. Information and data are input to this
govern the order process. system, represented in its memory as declarative
Mintzberg further suggests that environmental knowledge, and then used in problem solving
uncertainty is an important determinant of the following algorithmic or heuristic steps (Wang
mode for interactions and coordination. The more et al., 2006a).
stable and predictable the situation, the greater A complex system interacts with and adapts
the reliance on coordination based on structured to its environment for survival, in addition to in-
and specifiable schedules, such as coordination teractions of internal components. Any adaptive
by plan and coordination by standardization. The system must develop correlations between goals
more variable and unpredictable the situation, and actions in the world of process. This requires
the greater will be the reliance on informal and continuous translation between the state and
flexible communication, such as coordination by process and discovery of a sequence of processes
feedback and coordination by mutual adjustment that will produce the goal state from an initial
(K. Kumar et al., 1996, 2007). When faced with state based on means-ends analysis (Holland,
increased uncertainties in dynamic environments, 1995; Simon, 2003). A basic idea underlying is
organizations need to use more flexible coordi- the control of complex dynamic systems or situ-
nation mechanisms to coordinate their business ations based on situation awareness. Awareness,
activities. Flexible coordination is portrayed by according to biological psychology, is a human’s
more bottom-up initiatives and less centralization or an animal’s perception and cognitive reaction
of decision-making in the top. This requires flatter to a condition or event. Situation awareness is the
hierarchies, decentralized autonomy-based units, perception and understanding of objects, events,
and decision-based coordination, which in turn people, system states, interactions, environmental
reduces direct hierarchical control and encour- conditions, and other situation-specific factors in
ages greater mutual adjustment and coordina- complex and dynamic environments (Endsley,
tion between the work-units (Mintzberg, 1979; 1995). Situation awareness underpins real-time
Volberda, 1999). reactions to environmental changes. In terms of
1067
cognitive psychology, situation awareness refers management is an important and complicated

to the active content of a decision-maker’s mental issue to the efficiency and effectiveness of busi-
model; its purpose is to enable rapid and appropri- ness processes or workflows. In addition to task,
ate decisions and effective actions. structure, and procedure, the resource aspects of
In a dynamic business process environment, business process should be taken into account.
an exact execution order of activities is imprac- However the traditional process model or defini-
tical; the interaction or relationship between the tion does not include the resource concept. Though
environment and activities is more appropriate some common principles for resource allocation
in determining how to manage and coordinate in workflow management (e.g., first-in first out,
activities (Wang et al., 2006a). The dynamicism shortest processing time, and earliest due date)
therefore requires spontaneous decisions and coor- have been recommended (van der Aalst et al.,
dination of processes based on situation awareness. 2002), the research on resource management, in
In this context, BPM should be able to coordinate particular on process flexibility through resource
the processes by sensing and comprehending the selection and coordination in business processes
situation, determining responses to it, while at is far from sufficiency.
the same time, taking actions to work towards In today’s business environment, business net-
business goals. In other words, the question of works of resources and actors can be temporarily
which task to execute and when to execute it is assembled, integrated, and driven by demands that
dependent on the current environment and under- emerge and operate for the lifespan of the market
lying business rules rather than a static process opportunity (K. Kumar, 2001). In this conception,
schema. To achieve this, knowledge or rules a firm is not considered as a black box guided
for process management has become important by the strategist, but as a bundle of firm-specific
foundation for BPM in dynamic environment. The resources of use for specific tasks. Along with this,
modular system architecture can greatly improve new business models have accordingly come into
the ability to identify and leverage knowledge for view, such as demand chain; virtual enterprise;
coordination in business processes (Sanchez et al., and electronic marketplace. They allow companies
2003). Through modularization, the knowledge to operate in dynamically changing environments
becomes embodied in specific BPM components, by quickly and accurately evaluating new market
which helps an organization to discover and focus opportunities or new products. The companies
on opportunities for organizational learning and may coordinate with potential partners in demand-
capability development in BPM. driven and resource-based soft connections that are
made for the duration of the market opportunity.
resource selection and As a result, a business process can be dy-
coordination (r) namically established at run-time by connecting
or composing several services together from
Business processes require actors and resources different organizations through alliances, partner-
to perform the tasks. These actors and their as- ships, or joint ventures. In this situation, attention
sociated resources may reside either within the on business processes should go beyond task and
organization, or in the case of inter-organizational procedure, and extend to other elements, which
processes, across a network of multiple organi- include resources discovery; selection; integra-
zations. In some situations, more than one actor tion; and coordination.
may be qualified and available for performing a What is new in this business process model
task; while in some others, more than one task is reliance on the idea of separating resource re-
requires the same actor and resource. Resource quirements from concrete satisfiers (Mowshowitz,
1068
1997). This separation allows for crafting process object oriented computing technologies to BPM
structures that enable switching between different (Jennings et al., 2002; Wang et al., 2005b; Ley-
resources options for implementing a process, as mann et al., 2002; Kammer et al., 2000; Weske,
shown in Figure 2. It creates an environment in 1998). The aspiration of these modular comput-
which the means for reaching a goal are evalu- ing technologies have facilitated the modeling
ated and selected for optimized performance. The of process architecture by modular software
success of the model is highly dependent on the architectures (object, component, Web service,
match between the requirements and satisfiers and software agent), thereby creating analogs of
that deliver the services. One way to ensure this the business process in software. Each paradigm
balance is to model the integration or composition represents a philosophy of perception, abstraction,
of business processes as a management problem and decomposition of complex systems in order
which involves: 1) the separation of requirements to deal with changes. As the modular computing
from the means for realization, and 2) the dynamic paradigms and technologies become popular,
selection and allocation of available resources to researchers often attempt to employ and integrate
requirements. them in creating business process management
solutions.
However, different modular technologies have
INVEstIGAtION OF MODULAr different philosophical beliefs with respect to
cOMPUtING tEcHNOLOGIEs how a complex system should be decomposed in
order to tackle changes (Elfatatry, 2007). These
There has been a proliferation of studies about technologies are usually applied in BPM without
the application of agent, service, component, and identifying the real rationale for their use in BPM
Figure 2. Resource selection and coordination in business processes
1069
scenarios. At present there is considerable ambigu- component-based

ity in differentiating between these overlapping Development (cbD)
terminologies and consequently their use for BPM
systems development. For example, we often Component-Based Development (CBD) is another
hear people discussing their proposed solutions branch of the software engineering discipline, with
as agent-based systems whereas they may just be an emphasis on decomposition of the engineered
simply using object abstraction. Furthermore, the systems into functional or logical components
common-sense understanding of these concepts with well-defined interfaces used for communi-
and technologies does not easily map onto each cation across the components. CBD includes a
other. Unless we have clarity on these terminolo- component model and an interface model. The
gies and the way how to use them, the application component model specifies for each component
and integration of these techniques is likely be how the component behaves in an arbitrary en-
problematic. vironment, and the interface model specifies for
In this section we outline the four concepts: each component how the component interacts
Agents and Agent-Oriented Computing; Services with its environment (Szyperski, 2002).
and Services-Oriented Architecture; Objects and OOP vs. CBD. A component is a small group
Object-Oriented Programming; and Components of objects working together to provide a system
and Component-Based Development. The purpose function. It can be viewed as a black box at the
is not to re-define these concepts and approaches, level of a large system function. At a fine level
but to clarify their similarities and differences, in of granularity, we use objects to hide behavior
particular towards their abstraction and decom- and data. At a coarser level of granularity, we
position techniques, and to make explicit some use components to do the same. OOP focuses
of the underlying assumptions inherent in the use on the encapsulation of both data and behavior
of the terminology. of an object. CBD goes further by supporting
public interfaces used for communication across
Object-Oriented Programming (OOP) the components.
Object-Oriented (OO) technologies provide rich
Object-Oriented Programming (OOP) is a soft- models to describe problem domains, however it
ware engineering paradigm that uses “objects” is not enough to adapt to changing requirements
and their interactions to design applications and of real-world software systems (Elfatatry, 2007).
computer programs (Rumbaugh, 1991). A program OOP assumes that it is possible to identify and solve
is seen as a collection of cooperating objects, as almost all problems before coding, while CBD
opposed to a traditional view in which a program and later SOA and AOC adopt a more pragmatic
is seen as a list of instructions to the computer. approach that believes business system develop-
The OOP paradigm addressed issues of reuse ment is an incremental process and changes are an
and maintenance by encapsulating data and its inescapable aspect of software design. Specifically,
corresponding operations within an object class. objects were too fine-grained and did not make a
To change a data structure, it is often necessary to clear separation between computational and com-
change all the functions related to the data struc- positional aspects. Components were then proposed
ture. OOP was deployed as an attempt to promote to encapsulate the computational details of a set of
greater flexibility and maintainability, since the objects. Using CBD, software development can
concepts of objects in the problem domain has a be improved since applications can be quickly as-
higher chance of being stable than functions and sembled from a large collection of prefabricated
data structures. and interoperable software components.
1070
Components inherit much of the character- by providing flexibility through service-based

istics of objects in the OO paradigm. But the abstraction of organizational applications.
component notion goes further by separating the SOA vs. CBD. Compared with a component, a
interface from the component model. OO reuse service is relatively coarse-grained which should
usually means reuse of class libraries in a particu- able to encapsulate more details. SOA is an ex-
lar OO programming language or environment. tension of earlier OOP and CBD concepts. CBD
For example, to be able to reuse a SmallTalk supports more close system architecture where the
or Java class in OO, you have to be conversant exact source of the required functionality and com-
with SmallTalk or Java language. A component, munication is predetermined. Propriety standards
by using public interface, can be reused without and implementation dependent specification of
even knowing which programming language or components have hindered CBD from achieving
platform it uses internally. its primary goal facilitating reuse. The point in
SOA is the service specification rather than the
service-Oriented Architecture (sOA) implementation. SOA focuses on the user’s view
of a computing object or application, i.e., the
A service is defined as an act or performance that services that are provided and the metadata that
one party can offer to another that is essentially define how the services behave. In SOA, a service
intangible and does not result in the ownership of has a published network-addressable interface.
anything. Its production may or may not be tied A published interface is exposed to the network
to a physical product. and may not be changed so easily, because the
A Web Service, defined by W3C (World Wide clients of the published interface are not known.
Web Consortium), a software application identi- The difference between component and service
fied by a URI (Uniform Resource Identifier), in the interface is analogous to an intranet-based
whose interfaces and bindings are capable of be- site only accessible by employees of the company
ing defined, described, and discovered by XML, and an Internet site accessible by anyone.
and which supports direct interactions with other A service does not define any structural con-
software applications using XML-based messages straints for loose coupling of services over the
via Internet-based protocols. Web services are Internet. CBD architectures, on the other hand,
self-contained and modular business applications represent a case of tight coupling. For example,
based on open standards (Leymann et al., 2002). in CORBA, there is a tight coupling between the
They can share information using standardized client and the server as both must share the same
communication protocols and ask each other to interface with a stub on the client-side and the
do something, i.e., ask for service. corresponding skeleton on the server side. Com-
Service-Oriented Architecture (SOA) utilizes posing a system from a number of components
Web services as fundamental elements for devel- is relatively controlled compared to dynamic
oping applications. It is an emerging paradigm for service composition. Moreover, CBD assumes
architecting and implementing business collabora- early binding of components, i.e., the caller unit
tions within and across organizational boundaries. knows exactly which component to contact before
SOA enables seamless and flexible integration of runtime. SOA adopts a more flexible approach
Web services or applications over the Internet. It where the binding is deferred to runtime, enabling
supports universal interoperability and location the change of source of provision each time.
transparency. SOA reduces complexity of business Most significantly, the idea of the service model
applications in large-scale and open environments differs from CBD by the fact that SOA supports
1071
the logical separation of service need from service (service providing party). Acting as an intermedi-
fulfillment. Delivering software as a service brings ary between two or more parties in negotiating
about new business models and opportunities of agreements, brokers use appropriate mediating
software development and service provision in techniques or processes to improve the dialogue
demand-driven dynamic environment. between the parties, aiming to help them reach
an agreement. Normally, all parties must view the
Agent-Oriented computing (AOc) mediator as neutral or impartial.
Autonomy is the power or right of self-gov-
Recently the Agent-Oriented Computing paradigm ernment. It refers to the capacity of a rational
has gained popularity among researchers who at- individual to make an informed, uncoerced
tempt to develop complex systems for business decision. “Autonomous” means that the actor is
process management (Jennings et al., 2002; Wang independent, i.e., the actor can decide what to do
et al., 2005b). Terms such as “autonomous agent” and how to do it. An autonomous agent therefore
and “agency” are now commonly used in computer is a system situated in, and part of, an environ-
science literature. On the other hand, a rich body of ment, which senses the environment, and acts on
literature on the concept of Agency and the role of it, over time, in pursuit of its agenda as derived
agents already exists in the institutional econom- from its principal. As an agent acts on behalf of the
ics and business field. We attempt to reconcile the principal, the agent cannot be fully autonomous.
various terms from the two research traditions. The principal may give the agent different levels
Actor is someone who performs an act, i.e., of choice in performing the task. For example,
does something. An actor may be a person, an the principal can tell the agent what to do, but
organizational unit, or a computer program. An leave it to the agent to decide as to how to do it.
actor may be completely autonomous, that is, it Software Agent. In computer science, the term
acts of its own volition. If the actor is authorized “agent” is used to describe a piece of software or
to do something on behalf of someone else, the code that acts on behalf of a human user or an-
actor is an “agent” of the other party. other program in a relationship of agency. It may
Agent is an actor (performer) who acts on the denote a software-based entity that could enjoy
behalf of a principal by performing a service. the some properties of autonomy (agents operate
The agent provides the service when it receives without the direct intervention of its principal
a request for service from the principal. The humans), social ability (agents communicate with
principal-agent relationship is found in most em- other agents), reactivity (agents perceive their
ployer-employee relationships. A classic example environment and respond to changes in a timely
of agency relationship occurs when stockholders fashion), and pro-activity (agents do not simply
hire top executives to run the corporation on their act in response to their environment, but are able
behalf. To manage the relationship between a to exhibit goal-directed behavior by taking some
principle and an agent of the principle, agency initiative) (Jennings et al., 2002). The agent-based
theory is concerned with various mechanisms used computing paradigm is devised to help computers
for aligning the interests of the agent with those know what to do, solve problems on behalf of hu-
of the principal such as piece rates/commissions man beings, and support co-operative working.
and profit sharing (Eisenhardt, 1989). The behavior of software agents is empowered by
Broker is a special type of agent that acts on human and implemented by software.
behalf of two symmetrical parties or principals – Agent-Oriented Computing (AOC) is based on
the buyer and seller. A broker mediates between the idea of delegating tasks and responsibility of a
the buyer (service requesting party) and the seller complex problem to a group of software agents. It
1072
emphasizes autonomy and mutual co-operation of properties for solving complex problems, while
agents in performing tasks in open and complex Web services are characterized by their open ac-
environments. A complex system can be viewed cess standards and protocols over the Internet.
as a network of agents acting concurrently, each While a Web service may only know about itself,
finding itself in an environment produced by its agents often have awareness of other agents and
interactions with the other agents in the system. their capabilities as interactions occur among the
AOC is used to model and implement intelligent agents. Agents are inherently communicative,
solutions to semi- or ill-structured problems, which whereas Web services are passive until invoked.
are too complex to be completely characterized Agents cooperate autonomously and flexibly, and
and precisely described. AOC offers a natural by forming teams and coalitions can assemble
way to view and describe systems as individual higher-level and more comprehensive services.
problem-solving agents pursuing high-level goals However, current standards or languages for Web
defined by their principals. It represents an emerg- services do not provide for flexible composing
ing computing paradigm that helps understand and functionalities, such as brokering and negotiation
model complex real-world problems and systems, in e-marketplaces (Huhns, 2002). Web services
by concentrating on high-level abstractions of are inherently less autonomous and independent
autonomous entities (Wooldridge et al., 1999). than software agents.
AOC vs. OOP. From a software engineering Against this background, there is a movement
point of view, Object-Oriented methodologies towards combining the concept of Web services
provide a solid foundation for Agent-Oriented with software agents. W3C introduced a concept,
modeling. AOC can be viewed as a specialization where software agents are to be treated as the
of OOP. OOP proposes viewing a computational foundation for Web services architecture - “A
system as made up of modules that are able to Web service is an abstract notion that must be
communicate with one another. AOC specializes implemented by a concrete agent”. AOC may take
the framework by representing the mental states SOA into new dimensions to model autonomous
and rich interactions of the modules (agents). and heterogeneous components in uncertain and
While objects emphasize passive behavior (i.e., dynamic environments. The integration of Web
they are invoked in response to a message), agents services with software agents can function as
support more autonomous behavior, which can computational mechanism in their own right,
be achieved by specifying a number of rules for thus significantly enhancing the ability to model
interpreting the environmental states and knowl- and construct complex software systems. It will
edge for governing multiple degrees of freedom be a promising computing paradigm for efficient
of activities. In relation to this, mechanisms of enterprise service selection and integration.
knowledge acquisition, modeling, and mainte-
nance have become important foundation for reconciling OOP, cbD,
building of autonomous agents. sOA, and AOc
AOC vs. SOA. Software agent is a software-
based entity that enjoys the properties of autonomy, From OOP and CBD to SOA and AOC, the
social ability, reactivity, and pro-activity. Web practice of software programming has evolved
service is a software application in the Web based through different development paradigms. At
on open standards. Though both of them are com- the conceptual level, these concepts and ap-
puter applications that perform tasks on behalf of proaches are complementary and built upon each
principals (human beings or other programs), the other; all have a role to play in designing and
focus of software agents is on their autonomous managing software systems. Each method shift
1073
came about in part to deal with greater levels services. This special type of agent works as
of software complexity. In all cases, the way to an intermediary between service requesters and
manage complexity is by decomposing a complex service providers, coordinating on behalf of the
system or process into smaller modules that can two parties by taking into account of service re-
be designed independently, i.e., modularization. quirements, qualities, costs, constraints, and so on.
Modularization ensures easy maintenance and
updates of complex systems by separating the
high-frequency intra-module linkages from the APPLyING MODULAr cOMPUtING
low-frequency inter-module linkages, limiting tEcHNOLOGIEs IN cOMPLEX bPM
the scope of interactions between the modules by
hiding the intra-module relations inside a module In the 5th section we propose the DCAR architec-
box. Based on the idea of modularity, constructs ture for developing complex process management
such as objects, components, software agents, and systems, which includes decomposition of com-
Web services have been continuously invented plex processes (D); task coordination (C); aware-
and evolved for developing software applications. ness of environmental changes (A); and resource
OOP provides a foundation for software engi- selection and coordination (R). This provides the
neering that uses objects and their interactions to foundation to identify the BPM requirements for
design applications and computer programs. CBD employing appropriate technologies in develop-
provides a coarser grained construct for larger sys- ing flexible BPM systems. In the 6th section we
tems, and separates interface from the behavior of clarify and explicitly define the similarities and
the construct for supporting public communication differences between the four modular technolo-
between the components which know about each gies: OOP, CBD, SOA, and AOC. In this section
other before runtime. SOA goes further by using we will show how these technologies can be used
XML-based and network-addressable interface as to implement the proposed BPM architecture.
well as XML-based messages and standard proto-
cols for open communication among all software Decomposition of complex
applications in Internet. Different from OOP, CBD, Processes (D)
and SOA, AOC is used to model and implement
solutions to semi- or ill-structured problems, which Business processes display complexity as a result
are too complex to be completely characterized of interactions of their internal components and
and precisely described. Agent is used to perform interaction of the process with its environment.
more autonomous activities in solving complex A process can be decomposed into a set of tasks,
problems. To achieve this, knowledge or rules for task into sub-tasks, and so on, through several
governing the behavior of an agent are separated layers in a network hierarchy. Tasks or sub-task
from the behavior of the agent. components can be delegated to software objects,
In computer science, the terms object, compo- components, agents, and services, as actors of
nent, software agent, and Web service describe a the tasks, which interact and communicate in
piece of software that performs some action on performing the process.
behalf of human beings, like an agent or actor. In To deal with interactions across different
addition to an actor, the agent can also be a broker, organizations, SOA proposes “service” as a high
which mediates between the service requesting level view of the building block of a process. A
party and the service providing party. In terms process is composed of a set of services, each of
of broker, software agent can be used to search which is provided by an individual organization.
appropriate applications to perform requested By using SOA, the inter-service interactions are
1074
separated from intra-service interactions; the of autonomous software entities, which are able
complexity of both is maintained at different to perform decision-based coordination of their
layers. Moreover, we can take advantage of reus- activities.
ability, inter-operability, and extensibility of Web In AOC, after decomposing a complex pro-
services on the basis of open standards to cater cess into a number of loosely coupled tasks in
for business process integration and interopera- a flat hierarchy, we may delegate the tasks to a
tion over the Web. number of autonomous agents, each working both
The highly dynamic and unpredictable nature autonomously and collaboratively throughout the
of business processes makes agent-based approach whole process. In complex process management,
appealing. AOC assigns business applications’ it is impossible to predefine all activities and
main activities to autonomous agents. Such agents interactions at design time. Instead, we define
are flexible problem solvers that have specific the goal or role of each agent, and specify a set
goals to achieve and interact with one another to of rules for governing the behavior of the agent.
manage their autonomy and interdependency in Agents operate asynchronously and in parallel.
business processes. AOC is well suited for complex This results in an increase in overall speed and
process situations that are not all known a priori, robustness in BPM. The failure of one agent does
cannot be assumed to be fully controllable in their not necessarily make the overall system useless,
behaviors, and must interact on a sophisticated where other agents may adjust and coordinate their
level of communication and coordination (Wang behavior reactively and proactively to the change.
et al., 2005b).
Awareness of Dynamic
Flexible task coordination (c) Environments (A)
A business processes is made up of a number The complexity of business processes comes not
of task components that interact with dynamic only from interactions of their internal compo-
behavior. OOP uses objects to hide behavior and nents, but also from interaction of the process with
data, supporting communications among small its environment. To manage business processes
objects, e.g., functions of tasks. CBD extends in a dynamic environment, we need to be able
OOP by supporting interaction among components to continuously perceive the environment, and
(e.g., tasks), i.e., coarser grained constructs, using make real-time decisions on the process. Objects,
public communication interface. SOC goes further components, and services are normally unable to
by using XML-based and network-addressable behave in dynamic environments. Agent-based
interface as well as XML-based messages and software entity is able to sense and recognize the
standard protocols for open communication among situation and determine appropriate actions upon
BPM applications over the Internet. the situations.
While OOP, OBD, and SOA mainly support Unlike the ECA (event-condition-action)
structured communications among tasks or task rules in workflow systems that enable reaction to
components, AOC are able to support ill-structured certain events, AOC goes further by incorporat-
interactions. To coordinate the interactions in dy- ing all environmental information into a mental
namic situations, flatter hierarchies, decentralized state that watches over the whole environment.
autonomous units, and decision-based coordina- Individual events are put together for a compre-
tion mechanisms are required, where AOC is hensive understanding; ambiguous information
directly applicable. AOC supports decentralized is understood after appropriate interpretation and
control and asynchronous operations by a group reasoning (Wang et al., 2006a). As shown in Figure
1075
3, information about the environment (e.g., events, under various constraints. The complexity arises
activities, and resources) is sensed and interpreted from the unpredictability of solutions from service
by the agent on the basis of predefined scheme and providers (e.g., availability, capacity, and price),
rules. In case that information is unanticipated or the constraints on the services (e.g., time and cost
comes as a complete and total surprise, it will be constraint), and interdependencies among the ser-
sent to human manager for manual processing. vices. A service solution to an individual service
Moreover, AOC supports prediction of future involved in an integrated process may not have
state of the environment for purpose of proactive a view of the whole service, very often resulting
actions. Different from passive response to cur- in incoherent and contradictory hypotheses and
rent events, proactive behavior has an orientation actions (Wang et al., 2006b).
to the future, anticipating problems and taking To deal with the problem, AOC can be used
affirmative steps to deal with them rather than for decentralized decision making and coordina-
reacting after a situation has already occurred. It tion. In process integration, decision-making and
refers to the exhibition of goal-oriented behaviors coordination among services can be modeled as
by taking initiatives. a distributed constraint satisfaction problem, in
which solutions and constraints are distributed
Flexible resource coordination (r) into a set of services and to be solved by a group
of agents (brokers) on behalf of service requesters
As discussed, the rise of Internet-mediated e- and providers. In this context, service-based pro-
Business brings the era of demand-driven and cess integration is mapped as an agent-mediated
resource-based soft connections of business orga- distributed constraint satisfaction or optimiza-
nizations. A business process can be dynamically tion problem. Individual services are mapped to
established by connecting or composing services variables, and solutions of individual services
provided by different organizations. SOA provides are mapped to values. A distributed constraint
a real platform for resource selection and alloca- satisfaction or optimization problem consists of a
tion in order to implement seamless and flexible set of variables, each assigned to an agent, where
integration of business processes over the Internet. the values of the variables are taken from finite
However, it is a complex problem to search and discrete domains. Finding a global solution
appropriated services from a large number of re- to an integrated process requires that all agents
sources as well as schedule and coordinate them find the solutions that satisfy not only their own
Figure 3. Agent-based Process Management in Dynamic Environment
1076
constraints but also inter-agent constraints (Wang rEFErENcEs

et al., 2008b).
Baldwin, C. Y., & Clark, K. B. (1997). Managing
in an age of modularity. Harvard Business Review,
cONcLUsION 75(5), 84–93.
Casati, F., Ceri, S., Paraboschi, S., & Pozzi, G.
In this paper, we have investigated the challenges (1999). Specification and implementation of
and solution for complex BPM. Based on the exceptions in workflow management systems.
analysis of business process complexity and the ACM Transactions on Database Systems, 24(3),
modularity theory for complex systems, we pres- 405–451. doi:10.1145/328939.328996
ent the DCAR architecture for developing BPM
systems in turbulent environment. On the other Casati, F., & Shan, M. (2001). Dynamic and
hand, we investigated relevant modular program- adaptive composition of e-services. Information
ming technologies that can be used to model and Systems, 26(3), 143–163. doi:10.1016/S0306-
develop complex systems. We analyze the overlap- 4379(01)00014-X
ping technical concepts and techniques as well as
Chiu, D. K. W., Li, Q., & Karlapalem, K. (1999).
clarify the differences and relationships between
A meta modeling approach for workflow manage-
these terminologies and techniques in the context
ment system supporting exception handling. In-
of BPM. Based on the examination of 1) the re-
formation Systems, 24(2), 159–184. doi:10.1016/
quirements of complex BPM based on the DCAR
S0306-4379(99)00010-1
architecture and 2) the supporting technologies for
complex systems, we have made a clear picture Eisenhardt, M. K. (1989). Agency theory: An as-
with a systemic approach on how these technolo- sessment and review. Academy of Management
gies can be applied and integrated in developing Review, 14(1), 57–74. doi:10.2307/258191
systems for complex process management. This
El Sawy, O. A. (2001). Redesigning enterprise pro-
work will benefit professionals, researchers, and
cesses for e-business. Boston: Irwin/McGraw-Hill
practitioners by advanced analysis and theoreti-
cal investigations of problems and solutions in Elfatatry, A. (2007). Dealing with changes: Com-
developing solutions for complex BPM. ponents versus services. Communications of the
ACM, 50(8), 35–39. doi:10.1145/1278201.1278203
Endsley, M. R. (1995). Toward a theory of situation
AcKNOWLEDGEMENt
awareness in dynamic systems. Human Factors,
37(1), 32–64. doi:10.1518/001872095779049543
This research is supported by a General Research
Fund (No. RGC/HKU7169/07E) from the Hong Garud, R., & Kumaraswamy, A. (2003). Tech-
Kong SAR Government, and a Seed Funding nological and organizational design for realizing
for Basic Research (200611159216) from The economies of substitution. In R. Garud, A. Ku-
University of Hong Kong. maraswamy, & R.N. Langlois (Eds.), Managing
in the modular age: Architectures, networks, and
organizations (pp. 45-77). Blackwell Publishing
Limited.
1077
Hammer, M., & Champy, J. (1993). Reengineer- Kumar, K., van Fenema, P. C., & von Glinow, M.
ing the corporation: A manifesto for business A. (2007). Offshoring and the global distribution
revolution. London: Brealey. of work: Implications for task interdependence
theory and practice. In First Annual Research
Holland, J. (1995). Hidden order: How adapta-
Conference and Workshop on Offshoring. North
tion builds complexity. Cambridge, MA: Perseus.
Carolina
Huhns, M. N. (2002). Agents as Web ser-
Leymann, F., Roller, D., & Schmidt, M. T. (2002).
vices. IEEE Internet Computing, 6(4), 93–95.
Web services and business process management.
doi:10.1109/MIC.2002.1020332
IBM Systems Journal, 41(2), 198–211.
Jennings, N. R., Faratin, P., Norman, T. J., O’Brien,
Melão, N., & Pidd, M. (2000). A conceptual frame-
P., & Odgers, B. (2002). Autonomous agents for
work for understanding business processes and
business process management. International
business process modeling. Information Systems
Journal of Applied Artificial Intelligence, 14(2),
Journal, 10(2), 105–129. doi:10.1046/j.1365-
145–189.
2575.2000.00075.x
Kammer, P. J., Bolcer, G. A., Taylor, R. N., Hi-
Mintzberg, H. (1979). The structuring of organi-
tomi, A. S., & Bergman, M. (2000). Techniques
zations. NJ: Prentice Hall.
for supporting dynamic and adaptive workflow.
Computer Supported Cooperative Work, 9(3-4), Mowshowitz, A. (1997). Virtual organization.
269–292. doi:10.1023/A:1008747109146 Communications of the ACM, 40(9), 30–37.
doi:10.1145/260750.260759
Kumar, A., & Zhao, J. L. (2002). Workflow support
for electronic commerce applications. Decision Petrie, C. J., & Bussler, C. (2003). Service
Support Systems, 32(3), 265–278. doi:10.1016/ agents and virtual enterprises: A survey. IEEE
S0167-9236(01)00114-2 Internet Computing, 4, 68–78. doi:10.1109/
MIC.2003.1215662
Kumar, K. (2001). Technology for support-
ing supply chain management: Introduction. Rumbaugh, J. (1991). Object-oriented modeling
Communications of the ACM, 44(6), 58–61. and design. Englewood Cliffs, N.J.: Prentice Hall
doi:10.1145/376134.376165
Sanchez, R., & Mahoney, J. T. (2003). Modular-
Kumar, K., & Narasipuram, M. M. (2006). Defin- ity, flexibility, and knowledge management in
ing requirements for business process flexibility. In product and organization design. In R. Garud, A.
Seventh Workshop on Business Process Modeling, Kumaraswamy, & R.N. Langlois (Eds), Managing
Development, and Support. CAiSE in the modular age: Architectures, networks, and
organizations (pp. 362-389). Blackwell Publish-
Kumar, K., & van Dissel, H. (1996). Sustainable
ing Limited.
collaboration: Managing conflict and cooperation
in interorganizational systems. MIS Quarterly, Schilling, M. A. (2003). Towards general modular
20(3), 279–300. doi:10.2307/249657 systems theory and its application to interfirm
product modularity. In R Garud, A. Kumaraswamy,
& R.N. Langlois (Eds.), Managing in the modular
age: Architectures, networks, and organizations
(pp. 172-216). Blackwell Publishing Limited.
1078
Simon, H. A. (1977). The new science of manage- Wang, M., & Wang, H. (2002). Intelligent agents
ment decision. Englewood Cliffs, N.J.: Prentice- supported flexible workflow monitoring system. In
Hall Proceedings of the14th International Conference
on Advanced Information Systems Engineering
Simon, H. A. (1981). The sciences of the artificial.
(CAiSE’02) (LNCS 2348, pp. 787-791).
Wang, M., & Wang, H. (2005b). Intelligent agent
Simon, H. A. (2003). The architecture of com-
supported business process management. In Pro-
plexity. In R Garud, A. Kumaraswamy, & R.N.
ceedings of 38th Hawaii International Conference
Langlois (Eds), Managing in the modular age:
on System Sciences (HICSS-38). IEEE Computer
Architectures, networks, and organizations (pp.
Society Press.
15-44). Blackwell Publishing Limited.
Wang, M., & Wang, H. (2006a). From process
Szyperski, C. (2002). Component software:
logic to business logic - A cognitive approach
Beyond object-oriented programming. Bosoton:
to business process management. Information
Addison-Wesley Professional.
& Management, 43(2), 179–193. doi:10.1016/j.
van der Aalst, W. M. P., & van Hee, K. M. (2002). im.2005.06.001
Workflow management: Models, methods, and
Wang, M., Wang, H., & Xu, D. (2005a). The design
systems. Cambridge, MA: MIT Press.
of intelligent workflow monitoring with agent
Volberda, H. W. (1999). Building the flexible firm: technology. Knowledge-Based Systems, 18(6),
How to remain competitive. Oxford University 257–266. doi:10.1016/j.knosys.2004.04.012
Press
Weske, M. (1998). Object-oriented design of
Wang, M., Cheung, W. K., Liu, J., Xie, X., & a flexible workflow management system. 2nd
Lou, Z. (2006b). E-Service/process composition East-European Symposium on Advances in Da-
through multi-agent constraint management. tabases and Information Systems (LNCS 1475,
International Conference on Business Process pp. 119-130).
Management (BPM 2006) (LNCS 4102, pp.
Weske, M. (2001). Formal foundation and concep-
274-289).
tual design of dynamic adaptations in a workflow
Wang, M., & Kumar, K. (2008a). Developing flex- management system. In Proc. HICSS-34. Maui,
ible business process management systems using Hawaii
modular computing technologies. In Proceedings
Wolters, N. J. (2002). The business of modularity
of Eighth Global Conference on Flexible Systems
and the modularity of business. PhD Thesis, Eras-
Management (GlOGIFT-08). Hoboken, NJ.
mus Research Institute of Management, Rotterdam
Wang, M., Liu, J., Wang, H., Cheung, W., & Xie, X.
Wooldridge, M., & Jennings, N. R. (1999).
(2008b). On-demand e-supply chain integration:
software engineering with agents: Pitfalls and
A multi-agent constraint-based approach. Expert
pratfalls. IEEE Internet Computing, 3(3), 20–27.
Systems with Applications, 34(4), 2683–2692.
doi:10.1109/4236.769419
doi:10.1016/j.eswa.2007.05.041
1079
KEy tErMs AND DEFINItIONs Component-Based Development: Compo-

nent-Based Development (CBD) is software en-
Agent-Oriented Computing: Agent Oriented gineering discipline, with an emphasis on decom-
Computing (AOC) is based on the idea of delegat- position of the engineered systems into functional
ing tasks and responsibility of a complex problem or logical components with well-defined interfaces
to software agents. It emphasizes autonomy and used for communication across the components.
mutual co-operation of agents in performing tasks Modularity: Modularity refers to a particular
in open and complex environments. design structure, which refers to development
Business Process: A business process can be of a complex product or process from smaller
simply defined as a collection of activities that subsystems that can be designed independently.
create value by transforming inputs into more Object-Oriented Programming: Object-
valuable outputs. These activities consist of a oriented Programming (OOP) is a software en-
series of steps performed by actors to produce a gineering paradigm that uses “objects” and their
product or service for the customer. interactions to design applications and computer
Business Process Management: Business programs
Process Management (BPM) refers to activities Situation Awareness: Situation awareness
performed by organizations to design (capture is the perception and understanding of objects,
processes and document their design in terms of events, people, system states, interactions, envi-
process maps), model (define business processes ronmental conditions, and other situation-specific
in a computer language), execute (develop soft- factors in complex and dynamic environments
ware that enables the process), monitor (track Service-Oriented Architecture: Service-
individual processes for performance measure- Oriented Architecture (SOA) utilizes Web services
ment), and optimize (retrieve process performance as fundamental elements for developing applica-
for improvement) operational business processes tions. It is an emerging paradigm for architecting
by using a combination of models, methods, and implementing business collaborations within
techniques, and tools and across organizational boundaries.
Adaptability in Turbulent Environments, edited by Minhong Wang and Zhaohao Sun, pp. 1-22, copyright 2010 by Information
1080
1081
Chapter 4.13
Multiple-Step Backtracking
of Exception Handling
in Autonomous Business
Process Management
Mingzhong Wang
University of Melbourne, Australia
Jinjun Chen
Swinburne University of Technology, Australia
Kotagiri Ramamohanarao
Amy Unruh
AbstrAct the flexible backtracking approach is introduced

by utilizing the beneficial features of event-driven
This chapter proposes a multiple-step backtracking and means-end reasoning of BDI agents. Finally,
mechanism to maintain a tradeoff between replan- we incorporate open nested transaction model to
ning and rigid backtracking for exception handling encapsulate plan execution and backtracking to gain
and recovery, thus enabling business process man- the system level support of concurrency control and
agement (BPM) systems to operate robustly even in automatic recovery. With the ability of reasoning
complex and dynamic environments. The concept about task characteristics, our approach enables the
of BDI (belief, desire and intention) agent is applied system to find and commence a suitable plan prior
to model and construct the BPM system to inherit to or in parallel with a compensation process when
its advantages of adaptability and flexibility. Then, a failure occurs. This kind of computing allows us
to achieve business goals efficiently in the presence
of exceptions and failures.
DOI: 10.4018/978-1-60566-669-3.ch010
Multiple-Step Backtracking of Exception Handling in Autonomous Business Process Management
INtrODUctION ecution path is found. Moreover, they follow a

defensive route of recovery-then-try pattern. Due
A critical challenge in building practical business to the dynamic and nondeterministic features of
process management (BPM) systems is to allow complex environments, rigid recovery on the
users to maintain the system robustness and reli- reverse chronological order of execution history
ability with respect to the correct execution even will in many instances become meaningless and
in the presence of abnormalities. As the operat- inappropriate. Some attempt to consider starting
ing environment becomes increasingly complex, a fresh plan after the exception. However, this
dynamic and error-prone, it is extremely challeng- approach is too computationally expensive and
ing for designers and programmers to find out rarely practical to get deployed because it requires
all possible combinations of exceptions as well a complete knowledge about the running environ-
as designing corresponding handling methods. ment and needs to consider all previous actions.
Therefore, a flexible, systematic and autonomic As a result, developers are forced to consider
approach for exception handling is essential for low-level details of disturbances, failure, or un-
the success of applying complex BPM systems controlled interactions between workflow actors
into wider fields of application. for the requirement of robustness and reliability.
Multi-agent systems have been extensively To address this issue, we propose to extend
studied as a powerful high-level decomposition existing rigid backtracking strategy to support and
and abstraction tool in analyzing, designing, and enable execution backtracking in a multiple-step
implementing complex software systems (Jen- fashion (reverse chronological order with certain
nings, 2001). Many researchers and practitioners steps skipped). Instead of step-by-step backtrack-
have noticed the fundamental relationship between ing through the execution tree, the system can
agents and workflow systems (Ehrler, Fleurke, “jump” back to an arbitrary history node and try
Purvis, & Savarimuthu, 2006) and proposed vari- another eligible execution path. Our approach
ous approaches to build dynamic and adaptive maintains a tradeoff between replanning and
workflow systems with the concept of agents, traditional backtracking strategies.
and vice versa. To provide automatic and system-level support
To manage the complexity arisen from the for concurrency control and exception handling,
dynamic and complex running environment, we open nested transaction (Weikum & Schek, 1992)
propose to apply the reactive BDI (Belief, Desire is integrated into our multiple-step backtracking
and Intention) (Rao & Georgeff, 1995) agent model. Compared with other approaches which ap-
system to model and construct BPM systems, ply transaction models (Gray & Reuter, 1993) into
thus benefiting from its sound features of event- workflow (Georgakopoulos, Hornick, & Manola,
driven and means-end reasoning for the purpose 1996) or multi-agent systems (Ramamohanarao,
of robustness, flexibility and adaptability. Bailey, & Busetta, 2001; Nagi, 2001), our method
Providing a higher level of abstraction, agents unites the beneficial features of event-driven and
help to simplify modeling and construction of means-end reasoning from BDI agent systems
flexible BPM systems operating in open and dis- and utilizes a flexible backtracking approach
tributed environments. To deal with exceptions, to allow the execution “jumping” back several
most multi-agent systems apply the mechanism levels at once to continue its execution towards
of backtracking to recover and retry the failed the goal in case that backtracking to one level in
execution. However, they back track in a rigid the execution tree does not solve the problem.
step-by-step manner until one alternative ex-
1082
bAcKGrOUND about how to achieve a certain goal or re-

act to a specific situation. Intentions are
bDI Agents and Exception Handling selected plans for execution and represent
the deliberative state of an agent.
Software agents have the ability to provide au- • The execution of an agent is event driven.
tonomous and reactive behaviors, and to support Plans, which are usually represented in the
decomposition and abstraction of functionality, form of event ¬ preconditions | action_se-
making agent technology useful in analyzing, quence, are defined to react to a certain
designing, and implementing complex software event which can be internal modifications
systems (Jennings, 2001). They are persistent to its goals and beliefs or external changes
entities that can perceive, reason and act in some of environment. After event is triggered,
environment. Often, agents are autonomous, reac- the preconditions will be tested before ac-
tive, and sociable (Woolridge, 2001). tion_sequence can be chosen for execu-
In order to push agent technology into the tion. Because events can occur non-deter-
mainstream of software development, various ministically, plans are executed reactively.
agent architectures (Woolridge, 2001) and agent • The execution path to achieve a goal of an
programming languages have been proposed. agent is generated by means-end reason-
Among them, BDI (Rao & Georgeff, 1995) is probing. That is, the goal is treated as an ini-
ably the most mature and accepted model. Most tial event triggering a corresponding plan
BDI platforms with PRS (Procedural Reasoning to run. action_sequence of the plan may
System) style share the following three features. contain primitive actions as well as sub-
goals. All sub-goals will be in turn treated
• An agent contains four key data structures as events to trigger sub-plans to run. This
as shown in Figure 1. Beliefs are the infor- process continues recursively until all ac-
mational state representing what an agent tions in sub-plans are primitive or atomic.
knows about itself and the world which
may be incomplete or even incorrect. The execution process of a BDI agent can be
Goals are the motivational state and corre- abstractly depicted in Figure 2. This process is
spond to what the agent wants to achieve. described by (Rao & Georgeff, 1995) and applied
Plans represent the procedural knowledge directly or indirectly by dMARS (d’Inverno et al.,
Figure 1. PRS-style structure for BDI agent. (Adapted from (d’Inverno, Luck, Georgeff, Kinny, & Wool-
dridge, 2004))
1083
Figure 2. Event driven and means-end reasoning for BDI system execution
2004) and 3APL (Dastani, Riemsdijk, Dignum, other extreme, backtracking is applied to go back
& Meyer, 2003). through all the work that previously been done
The plan matcher will retrieve an event from in planning in a depth-first searching manner.
the queue and search through the plan library to Despite the feasibility of backtracking requires
find the set of plans which can handle this event a rigid execution tree structure, it becomes the
in that certain situation (determined by its beliefs). most widely applied for exception handling in
There might be more than one suitable plan found, multi-agent systems because it fits nicely to the
and the plan selector will choose one of them and hierarchical program architecture and enables
append it into the intention stack. Finally, the systematic and platform-level recovery support.
intention is executed, which results in internally To back track one step in the execution tree
updating the BDI state, including beliefs, desires needs some compensation. Many existing agent
and intentions, or externally operating to sense programming languages provide some basic
and change the environment. support for modeling exceptions and their com-
Since agents usually work in open, dynamic and pensations. They usually use a specified event to
error-prone environment, they are more liable to trigger a new plan to define compensating actions.
conflicts and eventually failures. Without dealing For example, 3APL (Dastani et al., 2003) allows
with these problems appropriately, agent system using plan revision rules to define compensating
can only remain as an experimental toy. Therefore, actions for a specified event. dMARS (d’Inverno
substantial effort is carried out with attentions et al., 2004) allows encoding of maintenance
to automate exception handling and execution conditions, as well as success and failure actions,
resumption, thus making agent programming a into a task definition. However, all of them lack
serious platform to developers for developing a systematic way to organize and manage those
complex applications. exceptions and compensations, making complex
On one extreme, replanning is proposed to have system design and interaction management dif-
a fresh start by throwing away all the work that ficult and sometimes impossible.
has been previously done in planning. However, Some researchers propose to apply trans-
it is too computationally expensive and rarely action concepts to help build systematic and
practical because it requires complete knowledge automatic platform for robust agent execution.
about the environment and business domain and TOMAS (Ramamohanarao, Bailey, & Busetta,
involves a lot of programming efforts. On the 2001) applies a nested transaction model as the
1084
concurrency control and recovery mechanism to Exception Handling in

avoid performing conflict update operations on Workflow system
the agent. However, since a nested transaction
model requires full control on the resources and A primitive form of multi-agent system can be
no exposure of partial results, it is not feasible in viewed as a workflow model in the sense that it
most situated agent systems. Even if all agents has a predefined finite (static) execution flow.
perform database manipulation only, the model However, they share the same problem with regard
is still too rigid to be practical in a multi-agent to exception handling. It is therefore essential for
system, because long-running activities can lead the success of a workflow management system that
to locking resources for very long periods and it provides a powerful yet easy-to-use mechanism
can cause deadlocks. (Nagi, 2001) takes a more for maintaining system robustness and reliability,
workflow-like approach by treating every action with respect to the correct execution of tasks even
as an ACID (Atomicity, Consistency, Isolation in the presence of abnormalities and exceptions.
and Durability) entity and putting them into open (Georgakopoulos, Hornick, & Manola, 1996)
nested transaction model to create transaction argues that Customized Transaction Management
trees. ECA (Event, Condition, and Action) rules (CTM) is one of the key infrastructure technolo-
are used to link two agents if one uses the partial gies for effective workflow management system.
results of another. But this approach does not allow CTM can ensure the correctness and reliability of
changes of execution plans, which are quite com- applications implementing business or informa-
mon in agent applications. What is more, arbitrary tion processes, while permitting the functionality
and unstructured linkages among different agents each particular process requires (e.g., isolation,
breach the design principles of modularization coordination, or collaboration between tasks).
and loose coupling. The concept of transaction is a concise but
As an extension to the compensation concept powerful abstraction tool with the properties of
in advanced transaction models, (Unruh, Bailey, concurrency control and failure atomicity. Tradi-
& Ramamohanarao, 2004) discusses the use of tional transactions have ACID properties (Gray
goal-based semantic compensation in the context & Reuter, 1993), which prevent inconsistency
of agents. and integrity problems. Atomicity ensures that
Compared with transaction-based solutions, either all or none of operations of a transaction are
Guardian (Tripathi & Miller, 2001) and Citizens performed. Consistency ensures that a transaction
(Klein, Rodriguez-Aguilar, & Dellarocas, 2003) maintains its integrity constraints on the objects it
separate the mechanisms and knowledge about manipulates. Isolation ensures that a transaction
exception handling from the agent system to a executes as if it were running alone in the system
centralized exception manager. However, this and intermediate transaction results are hidden
separation only works well for some domain- from other concurrently executing transactions.
independent exceptions. SaGE (Souchon, Dony, Durability ensures the changes once made by
Urtado, & Vauttier, 2004) proposes to organize successfully completed transactions are persistent
agents as well as their exception handlers in a even when systems crash.
hierarchical structure, but it lacks a precise model In database applications, each transaction is
of what to do after the exception has been handled enforced to have the ACID properties. Therefore,
based on the exception-handling plan and the whenever there is an exception occurring, all work
result of executing that plan. already done in the transaction is aborted and the
system is restored to the starting state as if noth-
1085
ing has happened. However, ACID transactions each level starting from that ancestor (Kamathy
require an activity to obtain full control of, and & Ramamritham, 1996).
exclusive access to, its resources. In contrast to Another proposal is called sphere of joint
database applications where transactions can lock compensation (Kamathy & Ramamritham, 1996).
records to get exclusive access, and can restore any A collection of tasks in a workflow is grouped
data from history, workflow tasks in nature usu- into a sphere S such that either all the tasks of S
ally work in an open environment and operate on complete successfully or all of them are compen-
physical objects where actions “always commit” sated. Thus a sphere is basically a failure-atomic
and it is impractical or even impossible to satisfy unit. Spheres can overlap and be nested. Actually
either requirement. For example, a flight reserva- spheres are relaxed transactions with the property
tion task cannot lock the schedule to avoid flight of atomicity and consistency. The problem is how
changes or restore a bank account to the original to define the scope of the sphere.
amount by itself when cancelling a booking. Although these methods can guarantee the
Thus, traditional transaction mechanisms need to correctness of the system execution, they do not
be extended for an open and shared environment consider issues related to the context changes
before they can be applied to workflow systems. arising from dynamic and complex environment.
The extended transaction models in workflow For example, when a branch of the execution tree
systems usually focus on the inter-dependency fails, other possible solutions need to wait until the
among the participating tasks. The overall recovery completes. If the running environment
workflow process is treated as a big transaction for recovery has become different from initial
which organizes all participating tasks in a tree execution or design assumption, the overall sys-
structure to ensure the criterion of correctness and tem is taken down even though sometimes there
reliability. Each task node is considered as sub- are clear other approaches existing to continue
transaction which may not be required to hold all the execution for achieving the business goals.
ACID properties. The relationships among these As such, a more flexible recovery mechanism is
sub-transactions are depicted by task dependen- required to organize transactional recovery and
cies which can ensure the execution order and forward execution in complex interacting systems.
execution correctness.
Based on the dependencies among tasks, the A Motivating Example
workflow management system can generate plans
to execute. Theoretically when exceptions occur, A travel management system with the goal of pre-
the system can retrieve all affected tasks and co- paring a holiday for the customers is illustrated in
ordinate them to handle the exceptions together. Figure 3. All alternatives at the choice points are
Generally, the failure of a task at a given level may eligible for selection. Let us assume the system
or may not affect its parent step. If it does, then chooses California as the tour destination from
the parent is to be rolled back and the procedure all three possible places. Then it tries to book the
is repeated until a parent task is reached that is not flight and the hotel. After that, it chooses Sailing
affected. From this point on, a parent task may try as the entertainment plan. Finally, it begins to rent
an alternative child step to compensate or restart. a yacht. Unselected branches during the execution
Then the overall process is a two-phase remedy are cut off to simplify the presentation.
where the first phase, called the bottom-up phase, If yacht renting fails, traditional backtracking
determines the highest ancestor task affected by mechanism will go back to the choice point of
the failure of the current task and a second phase, Entertainment and try Disneyland. However, if
called the top-down phase, undoes the changes at tickets for the Disneyland are sold out, the system
1086
Figure 3. Travel management system
needs to keep on backtracking by undoing hotel Addressing these drawbacks, we propose a

booking and flight booking to the choice point of multiple-step backtracking mechanism for recover-
Destination and try either Hawaii or Hongkong ing exceptions occurred in BPM systems running in
as the new destination. dynamic and open environments. In our approach,
However, this rigid step-by-step backtracking when the system knows going back one level in
approach is sometimes not appropriate when the the execution tree can not deal with the exception,
environment keeps on changing. First, its require- it can jump back several levels at once instead to
ment of keeping the execution structure static for continue its execution towards the goal. We will
recovery purpose is hard to fulfill in a dynamic also show how open nested transaction model is
and non-deterministic environment, and recovery integrated into our flexible backtracking approach
on the reverse order of the execution history will to provide systematic and automatic support for
probably become meaningless in many situa- exception handling and concurrency control.
tions. For example, the closure of the hotel after
the booking has been made will invalidate any
rollback or compensation attempts and result in bDI-styLE EXEcUtION
the discontinuity of achieving the goal of holiday MODEL FOr bPM
travel. Second, after exceptions occur, it requires
complete compensation of failed plans before try- In order to operate in complex and dynamic envi-
ing any further efforts to achieve the goal. Because ronment, the execution tree of a business process
applying other plans to approach the goal may not must be generated at run-time reactively to its sur-
always conflict with the process of compensa- roundings. We propose to apply BDI framework to
tion, this requirement results in inefficiency with construct the dynamic composition of individual
respect to goal achieving and sometimes brings autonomous participants in the system to achieve
unnecessary costs. a certain business goal.
1087
The application domain has a predefined set achieve the goal of holiday arrangements for the
of business rules. For a certain goal in a specified system than to undo its payment of the booking
environment, it is broken down into sub-goals deposit.
by matching it with corresponding plans. Each Principle 2. Cleaning up a side effect of a
subgoal can be achieved by delegating to some failed selection Scp (i ) can have lower priority
agents, or being decomposed further with appropri- compared with trying its alternatives, or achieving
ate plans. Compared with conventional workflow the system’s goal.
systems, the proposed BPM execution model is Compensation is usually used to release the
more dynamic and adaptive to the environment consumed resources which are necessary for
because the execution tree is constructed at run other attempts. However, it is the approach, not
time according to the real situation. For the pur- the purpose. All unrelated compensations can be
poses of the approach we describe in this chapter, processed in parallel with or even after the achieve-
we can assume without loss of generality a plan ment of the goal to improve system throughput
is a task to achieve. Thus, we will use task and and efficiency. The applicability of this principle
plan interchangeably. is based on the fact that system plans are dynami-
During the plan matching process for a goal, cally composed when their events are triggered
there may exist more than one feasible plan. Then, and their preconditions are satisfied. However,
it is said the system has a choice point which is in resource-bounded situation compensation may
a key concept in our model. have a higher priority.
Definition 1. A choice point of the execution Principle 3. If the system is consistent, any
tree records the applicable plans found by plan plan in its plan library can be selected for execu-
matcher to deal with a certain goal. It is denoted as tion if the plan preconditions are met.
cpgoal = {p1, p2 , , pn } where pi is an applicable These rules allow us to continue system execu-
plan. Scp (i ) stands for the selection of plan pi tion at any legal choice point after current execu-
by plan selector. tion encounters interruption. In other words, it is
Because the BPM system is modeled and built not necessary to conservatively roll back before
by following BDI agent systems, its execution trying other paths. The execution flow just needs
shares the same features as multi-agent systems. to “jump” back to an appropriate choice point and
Principle 1. After a plan has made a change continue. We will give a more detailed description
to the environment, if the whole system is still in of this concept in the next section.
a semantically consistent state satisfying system
constraints, this state is acceptable even though
the actions are later shown to be futile in working FLEXIbLE EXcEPtION
towards a goal. HANDLING MODEL
This rule allows the execution of the BPM
system to be treated in terms of independent small We first explain the multiple-step backtracking
plan fragments. Each such fragment can be en- mechanism which maintains a tradeoff between
capsulated in a transaction which can externalize replanning and rigid backtracking for exception
its result after termination without causing long handling and recovery. We then describe how we
running transactions. For example, even if the can incorporate the open nested transaction model
travel management system cannot arrange any to encapsulate plan execution and backtracking
entertainment in California, the existence of flight to gain the system level support for concurrency
and hotel booking does not prevent it from trying control and automatic recovery.
other options. Moreover, it is more important to
1088
Multiple-step backtracking recovery Definition 3. A jump is a continuation of the

system execution flow at another selectable plan
Traditional Backtracking recovery tries to ex- implied by elements of its choice point stack. To
plore the solution domain in a depth-first search enable the jump to the jth choice of cpi , the pre-
manner. The execution of the system is usually conditions of the plan Scp ( j ) must be satisfied.
organized as a tree. When a task node fails, the i
Preconditions of a plan usually specify the

system would go one step back to its parent by
required resources and execution context. For
undoing or compensating the result done by the
example, after the failure of hotel booking, the
task. If there is still no solution in the parent node,
travel management system may find there is not
the system would recursively back track to the
enough money left for Hong Kong or Hawaii
upper level until reaching the root. Compared
because money has been paid for air ticket to
with this rigid search strategy, our approach can
track back multiple levels at once by skipping California. The failed branch ( Scp ( f ) ) and the
n
intermediate steps according to the evaluation of new selected branch ( Scp (s ) | i £ n ) may have
i
the real time condition. Therefore, even in case three types of coordination with respect to their
step-by-step backtracking is invalid because of the resource share or competition. Different type of
environment changes, the system can still survive coordination leads to different processing strategy.
by recovering and continuing the execution flow
from a higher level of control. This section at first • Scp ( f ) does not consume any resources
n
introduces and defines the key concepts related which will be used by Scpi (s ) . That is,
to the flexible backtracking mechanism, and then Scp ( f ) has no access to the resources
shows the details about the backtracking algorithm n
which must be guaranteed as precondi-
for the recovery. Finally, the proposed approach tions of Scp (s ) . In this case, Scp (s ) can be
is summarized and compared with the generally i i
started directly prior to addressing the fail-
applied handling methods. ure. However, there could be a background
thread reclaiming the resources consumed
Key Concepts for Multiple- by Scp ( f ) .
Step Backtracking n
• Scp (s ) shares some resources with Scp ( f )
i n
However, there are still enough resources
The system can use the knowledge of its set of
choices at each choice point to decide which sub- left for Scp (s ) after the consumption by
i
stitutable plan to choose if exceptions occur later. Scp ( f ) . The handling method of this case
n
Definition 2. A choice point stack contains the is the same as the first type. As shown in
choice points the system has met chronologically. Figure 4 (a) and (b), the execution can
It is denoted as cp _ stack = {cpn , cpn −1, , cp1 }
jump directly to arrange travel at different
where cpi is the choice point occurred at time i .
destination because flight and hotel book-
When there is a failure of execution, it must
ings have not spent anything, or only spent
happen at one branch of cpn . The system can look a small amount of money. The refund pro-
through its stack for another applicable branch of cess can be ignored, or be carried out in
a certain past choice point to continue achieving parallel with the goal achieving or even af-
its goal. If possible, the new selected branch can ter the goal of the agent has been achieved.
be identical to the failed one as a retry. We call
• Scp (s ) shares some resources with Scp ( f )
this process a “jump”. i n
of which too much has been used for
1089
Figure 4. Different recovery strategy for different type of path coordination
Table 1. Algorithm 1. Execution flow of “jump”

Scp (s ) to remain applicable. In this case,
i
compensation must be applied first be- Func buildStack() {
fore execution can continue at Scp (s ) . As if number of eligible plans > 1then
i save plan choices into choice point stack;
shown in Figure 4 (c), because the booking }
of flight and hotel has cost too much mon- Func evalAndTry() {
ey, the agent must obtain the refund before fori = n; i > 0; i--do
continuing towards the goal. foreachchoice point Scp ( j ) do

i
ifisApplicable( S ( j ) ) then
cp
i
Based on these different types of coordination, continue at Scp ( j ) ;
we have designed an innovative failure recovery i
compensate previous work after cpi ;
algorithm to achieve non-stop execution towards remove cp , cp , , cp from stack;
the agent’s goal. It is embedded in an open nested break;
n n −1 i +1
end if
transaction structure tailored for BDI agents, end foreach
thus enabling the build of robust and reliable end for
BPM systems with architecture-level support of }
concurrency control and higher efficiency and Func backOneStep() {
throughput to deal with exceptions. randomly select Scp (r ) ;

n
add preconditions of Scp (r ) as new subgoal;

Recovery Mechanism and Algorithm } n
Func jump() {
ifsystem execution failsthen
As shown in Algorithm 1, the overall execution ifevalAndTry() failsthen
flow of “jump” mainly consists of three parts backOneStep();
end if
operating on the data structure of choice point end if
stack. (Table 1) }
The choice points (including their sets of choic-
es) encountered during the execution are stored
into the stack as one static data structure of the ing the execution in accordance with the traversal
BPM system. The stack is built and updated dur- along its execution tree structure. As the system
1090
traverses down the execution tree, it pushes the a new selected branch. backOneStep() randomly
choice point it meets into the stack. When there selects a branch at the last choice point and cre-
is a failure preventing forward execution, the ates new subgoals to achieve its preconditions if
execution jumps back to a previous choice point. “jump” can not find an appropriate substitutable
Meanwhile, all choice points coming after the se- plan. This step relies on the means-end reasoning
lected one are removed from the stack altogether. ability of BDI agents, and suspends the forward
Note that the selected choice point becomes the execution of the agent until compensation or other
top element of the stack to indicate the current measures are taken. The compensation process
execution flow. These operations guarantee the generally comes from two methods. One is to
correctness of the following theorem. compensate back step by step like sagas (Hec-
Theorem 1. If a choice point cpi appears in tor & Kenneth, 1987); another is to generate a
the stack, all and only its ancestor choice points compensation plan for all the backtracked steps
are present in the stack at the same time. altogether (Eiter, Erdem, & Faber, 2004). With
Proof. Let cpa be the ancestor choice point the choice point stack, the method jump() can be
of cpi . Only two possibilities can result in that added into the system deliberation cycle (Figure
cpa is not present in the stack: cpa has not been 2) as a general exception handling mechanism.
For the failed branch, its compensation may
visited at all or cpa has been jumped by. In the
be carried out in parallel or after the execution of
first case, cpi should also not be visited, and in
its alternatives. Thus, the result of the compensa-
the latter case cpi should also be jumped by.
tion may falsify the validity of the plan chosen in
However, this is conflict to the fact that cpi is in “jump” process.
the stack. Therefore, all ancestor choice points of However, if the compensation does not over-
cpi are in the stack. correct, but only partially or completely recover
Let cp j be a choice point in the stack which is existing side effects, the order of execution will
not the ancestor of cpi , then cp j can only either not affect the final result. Non-over-correction
be a descendant of cpi , or have no relationship is denoted as comp(r ) £ invl (r ) where invl (r )
with cpi at all. In the first case that cp j is the represents the amount of involved resource r in
descendant of cpi , cpi can not be the top ele- normal execution and comp(r ) is the reversal
ment of the stack because cp j should be visited of r in the corresponding compensation. For
later than cpi . In the second case, cp j is not on example, if there is a refund for flight booking,
the passenger will not pay more fines than the
the path from the root to cpi . In other words, cp j
airfare, nor the air company refund more than
can not be visited since it is not reachable in the
the airfare.
execution towards cpi , thus being unable to be in
Theorem 2. The result of compensation to the
the stack. Therefore, only ancestor choice points
failed branch ( Scp ( f ) ) will not affect the valid-
of the top element are in the stack.
ity of the execution of the new selected branch
buildStack() is invoked by the plan selector in
Figure 2 to build up the choice point stack for the ( Scp (s ) ) if the compensation has the feature of
“jump” process. evalAndTry() is the core method non-over-correction.
which searches through the stack to find a quali- Proof. If the compensation occurs before any
fied plan to run in case of exception. This method other progress is made, it can not invalidate the
utilizes event-driven feature of BDI agents. During latter “jump” process.
this part, compensation to a failed branch may be For the compensation to occur after
carried out in parallel with or after the execution of “jump”, the initial resources r and the mini-
1091
mum requirement rmin for Scp ( f ) and Scp (s ) we introduce the operation evalAndTry(), which
satisfy r ³ rmin . Otherwise, they cannot be evaluates each choice point to find if there is a di-
saved as choice point options in the stack. If rectly applicable plan. If one is found, it is selected
Scp ( f ) consumes resources cons(r ) , the se- for execution without waiting for the completion
of compensation. So, the backtracking in our
lection of Scp (s ) means r − cons(r ) ≥ rmin .
approach follows the journey of evalAndTry() ||
Thus, the compensation is unable to in-
compensate(). If this procedure fails to find an
validate the execution of Scp (s ) after releas-
applicable plan from any branch of the choice
ing some resources back. Conversely, if
point stack, the execution pattern is converted to
Scp ( f ) produces resources prod (r ) , we get
compensate(nearestCP ) || try() . The compen-
r + prod (r ) − comp(r ) ≥ r ≥ rmin under the sation must complete before try() if and only if
feature of non-over-correction after the com- the previous execution has consumed too much
pensation. Then the precondition of Scp (s ) is resources to allow continuation. Our approach will
still guaranteed. have higher throughput because compensation and
the substitutable plan can execute in parallel, and
Relation to Existing Approaches goal achievement becomes more efficient.
The approach implicitly makes the assump-
There are three primary possible strategies for tion that certain domain knowledge, such as goal
recovery: replanning, which discards previous preconditions and action effects, can be modeled
work; plan patching, which continues the current with sufficient accuracy for such decisions. If this
execution plan after repairing the cause of failure; were not the case in some contexts, compensa-
and backtracking, which tries all alternatives in a tion can make subsequent actions more robust,
depth-first searching manner. since it helps avoid interactions that are not well
Replanning should be the last resort because modeled. Thus doing the compensation first can
it is expensive to deploy. Plan patching is the be adopted as the default handler to make the
most preferable approach; however, it is usu- system more robust.
ally not available because of uncontrollability In fact, we can make some modifications to
of the environment. Backtracking remains as the last two functions of Algorithm 1 to simulate
the most feasible strategy which can be carried different backtracking strategies. For example,
out systematically and protects existing work as if evalAndTry() is constrained so that it will not
much as possible. Moreover, it can be mapped return a directly executable plan, our approach
to nested transaction structures and fulfilled by becomes very similar to the standard one.
compensation. In case no eligible plans are found from the
Our model is essentially a variant of backtrack- choice point stack, the system will make the replan
ing strategy to support and enable flexible recovery. from the existing situation to continue the execu-
However, we do not follow the standard backtrack- tion, as described in the function backOneStep().
ing journey of compensate(nearestCP ) - try() Compared with merely replanning, our approach
where compensate(nearestCP ) compensates uses the previous plan as a guide for further plan-
the works done after the nearest choice point, ning; therefore tries to keep finished work as much
and try() will select a different branch to run. as possible to be valid.
Here, “–” denotes sequential execution while “||” Our method is more general than the conser-
denotes parallel. vative backtracking when plans in different tree
Instead, we argue that achieving the goal has branches become more and more independent.
higher priority than doing compensation. Thus, The introduction of a transaction manager also
1092
frees programmers from considering low-level drawbacks of applying open nested transaction
error-prone details of concurrency control. (Eiter, directly into BPM systems, such as the requirement
Erdem, & Faber, 2004) describes a similar method of a rigid and static execution structure.
which recovers from execution problems by back- A plan may contain complex internal structure
tracking to a past nondeterministic choice point, with respect to the goal decomposition, but it is its
from which the system tries to “repair” the causes interfaces, not its implementation details, that are
of failure and then continues. However, their aim the concerns of the user. The internal execution
is to generate a reverse plan to compensate back of the plan is a black box for its caller.
to a previous point and retry from there, which Definition 4. A plan is an atomic unit of
also follows the compensate-then-retry pattern. work from the observer’s point of view. Before
Their approach can be adopted to generate a the plan p is performed, it is in the start state
compensation plan for the failed execution path. S 0 , and after it finishes, it is in the end state Sn
denoted as S 0   p
→ Sn .
transactional Execution Figure 5 shows the interface of a plan. Plan p
and its state changes are constrained by enabling
The open nested transaction model will be inte- conditions (EC), invariants (I) and termination
grated into the BPM execution platform to provide conditions (TC). p can begin to execute if and
system level support of automatic recovery and only if its enabling conditions are satisfied, that
concurrency control. The proposed platform brings is S 0 ⊨ EC Ù I where ⊨ means satisfying. The
the following benefits: change from S 0 to Sn is consistent if the invari-
ants of the plan are satisfied. The termination
• It guarantees modularity and failure ato- conditions specify the desired outcome, or cor-
micity of plans. Thus, programmers only rectness criteria, of executing p. It is also possible
need to deal with two situations: the plan that exogenous events, unanticipated interactions
succeeds completely or fails without side between agents, or non-deterministic action results
effects. For example, if renting a yacht may cause plan to be aborted when invariants are
fails, the system will return back to the violated. The possible outcomes of p are:
choice point of Entertainment with the
details about the incomplete deal with the
• Consistent if Sn satisfies I, denoted as
yacht company abstracted out.
I (Sn ) Þ Sn ⊨ I .
• It enforces automation of exception han-
• Inconsistent if Sn dissatisfies I, denoted as
dling and execution resumption. For ex-
ample, after returning back to the choice ¬I (S n ) ⇒ Sn ⊭ I .
point of Entertainment automatically, the • Correct if Sn is consistent and satis-
system can retry renting again or choose to fies TC, denoted as TC (Sn ) Þ Sn ⊨
visit Disneyland instead. TC ∧ I ⊆ TC .
• The nested structure of transactions also • Consistent but not correct if Sn is con-
matches the tree structure of plan composi- sistent but not satisfies TC, denoted as
tion elegantly. It provides inter- and intra- ¬TC (Sn ) ∧ I (Sn ) ⇒ (Sn ⊭ TC ) Ù (Sn ⊨
task level support for distribution and con- I ) ∧ I ⊂ TC .
currency control.
Two kinds of plan models, an atomic plan and
The flexible backtracking recovery strategy a composite plan, have been discussed above, they
introduced in this chapter also helps to avoid the
1093
Figure 5. Interface of a plan
map to the same underlying model. Whether a plan cause it is fully integrated with the Java platform
is viewed as atomic, or composed of other plans, and programming environments, 3APL-M is a
depends upon the abstraction level at which it is good prototyping tool for cognitive agents. Fur-
considered. If we need to access a plan’s internal ther, programs written in 3APL-M can be easily
organization, we model it as compositional. If migrated into 3APL because they share similar
we just need to use its functionality as a building underlying language concepts and programming
block, we model it as an atomic plan. constructs.
To avoid conflicts when accessing the shared JBoss Transaction Service is employed as the
resources such as bank account or available hotel transactional execution manager, which guaran-
rooms, traditional lock mechanisms are applied. tees the isolation of parallel plans. Its TxCore
However, it is a requirement that the lock is ob- transaction engine supports both closed and open
tained only when the resource is visited and re- nested transaction and presents programmers with
leased as soon as possible after the visit. Otherwise, a toolkit of Java classes which can be invoked by
it is likely to cause the problem of long-running 3APL-M agents directly to obtain desired proper-
transactions which result in low efficiency and ties, such as persistence and concurrency control.
even deadlock. As long as the system is consistent, As shown in Figure 3, a root goal of the system
cascading rollback is avoided. is decomposed into subgoals recursively, result-
ing in a tree structure. Each subgoal is achieved
through a transaction-like function as shown
IMPLEMENtAtION AND in Algorithm 2. They are then organized in an
EXPErIMENts open nested transaction structure. Whenever an
exception occurs, the execution of that transac-
We have built a prototype of reactive workflow tion is taken over by the function jump() shown
using 3APL-M (Koch, 2005) and JBoss Transac- in Algorithm 1. jump() will fork two threads: one
tion Service (formerly Arjuna Transaction Service) is used to continue the execution at an earlier
(JBoss Inc., 2006) to simulate the travel arrange- choice point, and the other is used to compensate
ment example. the failed plan. (Table 2)
3APL-M is a lightweight version of 3APL Because the rollback operation of traditional
(Dastani et al., 2003) and distributed under the database transactions is usually not applicable in
GNU GPL. Its source code remains much simpler dynamic and open environments, it is not invoked in
by leaving out supplementary components of the prototype. Instead, if jump() succeeds, the failed
3APL, such as integrated development environ- task is allowed to commit, and its side-effects will be
ment. It behaves as a programming library whose undone by its counterpart compensating transaction.
API allows a Java application to call 3APL logic The “jump” algorithm is opportunistic, as
and deliberation structures, and vice-versa. Be- it makes the assumption that the failure of one
1094
Table 2. Algorithm 2. Pseudo-code for the travel

path of execution tree will not block others. In
agent
the worst case, where a failure state always holds
Func transactionFlightBooking() { resources required by other actions, the cost of
beginTrans(); maintaining and iterating the choice point stack
sequence of actions for flight booking;
ifaction failsthen
increases. However, as different execution paths
system.jump(); increase in independence from each other, the
commitTrans();
}
performance is improved.
Func transactionHotelBooking() {
beginTrans();
sequence of actions for hotel booking; FUtUrE trENDs
ifaction failsthen
system.jump();
commitTrans(); The shared theme of transactions, workflows,
}
and multi-agents is to ensure the system correct-
……
ness by executing concurrent tasks cooperatively
Func TravelSystem.run() {
beginTrans(); towards their goals during normal execution, and
switchsystem.select(Destination)do to guarantee the system reliability by repairing af-
caseCalifornia
transactionFlightBooking();
fected participating tasks coordinately to reach a
transactionHotelBooking(); semantically correct state when exceptions occur.
break;
As shown in Figure 6, a workflow system evolves
… ...
end switch into a multi-agent system when its operating en-
switchsystem.select(Entertainment)do vironment becomes more and more complex and
caseSailing
transactionYachtRenting(); dynamic, and condenses into a database transac-
break; tion when it obtains full control to the running
caseDisneyland
transactionTicketBuying(); environment.
break; As these three domains are closely related,
end switch
commitTrans(); more work is required to explore and identify the
} correlation among them. In fact, the boundary and
criteria to classify real applications into one of the
three types of systems are not clear.
Figure 6. Relations among transactions, workflows, and multi-agent systems
1095
To build a robust and reliable BPM system cONcLUsION

working in complex and dynamic environment,
we have to assimilate the key concepts from This chapter has described a multiple-step back-
transactions for structured concurrency and re- tracking mechanism to maintaining a tradeoff
covery management, and from multi-agents for between replanning and rigid backtracking for
adaptability and reactivity. exception handling and recovery, thus not suffer-
Due to the growing complexity and unpredict- ing from the drawbacks of rigid backtracking. Our
ability, transactions as well as their extensions can model encapsulated the business process execu-
not be straightforwardly adopted as a satisfactory tion in an open nested transaction to inherit the
solution for systems operating in dynamic and benefits of concurrency control and distribution
open environment. However, they still provide management of participating plans.
some invaluable concepts and features (e.g., failure Our approach mainly contained three parts.
atomicity, concurrency control, nested structure, We first applied BDI agent model to design and
compensation, and forward recovery) which can be construct BPM systems to operate in complex
used as components of the BPM system, especially and dynamic environments with the advantages
for the problem of error handling and recovery. of adaptability and flexibility. Then we introduced
We can integrate these essential concepts into the the “jump” approach for exception handling and
sound features, such as autonomy and reactivity, recovery which utilizes the beneficial features of
brought by agent systems, thus designing a flex- event-driven and means-end reasoning of BDI
ible and well-organized BPM recovery structure agents. Finally, we used open nested transac-
from a programming and software engineering tion model to encapsulate plan execution and
perspective. backtracking to gain the system level support
This chapter proposes an innovative backtrack- of concurrency control and automatic recovery.
ing mechanism utilizing the concept and features The nested tree structure was applied to
of BDI agent, and integrating open nested trans- depict the BPM system execution. During the
actions to support its execution. A more general construction of the tree structure along with the
expression of our algorithm can be formalized to system execution, the choice points were stored
incorporate the full range of exception handling and maintained in a stack. By iterating the stack,
strategies between the two extreme ends of replan- the system can find and execute a suitable plan
ning and rigid step-by-step backtracking. from previously applicable ones to achieve its
Our work does not go deep into the issue of goal as soon as an exception occurs. This “jump”
distributed computation which is also essential procedure struck a balance between complete re-
for building BPM systems operating in complex planning and rigid step-by-step backtracking after
and open environments. As agents are naturally exceptions occur, by utilizing previous planning
distributed entities, multi-agent concepts can be results in determining response to failure. Because
further applied to benefit BPM systems to cope the substitutable path is allowed to start prior to
with a wide range of internal and external interac- or in parallel with the compensation process, the
tions and changes. As a result, our multiple-step system can achieve its goals more directly with
backtracking mechanism needs to be further higher efficiency.
extended to embrace the distributed execution of Our approach also frees system programmers
BPM tasks. This is our future work. from considering low-level details of concur-
rency control and exception handling, because
transactional execution automates these issues.
Combining and utilizing several beneficial fea-
1096
tures of BDI agents, the open nested transaction JBoss Inc. (2006). JBoss transactions 4.2.2 trans-
model is tightly integrated into the BPM system. action core programmers guide. Retrieved May
Both BDI data structures and the deliberation 7, 2008, from JBoss Inc.: http://labs.jboss.com/
cycle are leveraged to maximize the functionality jbosstm/docs/4.2.2/manuals/pdf/core/Program-
of transaction management. mersGuide.pdf
Jennings, N. R. (2001). An agent-based ap-
proach for building complex software systems.
rEFErENcEs
d’Inverno, M., Luck, M., Georgeff, M. P., Kinny, doi:10.1145/367211.367250
D., & Wooldridge, M. (2004). The dMARS archi- Kamathy, M., & Ramamritham, K. (1996).
tecture: A specification of the distributed multi- Correctness issues in workflow management.
agent reasoning system. Journal of Autonomous Distributed Systems Engineering, 3, 213–221.
Agents and Multi-Agent Systems, 9(1-2), 5–53. doi:10.1088/0967-1846/3/4/002
doi:10.1023/B:AGNT.0000019688.11109.19
Klein, M., Rodriguez-Aguilar, J. A., & Dellarocas,
Dastani, M., Riemsdijk, B. v., Dignum, F., & C. (2003). Using domain-independent exception
Meyer, J.-J. C. (2003). A programming language handling services to enable robust open multi-
for cognitive agents goal directed 3APL. In Pro- agent systems: The case of agent death. Autono-
gramming Multi-Agent Systems (pp. 111-130). mous Agents and Multi-Agent Systems, 7(1-2),
Springer Verlag. 179–189. doi:10.1023/A:1024145408578
Ehrler, L., Fleurke, M. K., Purvis, M., & Savari- Koch, F. (2005). 3APL-M: Platform for light-
muthu, B. T. R. (2006). Agent-based workflow weight deliberative agents. Retrieved 7 May, 2008,
management systems (WfMSs). Information from http://www.cs.uu.nl/3apl-m/docs/3aplm-
Systems and E-Business Management, 4(1), 5–23. manual.pdf
doi:10.1007/s10257-005-0010-9
Nagi, K. (2001). Transactional agents: Towards
Eiter, T., Erdem, E., & Faber, W. (2004). Plan a robust multi-agent system. Berlin, Heidelberg:
reversals for recovery in execution monitoring. In Springer-Verlag.
10th International Workshop on Non-Monotonic
Reasoning (pp. 147-154). Ramamohanarao, K., Bailey, J., & Busetta, P.
(2001). Transaction oriented computational
Georgakopoulos, D., Hornick, M. F., & Manola, models for multi-agent systems. In Proceedings
F. (1996). Customizing transaction models and of 13th IEEE International Conference on Tools
mechanisms in a programmable environment with Artificial Intelligence (ICTAI) (pp. 11-17).
supporting reliable workflow automation. IEEE IEEE Computer Society, Washington, DC, USA.
Transactions on Knowledge and Data Engineer-
ing, 8(4), 630–649. doi:10.1109/69.536255 Rao, A. S., & Georgeff, M. P. (1995, June 12-14,
1995). BDI agents: From theory to practice. In
Gray, J., & Reuter, A. (1993). Transaction pro- Proceedings of the First International Confer-
cessing: Concepts and techniques. San Francisco: ence on Multiagent Systems (pp. 312-319), San
Morgan Kaufmann. Francisco, California, USA.
Hector, G.-M., & Kenneth, S. (1987). Sa-
gas. SIGMOD Record, 16(3), 249–259.
doi:10.1145/38714.38742
1097
Souchon, F., Dony, C., Urtado, C., & Vauttier, S. objectives. It has four key data structures: beliefs,
(2004). Improving exception handling in multi- desires, intentions and plans.
agent systems. In Software Engineering for Multi- Choice Point: For a certain goal of the system,
Agent Systems II (Lecture Notes in Computer if plan matcher finds more than one eligible plan,
Science Vol. 2940) (pp. 167-188). Springer Verlag. the corresponding node in the execution tree is
marked as a choice point.
Tripathi, A., & Miller, R. (2001). Exception
Choice Point Stack: A choice point stack
handling in agent-oriented systems. In Advances
contains the choice points the system has met
in exception handling techniques (LNCS 2022,
chronologically.
pp. 128-146).
Consistent: The plan execution is consistent
Unruh, A., Bailey, J., & Ramamohanarao, K. if it terminates at state Sn in which invariants are
(2004). Managing semantic compensation in a satisfied.
multi-agent system. In International Conference Correct: The plan execution is correct if it
on Cooperative Information Systems (LNCS terminates at state Sn in which both invariants and
3290). termination conditions are satisfied.
Jump: A jump is a continuation of the system
Weikum, G., & Schek, H.-J. (1992). Concepts
execution flow at another selectable plan implied
and applications of multilevel transactions and
by elements of its choice point stack.
open nested transactions. In Database transaction
Non-Over-Correction: The compensation
models for advanced applications (pp. 515-553).
process only partially or completely recovers the
San Francisco: Morgan Kaufmann Publishers, Inc.
side effects made by the normal execution. Let
Woolridge, M. (2001). Introduction to multiagent invl (r ) represent the amount of involved resource
systems. John Wiley & Sons, Inc. r in normal execution and comp(r ) is the reversal
of r in the corresponding compensation, then
comp(r ) £ invl (r ) .
Plan: A plan is an atomic unit of work from
KEy tErMs AND DEFINItIONs the observer’s point of view. It is constrained
by enabling conditions (EC), invariants (I) and
BDI agent: A software entity embedded in
termination conditions (TC).
certain environment and behaves reactively to
the changing situation in order to meet its design
Adaptability in Turbulent Environments, edited by Minhong Wang and Zhaohao Sun, pp. 251-270, copyright 2010 by Informa-
tion Science Reference (an imprint of IGI Global).
1098
1099
Chapter 4.14
A Resource-Based Perspective
on Information Technology,
Knowledge Management,
and Firm Performance
Clyde W. Holsapple
University of Kentucky, USA
Jiming Wu
California State University–East Bay, USA
AbstrAct but also by broadening our understanding of the

relationships among IT, KM, and firm performance.
The resource-based view of the firm attributes su-
perior firm performance to organizational resources
that are valuable, rare, non-substitutable, and dif- INtrODUctION
ficult to imitate. Aligned with this view, the authors
contend that both information technology (IT) and For the last two decades, the investigation of the
knowledge management (KM) comprise critical return on investments in IT has become a key
organizational resources that contribute to superior objective of many studies. In pursuing this objec-
firm performance. The authors also examine the tive, researchers have developed two main theo-
relationship between IT and KM, and develop a new retical frameworks: one asserts that IT has a direct
second-order variable – IT-KM competence – with impact on firm performance (Bharadwaj, 2000),
IT capability and KM performance as its formative while the other proposes that the effect of IT on
indicators. Thus, this chapter contributes not only by firm performance is mediated by business process
investigating the determinants of firm performance (Tanriverdi, 2005). However, no matter which
theoretical framework has been employed, some
DOI: 10.4018/978-1-60566-659-4.ch016 studies have failed to find a significant correlation
A Resource-Based Perspective on IT, KM, and Firm Performance
between IT and firm performance. Because the the research models investigating determinants of
return on IT investments seems to be contingent, superior firm performance. Thus, the work would
scholars call for more research into why IT may be of value not only to practitioners striving to
not benefit business, how to make IT effective, achieve and sustain business success, but also to
and what are the key determinants of the success researchers interested in identifying determinants
of IT (Dehning & Richardson, 2002). of better firm performance.
Meanwhile, considerable research attention This study contributes to such investigation.
has been devoted to the importance of KM in More specifically, the purpose of this chapter is to
the rapidly changing, competitive, and dynamic theorize a triangle of relationships among IT, KM,
business environment (Holsapple & Wu, 2008). and firm performance, and to develop a theoreti-
Modern organizations are turning to KM practices cal model with testable hypotheses that improve
and applications to foster the creation, integration, our understanding of the effects of IT and KM on
and usage of knowledge assets that enable them firm performance. The theoretical foundation of
to compete in an increasingly global economy. In this paper is embedded in the resource-based view
light of this, researchers have attempted to provide of the firm and prior work by Holsapple and his
empirical evidence of the strategic consequences colleagues. The current study contributes to the
that KM can bring to organizations (Grant, 1996). literature in a number of ways. First, this study
For example, based on the survey data collected is among the first to recognize that KM may play
from 177 firms, Chuang (2004) finds that greater an important role in the link between IT and firm
KM capabilities are significantly associated performance. Thus, the study may provide a plau-
with greater competitiveness and that social KM sible explanation for why some previous research
resource has a significant impact on competitive has failed to discover a significant relationship
advantage. Similarly, in a survey-based inves- between IT and firm performance. Second, we
tigation of the link between KM activities and examine the determinants of firm performance
competitiveness, Holsapple and Singh (2005) by introducing and employing a new perspective,
observe that the KM activities of interest can be which focuses on the collective impacts of IT and
performed in ways that improve organizational KM. Such a perspective may broaden our approach
competitiveness, and can do so in each/all of four to identifying determinants of firm performance.
ways: enhanced productivity, agility, innovation, Third, we present methods to measure relevant
and reputation. variables. Therefore, the current chapter is useful
Although there exist studies on IT-firm perfor- and effective in guiding future empirical research
mance relationship and on KM-firm performance in this regard. Finally, this study also investigates
link, these studies have paid insufficient attention the relationship between IT and KM, which has
to the full map of relationships among IT, KM, so far received relatively little research attention.
and firm-level return, and have placed relatively The remainder of the chapter is organized as
less emphasis on the collaborative effect of IT follows. In the next two sections, we review the
and KM on firm performance (Wu, 2008). Given state of IT and KM. Then, we present the research
the inseparability of IT and KM, and the strategic model and hypotheses, followed by a section in
importance of the two, a thorough investigation which we discuss methods for measuring the
of both their joint and separate roles in firm per- variables. Finally, we provide a brief summary
formance is necessary. Such investigation would of the contributions provided by this research.
enrich not only the theoretical understanding of
the mechanism for competitive advantage, but also
1100
INFOrMAtION tEcHNOLOGy Laudon, 2006). Today in the U.S., more than 23

million managers and 113 million workers in the
Information technology can be defined as covering labor force rely on information systems to con-
a broad range of technologies involved in informa- duct day-to-day business and to achieve strategic
tion processing and handling, such as computer business objectives (Laudon & Laudon, 2006).
hardware, software, telecommunications, and Along with the rapid growth and development
databases (Huff & Munro, 1985). Realizing that of IT, the role of IT in business has greatly ex-
IT enables businesses to run efficiently and profit- panded, ranging from simple back-office functions
ably, organizations around the world have made to enabler of business process reengineering and
tremendous investments in it. As estimated by key driver of competitive advantage. Until the
market research organizations, world IT spend- 1960s, IT had played a very simple role in business
ing in 2000 was about $2 trillion and will reach operation: transaction processing, record-keeping,
$3.3 trillion in 2008, with an average growth rate accounting, and other data processing activities; by
of over 7% in these eight years (WITSA, 2000; the late 1970s, the major role of IT began to shift
Gartner, 2007). It is also estimated that such a toward providing managerial end users with ad hoc
growth rate will be sustained for several years and interactive support of their decision-making
after 2008 (InformationWeek, 2007). In the U.S. processes; in the 1980s-1990s, IT was mainly
economy, IT spending now accounts for nearly employed to support end users to use their own
40% of overall expenditure on capital equipment, computing resources for job requirements and to
making it the largest line item in American firms’ assist top executives in easily accessing critical
budgets for capital investment (Cline & Guynes, information in their preferred formats; now the
2001). Not surprisingly, IT spending has already primary role of IT is to help develop fast, reli-
accounted for approximately 4% of U.S. gross able, and secure Internet, on which e-commerce
domestic product (GDP) (BusinessWeek, 2001). and Web-enabled enterprise are based (Laudon &
IT has profoundly changed the way that busi- Laudon, 2006). Figure 1 shows the expanding role
ness gets done in nearly every industry. Using IT, of IT in business and organizational management.
organizations radically redesign their business Because IT is an area of rapid change and growth,
processes to streamline and simplify operations it is important and necessary for organizations and
and remain competitive in a changing environ- individuals to continually adapt and develop new
ment. With the help of computer-aided design skills and knowledge.
and operational systems, organizations can greatly
reduce overall cost and time of developing and
manufacturing its products and of providing its KNOWLEDGE MANAGEMENt
services. Key customer-related IT, such as cus-
tomer relationship management systems, allows Knowledge refers to a fluid mix of framed experi-
organizations to capture and maintain detailed ence, values, contextual information, and expert
information about customer interactions, thus insight that offers a framework for interpreting,
enabling them to provide quality customer service assimilating, and integrating new experiences and
and to increase sales. As a specific category of information (Davenport & Prusak, 1998). Knowl-
IT serving middle-level managers, management edge is highly human-related. More specifically, it
information systems summarize and report on a is originated from and applied in brains of human
company’s basic operations using transaction- beings. From one perspective, knowledge is a
level data, and thus help with monitoring, control- product of human reflection and experience em-
ling, and decision-making activities (Laudon & phasizing understanding and sense making (why
1101
Figure 1. The expanding role of IT in business (O’Brien & Marakas, 2007)
and how), while information can be considered as generating new knowledge through derivation
a message focusing on the awareness of something or discovery, acquiring valuable knowledge from
(who and what) (Bennet & Bennet, 2003). Others outside sources, selecting needed knowledge from
contend that modern computer technology can also internal sources, altering the state of knowledge
make sense of situations, learn from its experi- resources, and embedding knowledge into orga-
ences, and derive/discover new knowledge – in nizational outputs (Holsapple & Joshi, 2004).
addition to message handling (Holsapple 2005). In KM is becoming increasingly important and
this vein, knowledge is something that is conveyed prevalent for many reasons. To succeed in today’s
in representations (e.g., linguistic, symbolic, digi- dynamic global economy, organizations must re-
tal, mental, behavioral, material patterns) that are duce their cycle times in production, operate with
usable to some processor (e.g., human mind) and minimum fixed assets and costs, shorten product
can be categorized as being descriptive (character- development time, improve customer service and
izations of the state of some system – who, what, product quality, enhance employee productivity
when, etc.), procedural (characterizations of how and performance, provide innovative products
to do something), or reasoning (characterizations and services, modernize and reengineer business
of logic or causality). In this view, information process, and increase agility and flexibility (Gupta
is one gradation of descriptive knowledge, but it et al., 2004). All these critical business activities
can be operated on by other types of knowledge require continued efforts to acquire, create, docu-
(i.e., procedures and logic). ment, share, and apply knowledge by employees
Holsapple and Joshi (2004, p. 593) define and teams at all organizational levels. Because
knowledge management as “an entity’s system- of the importance of KM to success, organiza-
atic and deliberate efforts to expand, cultivate, tions have invested heavily in it. According to
and apply available knowledge in ways that add IDC, global business spending on KM was rising
value to the entity, in the sense of positive results from $2.7 billion in 2002 to $4.8 billion in 2007
in accomplishing its objectives or fulfilling its (Babcock, 2004). The company also estimated that
purpose.” Thus, KM involves any activities of in the United States, KM spending reached $1.4
1102
billion in 2001 and $2.9 billion in 2006, exhibit- study recognizes three categories of KM influ-
ing an average annual growth rate of over 20% ences: managerial, resource, and environmental.
in these five years (Motsenigos & Young, 2002). The study also identifies five major knowledge
KM has also attracted tremendous attention manipulation activities: acquisition, selection,
from researchers. Figure 2 exhibits the trend of generation, assimilation, and emission, as well as
publications for KM as tracked by Google Scholar four major managerial activities that constituted
from 1995 to 2007. For each year, the number of the managerial influences: leadership, coordina-
publications referring to “knowledge manage- tion, control, and measurement.
ment” is shown. Such publications were 513 in
1995, or about 10 per week, and exponentially
increased to 12,600 in 2005, or about 243 per rEsEArcH MODEL
week. This indicates that two weeks’ publications AND HyPOtHEsEs
in 2005 are almost equal to whole year’s publica-
tions in 1995 and that the average annual growth Drawing on the resource-based view (RBV) of the
rate in these 10 years is astonishing – 236%! To put firm and prior empirical findings, we introduce a
this KM trend in perspective, we compare it with conceptual model positing that IT and KM both
the traditional business discipline of operations play an important role in predicting firm perfor-
management (OM), for which Google Scholar mance, and that KM performance is highly related
reports 1,190 publications in 1995, ramping up to IT capability. As depicted in Figure 3, the model
to 3,760 in 2005. Figure 2 also shows the trend includes a new variable – IT-KM Competence,
of publications for OM. which is also conceptualized as a key anteced-
One important research stream in this field ent of firm performance. Below we describe and
focuses on the KM ontology that offers a com- discuss the new variable and the conceptual links
prehensive understanding of KM phenomena in the research model.
(Holsapple & Joshi, 2004). While specifying
the conceptualization of the KM domain, the
Figure 2. Publication trends for knowledge management and operations management. Source: Google
Scholar, March 24, 2008
1103
Figure 3. A conceptual model of IT, KM, and firm

2004). In performing knowledge acquisition ac-
performance
tivities, an IT-based network system can assist in
identifying, evaluating, analyzing, and qualifying
external knowledge that needs to be acquired to
support the firm’s growth (Holsapple & Singh,
2003). An IT-based knowledge selection system
can help a firm be more efficient and effective in
the process of knowledge selection. For example,
Buckman Laboratories uses K’Netix, an IT-based
knowledge selection system, to locate, collect,
select, and package appropriate knowledge re-
ceived from 11 resources and transfer it to the
It capability and KM Performance person requesting the knowledge (Holsapple &
Singh, 2003).
In the last decade, more and more researchers In the activity of knowledge generation, a deci-
have adopted the notion that IT plays a critical sion support system may draw on databases and
role in shaping firms’ efforts for KM. For ex- text-bases, plus banks of solvers and/or rule sets to
ample, in a study examining the link between derive knowledge in the sense of expectations, ex-
KM and computer-based technology, Holsapple planations, evaluations, solutions, recommenda-
(2005) argues that IT is of great importance not tions, and so forth (Bonczek et al., 1981; Holsapple
only for enabling or facilitating the knowledge & Whinston, 1996). In addition, such systems can
flows among knowledge processors (human or also help in other knowledge generation activities
computer-based) but also for assisting in the such as data mining, text mining, and sense-making
measurement, control, coordination, and leader- (Jones, 2004). In the activities of knowledge as-
ship of knowledge and the knowledge processors. similation, an IT-based organizational memory
Thus, he asserts that modern KM is inseparable system can help in modeling, representing, and
from a consideration of IT. Similarly, in a study archiving knowledge, while an IT-based less
investigating the relationships among IT, KM, and structured repository (e.g., discussion database
firm performance, Tanriverdi (2005) argues that and lessons-learned system) can be used to store
an IT-based coordination mechanism can increase insights and observations (Jones, 2004). Finally,
the reach and richness of a firm’s knowledge in the process of knowledge emission, IT-based
resources, and enable business units of the firm systems can support users in sharing and transfer-
to learn about knowledge sharing opportunities ring knowledge quickly and cost-efficiently. For
with each other. Thus, he posits that IT related- instance, to enhance knowledge sharing among
ness, which is defined as “the use of common employees in geographically dispersed locations,
IT infrastructures and common IT management Honda has established a full-service international
processes across business units,” (p. 317) is posi- communications network system (called Pentac-
tively associated with KM capability. cord) and a system to manage selected databases
Not surprisingly, previous research also sug- (sales, finance, and part ordering) on a global basis
gests that IT plays an important role in supporting (Holsapple & Singh, 2003). In short, the current
and enhancing aforementioned KM activities: literature suggests that KM performance is of
acquisition, selection, generation, assimilation, particular relevance to IT. Thus, we hypothesize:
and emission (Holsapple & Singh, 2003; Jones,
1104
H1:IT capability is positively related to KM broadly – including financial capital, physical

performance. assets, knowledge, brand image, IT, organizational
processes, and so forth (Bharadwaj, 2000). Thus,
Here, IT capability refers to an organization’s a resource is an observable but not necessarily
ability to identify IT that meets business needs, tangible asset that can be independently managed,
to deploy IT to improve business process in a appraised, and even valued (Karmi et al., 2007).
cost-effective manner, and to provide long-term RBV suggests that a resource held by a majority of
maintenance and support for IT-based systems competing firms (i.e., a non-rare resource) may not
(Karimi et al., 2007). By KM performance, this explain firm performance differences (Newbert,
chapter means the degree to which KM activities 2007). It also suggests that if a resource held by
harness organizational resources to achieve the just a few competing firms is not costly to imitate,
goals or purposes of KM initiatives (Wu, 2008). the resource is likely to be quickly obtained by
Through linking IT capability to KM performance, competitors, and thus may not explain differences
the first hypothesis highlights that KM and IT in firm performance, either (Ray et al., 2005).
are inseparable, and can play a communal and
collective role in an organization. It capability and Firm Performance
the resource-based As an important firm resource, IT capability

View of the Firm plays a key role in firm performance. IT capabil-
ity enables organizations to design innovative
Rooted in management strategy literature, the products and services, and to reduce the overall
RBV of the firm is developed to understand why cost and time of developing the products and pro-
firms are able to gain and sustain a competitive viding the services. For instance, IT giant Apple
advantage (Newbert, 2007). RBV states that a developed an innovative product – iPod, which
firm’s performance is mainly determined by a has dominated digital music player sales in the
unique set of firm resources that are valuable, United States and brings the company new sales
rare, non-substitutable, and difficult to imitate. records and great business success. Continuing to
RBV indicates that such resources are often rent- innovate, the company recently released iPhone,
yielding and likely to survive competitive imitation an Internet-enabled multimedia mobile phone.
when protected by isolating mechanisms such as Computer-aided design (CAD) systems assist
resource connectedness, historical uniqueness, Toyota’s designers to create and modify their
and causal ambiguity (Barney, 1991). In short, product specifications much faster than before,
RBV addresses firm performance differences by and thus achieve cost efficiency. CAD allows a
using resource asymmetry. That is, the resources designer to see his or her ideas as they take shape
needed to achieve strategic business objectives on a monitor display, in addition to clay models.
are heterogeneously distributed across firms, and Taking the advantage of CAD, Toyota designs
thus are posited to account for the differences in quality into its products.
firm performance (Grant, 1991). IT capability is the primary driver of business
Based on the RBV, a resource can be defined process reengineering, which integrates a strategy
as a rare and inimitable firm-specific asset that of promoting business innovation with a strat-
adds value to firms’ operations by enabling them egy of making major improvements to business
to implement strategies that improve efficiency process so that a company can gain and sustain
and effectiveness (Wade & Hulland, 2004). Ad- competitiveness (O’Brien & Marakas, 2007). The
vocates of RBV tend to characterize resources computation capability, information processing
1105
speed, and connectivity of computers and Inter- performance by allowing companies to achieve six
net technologies can considerably enhance the major business values: (1) generate new revenue
efficiency of a business process, as well as com- from online sales, (2) reduce costs via online trans-
munications and collaboration among the people action, (3) attract new customers through online
responsible for its management, implementation, marketing and advertising, (4) increase customer
and maintenance (Wade & Hulland, 2004). For loyalty and retention by providing Web-based
example, many Fortune 500 companies count on customer service and support, (5) develop new
enterprise resource planning (ERP) systems to Web-based markets and distribution channels, and
reengineer, automate, and integrate their market- (6) develop and sell digital goods such as music
ing, manufacturing, sales, distribution, finance, track, video stream, online game, and flight ticket
and human resource business processes (O’Brien (O’Brien & Marakas, 2007; Wheeler, 2002).
& Marakas, 2007). IT capability can be a key enabler of superior
IT capability used for production and op- firm performance by improving communication
erations can improve performance of companies and collaboration within an organization. IT,
that must plan, monitor, and control inventories, especially network technologies, provides basic
facilities, and the flow of products and services. infrastructure and platform for communication,
Many manufacturing and production systems can coordination, and collaboration among the mem-
efficiently deal with the operation and mainte- bers of business teams and workgroups. In other
nance of production facilities; the establishment words, such IT capability enables employees and/
of production goals; the acquisition, storage, or managers at all levels to work together more
and distribution of production materials; and the easily and effectively by helping them share infor-
scheduling of equipment, facilities, materials, mation with each other, coordinate their individual
and labor required to fulfill an order (Laudon efforts and use of resources, and work together
& Laudon, 2006). Thus, computer-integrated cooperatively on joint projects and assignments
(or –aided) manufacturing enables organizations (O’Brien & Marakas, 2007). For example, knowl-
to reduce the cost and time of producing goods edge experts, technicians, computer specialists,
by simplifying, automating, and integrating all and R&D engineers may form a virtual team for
production and support processes (O’Brien & a KM system development project. The commu-
Marakas, 2007). Moreover, such manufacturing nication, coordination, and collaboration among
helps companies achieve highest product quality the team members may rely heavily on IT-based
by bridging the gap between the conceptual design applications such as email, instant messaging,
and the manufacturing of finished goods. newsgroup, videoconferencing, discussion forum,
IT capability is the key to electronic commerce, and a Web-based database for convenient and im-
which refers to the use of digital technology and mediate access to work-in-progress information.
the Internet to execute major business processes Such improved communication and collaboration
of developing, marketing, selling, delivering, can significantly increase the quality of the team
servicing, and paying for products and services work.
(Laudon & Laudon, 2006; O’Brien & Marakas, Adopting the resource-based view of the firm,
2007). Electronic commerce is transforming information systems researchers suggest that IT
firms’ relationships with customers, employees, capability has an impact on firm performance.
suppliers, distributors, retailers, and partners into For example, Mata and colleagues (1995) point
digital relationships using networks and the In- out that managerial IT skills are scarce and firm
ternet (Laudon & Laudon, 2006; Wheeler, 2002). specific, and thus likely to serve as sources of
More important, it can dramatically improve firm sustained competitive advantage. Focusing on
1106
the differential effects of various IT resources is likely to stimulate organizational learning be-
on customer service performance, Ray and col- cause individuals and teams become to believe
leagues (2005) argue that such factors as IT are that learning can help their company to handle
valuable resources because they enable firms to change, uncertainty, and complexity in the ever-
increase the efficiency or effectiveness of business changing business environment. KM helps define
processes compared to what would be the case and specify what should be learned, when it should
if these resources were not exploited. Similarly, be learned, and who should be learning it. KM
Bharadwaj (2000) contends that organizations can also create a culture of peer collaboration and
successful in developing superior IT capability open communication, both leading to a setting
will enjoy superior financial performance by conducive to organizational learning. Moreover,
boosting revenues and/or reducing costs. In line KM activities of knowledge acquisition and
with the resource-based view of the firm and the generation promote organizational learning by
literature, we therefore hypothesize: motivating individuals to obtain new knowledge
from external sources or from existing knowledge,
H2:IT capability is positively related to firm and to make it suitable for future use.
performance. KM can improve firm performance not only
by facilitating organizational learning but also by
KM Performance and encouraging knowledge sharing. A core principle
Firm Performance of KM is to make knowledge sharing easier and
timely, and to encourage employees and manag-
As mentioned earlier, the RBV indicates that ers to work together in ways that will incorporate
knowledge is a unique company resource (Grant, knowledge shared among them. Consequently,
1996). Therefore, KM can also be viewed as such one important goal of KM is to boost productivity
resource important to firm performance because it and efficiency by building a set of methods and
allows the firm to better leverage its knowledge. tools to foster appropriate flows of knowledge.
KM facilitates organizational learning, which For instance, to align with the strategy of pos-
keeps organizations in tune with trends and sessing a platform for quick and easy knowledge
developments in their business, and thus helps sharing on global scale, Xerox developed Eureka,
them perform better. Here, organizational learn- an intranet based communication system, in 1996
ing refers to individual learning, team learning (Barth, 2000). The system is linked with a corpo-
(i.e., learning in small or large groups), or entire rate database that helps service technicians share
organization-level learning (Bennet & Bennet, repair tips. There are more than 36,000 tips in the
2003). All these levels of learning are necessary system which can be accessed by about 19,000
for an organization eager to possess the requisite Xerox technicians via their laptop computers
knowledge to improve performance. From a KM (Barth, 2000). The increasing importance of KM
perspective, organizational learning is critical and also motivates managers to develop a reward and
should be nurtured and made an integral part of personnel evaluation structure favoring knowl-
KM strategy. Organizational learning also reflects edge sharing activities. Reward and punishment
an organization’s capacity to acquire or generate standards help define acceptable behavior. By
the knowledge necessary to survive and compete incorporating desired KM behavior into annual
in its environment (Bennet & Bennet, 2003). performance evaluation, an organization may
KM can change an employee’s attitude toward improve its own performance by encouraging
learning and its impact on an organization’s such critical activities as knowledge sharing and
competitive position (Wu, 2008). Such change foregoing organizational learning.
1107
KM can strengthen an organization’s competi- nization. More specifically, KM activities such as

tive position by increasing its agility (Holsapple knowledge generation and sharing can broaden
& Singh, 2003). In general, agility refers to an understanding of relevant issues and concepts, and
organization’s ability to detect changes, oppor- push thinking beyond the constraints of presump-
tunities, and threats in its business environment tion, narrow rationality, and traditional method.
and to provide speedy and focused responses to Therefore, KM can be an important organizational
customers and other stakeholders by reconfigur- practice that spurs innovation.
ing resources and processes and/or by developing In summary, RBV suggests that a firm can
strategic partnerships and alliances (Mathiyalakan outperform its competitors by taking advantage of
et al., 2005). Thus, agility derives from both the its KM. As a unique company resource, KM plays
physical ability to act and the intellectual ability to a fundamental role in firm performance because
understand appropriate things to act upon (Dove, it facilitates organizational learning, encourages
2003). KM is recognized as a key success factor for knowledge sharing, increases agility, and fosters
agility because it enables an organization to apply innovation. Although it is complex to acquire and
effectively its knowledge of market opportunity, difficult to leverage KM resources, firms that suc-
production process, business practice, cutting- ceed in doing so are likely to experience learning
edge technology, quality service, management effects whereby they improve their abilities for
skills, the extent of a threat, and so forth. In a creating value. This directly leads to the follow-
continuously changing and unpredictable business ing hypothesis:
environment, it is crucial for an organization to
manage knowledge in a way to quickly absorb H3:KM performance is positively related to firm
new knowledge, fully assimilate it, and effectively performance.
exploit it (Holsapple & Wu, 2008). Consequently,
an organization with sufficient competencies in It-KM competence and
KM will be agile enough to deliver leading edge Firm Performance
and achieve a better competitive position.
KM can also improve an organization’s per- IT-KM competence is defined as a firm’s IT and
formance by fostering its innovation. As a subject KM ability and resources that are peculiar to
of research and practice, innovation refers to the achieving and sustaining business success. The
ability of creating valuable and useful new prod- new variable is conceptualized as a composite
uct, new services, new technology, or production construct with IT capability and KM performance
processes (Liao & Chuang, 2006). Innovation has as its two formative indicators. Such conceptu-
been recognized as a primary value creator for alization is in line with prior research and RBV,
organizations, in both times of generating revenues which suggest that KM and IT are inseparable
and in times of cutting costs. Innovation consists from each other and both are unique and impor-
of two important dimensions: magnitude, which tant firm resources. Thus, the current literature
reflects the extent or breadth of innovation, and supports the idea to represent IT capability and
speed, which shows an organization’s quickness KM performance by a single composite construct
to adopt an innovation, relative to its competitors that impacts firm performance. We contend that
(Liao & Chuang, 2006). KM plays a critical role such conceptualization can push our thinking
in the ability of an organization to be innovative beyond current theoretical boundaries and offer
because KM initiatives and activities often serve a new perspective for investigating determinants
as a key platform for creating new and inventive of firm performance. Thus, we advance the fol-
ideas that will benefit and add value to the orga- lowing hypothesis:
1108
H4:IT-KM competence is positively related to by survey questionnaire or results of studies

firm performance. conducted by public independent organizations.
Tanriverdi (2005) has developed a survey instru-
MEAsUrING tHE VArIAbLEs ment to assess the extent to which an organization
creates, transfers, integrates, and leverages related
Firm performance can be measured in a variety product, customer, and managerial knowledge
of ways, including financial performance, market resources. KM performance data may also be
performance, and business process performance. obtained by collecting and analyzing results of
Financial performance is usually evaluated by relevant studies conducted by independent KM
means of standard profit and cost ratios, which can research organizations such as KMWorld (http://
be calculated by using accounting data obtained www.kmworld.com), and Teleos and its KNOW
from Standard & Poor’s COMPUSTAT. A com- Network (http://www.knowledgebusiness.com).
mon way to assess market performance is to use
Tobin’s q, which can also be calculated by using
COMPUSTAT data. However, one factor research- cONcLUsION
ers need to be aware of is that for private firms and
not-for-profit organizations, accounting data are Over the past decade, one of the most striking
not readily available in COMPUSTAT. Perceived developments in business has been the rapid
business process performance can be evaluated by proliferation of KM. Organizations have launched
using a survey questionnaire. Often, researchers KM initiatives to consolidate and reconcile knowl-
can find well-developed survey instruments in the edge assets that enable them to compete in the
literature and adapt them for their specific needs. dynamic and changing global business environ-
In addition, it is very important to address data ment. Therefore, in parallel to the focus on the
validity and reliability issues when using survey relationship between IT and firm performance,
data to test research hypotheses. the role of KM in firm profitability has also re-
Past research suggests that IT capability can ceived considerable research attention. Drawing
be measured by IT spending/use, survey question- on the RBV of the firm, plus findings from prior
naire, or results of studies conducted by public research, this chapter argues that both IT capabil-
independent organizations. IT spending/use data ity and KM performance are primary antecedents
are often available in annual corporate financial of firm performance and that IT capability has
reports. InformationWeek and ComputerWorld are a significant impact on KM performance. The
the two publicly available sources of data on cor- current chapter also introduces a new composite
porate IT spending and other measures of IT use. variable – IT-KM competence – with IT capability
Survey instruments for some constructs related to and KM performance as its formative indicators.
IT capability have already been developed and As a result, this chapter broadens our understand-
applied to practice by prior IS research such as ing of the relationships among IT, KM, and firm
the aforementioned study by Tanriverdi (2005). performance by (1) viewing both IT and KM as
Results of independent organizations’ studies are unique and important firm resources, (2) suggest-
also a very valuable source for IT capability data. ing that KM can play a mediating role between IT
For example, the IT leader study by Information- and firm performance, and (3) proposing that IT
Week may provide the data useful for measuring and KM may be represented by a single composite
an organization’s capability to leverage its IT variable, which might play a more important and
resources on a continuous basis. Past research effective role in predicting firm performance.
suggests that KM performance can be measured
1109
NOtE Cline, M. K., & Guynes, C. S. (2001). A study

of the impact of information technology invest-
Authors are listed alphabetically and have con- ment on firm performance. Journal of Computer
tributed equally to this chapter. Information Systems, 41(3), 15–19.
Davenport, T. H., & Prusak, L. (1998). Working
knowledge. Cambridge, MA: Harvard Business
rEFErENcEs
School Press.
Babcock, P. (2004). Shedding light on knowledge Dehning, B., & Richardson, V. J. (2002). Returns
management. HRMagazine, 49(5), 46–50. on investments in information technology: A re-
Barney, J. (1991). Firm resources and sustained search synthesis. Journal of Information Systems,
competitive advantage. Journal of Management, 16(1), 7–30. doi:10.2308/jis.2002.16.1.7
17(1), 99–120. doi:10.1177/014920639101700108 Dove, R. (2003). Knowledge management and
Barth, S. (2000). Eureka! Xerox has found it. agility: Relationships and roles. In C. W. Holsapple
Retrieved from http://kazman.shidler.hawaii.edu/ (Ed.), Handbook on knowledge management 2:
eurekacase.html Knowledge directions (pp. 309-330). Berlin/
Heidelberg, Germany: Springer-Verlag.
Bennet, A., & Bennet, D. (2003). The partnership
between organizational learning and knowledge Gartner. (2007). Gartner says worldwide IT spend-
management. In C. W. Holsapple (Ed.), Handbook ing to surpass $3 trillion in 2007. Retrieved from
on knowledge management 1: Knowledge mat- http://www.gartner.com/it/page.jsp?id=529409
ters (pp. 439-455). Berlin/Heidelberg, Germany: Grant, R. M. (1991). The resource-based theory of
Springer-Verlag. competitive advantage: Implications for strategy
Bharadwaj, A. S. (2000). A resource-based per- formulation. California Management Review,
spective on information technology capability and 33(3), 114–135.
firm performance: An empirical investigation. MIS Grant, R. M. (1996). Prospering in dynamically-
Quarterly, 24(1), 169–196. doi:10.2307/3250983 competitive environments: Organizational capa-
Bonczek, R., Holsapple, C., & Whinston, A. bility as knowledge integration. Organization Sci-
(1981). Foundations of decision support systems. ence, 7(4), 375–387. doi:10.1287/orsc.7.4.375
New York: Academic Press. Gupta, J. N. D., Sharma, S. K., & Hsu, J. (2004).
BusinessWeek. (2001). How bad will it get? An overview of knowledge management. In J. N.
Retrieved from http://www.businessweek.com/ D. Gupta, S. K. Sharma (Eds.), Creating knowl-
magazine/content/ 01_11/b3723017.htm edge based organizations (pp. 1-29). Hershey,
PA: Idea Group Inc.
Chuang, S. (2004). A resource-based perspective
on knowledge management capability and com- Holsapple, C., & Whinston, A. (1996). Decision
petitive advantage: An empirical investigation. support systems: A knowledge-based approach.
Expert Systems with Applications, 27(3), 459–465. Minneapolis, MN: West Publishing.
doi:10.1016/j.eswa.2004.05.008 Holsapple, C. W. (2005). The inseparability of mod-
ern knowledge management and computer-based
technology. Journal of Knowledge Management,
9(1), 42–52. doi:10.1108/13673270510582956
1110
Holsapple, C. W., & Joshi, K. D. (2004). A for- Karimi, J. K., Somers, T. M., & Bhattacherjee, A.
mal knowledge management ontology: Conduct, (2007). The role of information systems resources
activities, resources, and influences. Journal of in ERP capability building and business process
the American Society for Information Science outcomes. Journal of Management Information
and Technology, 55(7), 593–612. doi:10.1002/ Systems, 24(2), 221–260. doi:10.2753/MIS0742-
asi.20007 1222240209
Holsapple, C. W., & Singh, M. (2003). The Laudon, K. C., & Laudon, J. P. (2006). Manage-
knowledge chain model: Activities for competi- ment information systems: Managing the digital
tiveness. In C. W. Holsapple (Ed.), Handbook on firm. NJ: Prentice Hall.
knowledge management 2: Knowledge directions
Liao, C., & Chuang, S. (2006). Exploring the role
(pp. 215-251). Berlin/Heidelberg, Germany:
of knowledge management for enhancing firm’s
Springer-Verlag.
innovation and performance. In Proceedings
Holsapple, C. W., & Singh, M. (2005). Perfor- of the 39th Hawaii International Conference on
mance implications of the knowledge chain. In- System Sciences.
ternational Journal of Knowledge Management,
Mata, F. J., Fuerst, W. L., & Barney, J. B. (1995).
1(4), 1–22.
Information technology and sustained competitive
Holsapple, C. W., & Wu, J. (2008). In search of a advantage: A resource-based analysis. MIS Quar-
missing link. Knowledge Management Research terly, 19(4), 487–505. doi:10.2307/249630
& Practice, 6(1), 31–40. doi:10.1057/palgrave.
Mathiyalakan, S., Ashrafi, N., Zhang, W., Waage,
kmrp.8500170
F., Kuilboer, J., & Heimann, D. (2005, May 15-18).
Huff, S. L., & Munro, M. C. (1985). Informa- Defining business agility: An exploratory study.
tion technology assessment and adoption: A In Proceedings of the 16th Information Resource
field study. MIS Quarterly, 9(4), 327–340. Management Association International Confer-
doi:10.2307/249233 ence, San Diego, CA.
InformationWeek. (2007). Global IT spending to Motsenigos, A., & Young, J. (2002). KM in the
reach $1.48 trillion in 2010, IDC says. Retrieved U.S. government sector. KMWorld. Retrieved
from http://www.informationweek.com/news/ from http://www.kmworld.com/Articles/Edito-
management/outsourcing/showArticle.jhtml;js rial/Feature/KM-in-the-U.S.-government-sec-
essionid=JB4IADNZACBP2QSNDLRSKH0C tor-9397.aspx
JUNN2JVN?articleID=196802764&_reques-
Newbert, S. L. (2007). Empirical research on the
tid=885619
resource-based view of the firm: An assessment
Jones, K. G. (2004). An investigation of activi- and suggestions for future research. Strategic Man-
ties related to knowledge management and their agement Journal, 28(2), 121–146. doi:10.1002/
impacts on competitiveness. Unpublished doctoral smj.573
dissertation, University of Kentucky.
O’Brien, J. A., & Marakas, G. (2007). Introduction
to Information Systems (13th ed.). McGraw-Hill/
Irwin.
1111
Ray, G., Muhanna, W. A., & Barney, J. B. (2005). tion processing and handling, such as computer
Information technology and the performance of hardware, software, telecommunications, and
the customer service process: A resource-based databases (Huff & Munro, 1985).
analysis. MIS Quarterly, 29(4), 625–652. IT Capability: Refers to an organization’s
ability to identify IT meeting business needs,
Tanriverdi, H. (2005). Information technology
to deploy IT to improve business process in a
relatedness, knowledge management capability,
cost-effective manner, and to provide long-term
and performance of multibusiness firms. MIS
maintenance and support for IT-based systems
Quarterly, 29(2), 311–334.
(Karimi et al., 2007).
Wade, M., & Hulland, J. (2004). Review: The IT Relatedness: Is defined as “the use of com-
resource-based view and information systems mon IT infrastructures and common IT manage-
research: Review, extension, and suggestions for ment processes across business units” (Tanriverdi
future research. MIS Quarterly, 28(1), 107–142. 2005, p. 317).
IT-KM Competence: Is defined as a firm’s IT
Wheeler, B. C. (2002). BEBIC: A dynamic ca-
and KM ability and resources that are peculiar to
pabilities theory for assessing net-enablement.
achieving and sustaining business success.
Innovation: Refers to the ability of creating
doi:10.1287/isre.13.2.125.89
valuable and useful new product, new services,
WITSA. (2000). Digital planet 2000. Arlington, new technology, or production process (Liao &
VA: World Information and Technology Services Chuang, 2006).
Alliance. Knowledge: Refers to a fluid mix of framed
experience, values, contextual information, and
Wu, J. (2008). Exploring the link between
expert insight that offers a framework for interpret-
knowledge management performance and firm
ing, assimilating, and integrating new experiences
performance. Unpublished doctoral dissertation,
and information (Davenport & Prusak, 1998).
University of Kentucky.
Knowledge Management: Is “an entity’s sys-
tematic and deliberate efforts to expand, cultivate,
and apply available knowledge in ways that add
KEy tErMs AND DEFINItIONs value to the entity, in the sense of positive results
in accomplishing its objectives or fulfilling its
Agility: Refers to an organization’s ability to purpose” (Holsapple and Joshi 2004, p. 593).
detect changes, opportunities, and threats in its KM Performance: Is the degree to which
business environment and to provide speedy and KM activities harness organizational resources to
focused responses to customers and other stake- achieve the goals or purposes of KM initiatives.
holders by reconfiguring resources and processes A Resource: Can be defined as a rare and
and/or by developing strategic partnerships and inimitable firm-specific asset that adds value to
alliances (Mathiyalakan et al., 2005). firms’ operations by enabling them to implement
Information Technology: Can be defined as a strategies that improve efficiency and effective-
broad range of technologies involved in informa- ness (Karmi et al., 2007).
This work was previously published in Handbook of Research on Contemporary Theoretical Models in Information Systems,
edited by Yogesh K. Dwivedi, Banita Lal, Michael D. Williams, Scott L. Schneberger and Michael R. Wade, pp. 296-310,
copyright 2009 by Information Science Reference (an imprint of IGI Global).
1112
1113
Chapter 4.15
A Decision Support System for
Selecting Secure Web Services
Khaled M. Khan
Qatar University, Qatar
INtrODUctION determining factors in selecting appropriate Web

services. The need for run-time composition of
Web service is becoming an important area of enterprise systems with third-party Web services
business processing and research for enterprise requires a careful selection process of Web ser-
systems. Various Web service providers currently vices with security assurances consistent with the
offer diverse computing services ranging from en- enterprise business goal. Selection of appropriate
tertainment, finance, and health care to real-time Web services with required security assurances
application. With the widespread proliferation of is essentially a problem of choice among several
Web Services, not only delivering secure services alternative services available in the market. The IS
has become a critical challenge for the service managers have little control of the actual security
providers, but users face constant challenges in behavior of the third-party Web services, however,
selecting the appropriate Web services for their they can control the selection of right services which
enterprise application systems. Security has become could likely comply their security requirements.
an important issue for information systems (IS) Selecting third-party Web services arbitrarily over
managers for a secure integration of Web services the Internet is critical as well as risky.
with their enterprise systems. Security is one of the With increasing security challenges to the en-
terprise systems, there is a need for an automatic
DOI: 10.4018/978-1-59904-843-7.ch025 decision support system (DSS) for the selection of
A Decision Support System for Selecting Secure Web Services
appropriate secure Web services. A DSS analyzes bAcKGrOUND

security profiles of candidate Web services and
compares them with the security requirements of Making decisions concerning the selection of Web
the enterprise system. The IS managers can make services with security compliances often strains the
decisions from such systems more easily regarding cognitive capabilities of the IS managers because
which Web service is to be integrated with their many complex attributes are involved. Analyzing
applications. A DSS could make a comparative these complex attributes and predicting the se-
analysis of various security properties between a curity outcome of independent Web services is a
candidate Web service and the enclosing enterprise daunting task. The human intuitive judgment and
system including the consequences of different decision making capability is rather limited, and
decision alternatives in selecting Web services. this ability deteriorates further with the complexity
It could also project the likely additional security of assessing security issues manually. The final
properties needed for the system if the candidate decision to select a particular Web service for an
Web service lacked required properties. The enterprise system is critical because such a deci-
complex nature of selecting secure Web services sion is considerably influenced by many complex
could not be easily managed without such a DSS security attributes of the service. A computer-
support. With the rapidly evolving nature of se- aided decision making process may manage this
curity contexts in the field of enterprise systems, complexity in a more optimal way. One of many
decision support systems for selecting secure Web decision-making approaches in which decisions
services can play an increasingly important role. are made with the help of computer-aided process
This article proposes an architecture of an easy- is generally called decision support system (DSS).
to-use security decision support system (SDSS) for A DSS can take many different forms. In general,
selecting Web services with security assurances a DSS is a computerized system for helping people
consistent with the enterprise business goal. The make decisions (Alter, 1980; Power, 1997, 2007).
SDSS stores security profiles of candidate Web According to Finlay (1994) and Turban (1995),
services, compares properties with the security a DSS is an interactive, flexible, and adaptable
requirements of the enterprise system, and gener- computer-based information system, especially
ates alternatives with consequences. Supporting the developed for supporting the decision making.
choice making process involves the evaluation and In our context in this article, we emphasize a
comparison of alternative Web services in terms of knowledge-driven decision that helps managers to
their security properties. To minimize the risks of make a choice between alternative Web services
selecting the wrong Web services for the enterprise based on their supporting security properties. It
systems, the SDSS can provide managers with con- is an interactive computer-based system that aids
sistent and concise guidance for the development IS managers in making judgments and choices
of security criteria. Our proposed SDSS has been regarding the selection of Web services which
developed to provide IS managers with information match their expectation. This article focuses pri-
necessary to make informed decisions regarding the marily on the components that process various
selection of Web services. The basic components criteria against the provided data and generates
of the SDSS include a knowledge base of various best alternatives.
security properties and an inference mechanism During the process of selecting appropriate
which uses a set of rules. The architecture consists Web services for the enterprises, IS managers often
of three components: (i) Defining security criteria; make decisions on which Web services should be
(ii) Security profiling of Web services; and (iii) integrated with their application. Considering the
Generating alternatives. value of the information assets of the organizations,
1114
it is unlikely that managers only assess the business ware component indicating a relative strength of
functionalities that Web services provide for their the security properties of the component. Payne
organizational need. They should also consider the (2002) proposes a seven step methodology to guide
security implications of using Web services with the process of defining security metrics. Berinato
their applications. The decision making process (2005) argues for a constant measure of security
of selecting security-aware Web services requires incidents and it could be used to quantify the ef-
a systematic approach. A decision support system ficiency of the deployed security functions. The
could aid managers with an automated system. National Institute of Standards and Technology
In the current practice, IS managers use Web (Swanson, Bartol, Sabato, Hash, & Graffo, 2003)
services without properly assessing the security defines a security metrics guide for information
compliances of the services (Khan, 2006, 2007). technology systems. The document provides
Managers could use decision support systems guidance on how an enterprise through the use
which could significantly improve the selection of metrics identifies the security needs, security
process of secure Web services. controls, policies, and procedures.
Many decision making techniques already
published in the literature can be used for the
selection of an entity among various alternatives. sEcUrIty DEcIsION
Classical multi-attribute utility theory (MAUT) sUPPOrt systEM (sDss)
by Keeney and Raiffa (1976), analytical hierarchy
process (AHP) by Saaty (1980), and a recent ap- The environment of the proposed SDSS system
proach by Besharati, Azram, and Kannan (2005) consists of a preprocess and the architecture
for the selection of product design are among these as illustrated in Figure 1. The preprocess has
approaches. MAUT has been used in many ap- two related activities: (i) specification of the
plication domains for modeling decision maker’s security requirements or security objectives
preferences for ranking a set of alternative deci- of the identified functionality, and (ii) gather-
sions. AHP has been extensively used in marketing security properties of the candidate Web
ing and management areas. However, using any services. These two preprocess activities are
of these models in security issues has not been required in the process of constructing the
reported yet. An agent-based DSS methodology alternatives. Managers identify what security
reported in Choi, Kim, Park, and Whinston (2004) requirements their enterprise system needs
and a market-based allocation methodology from the Web services. This essentially sets
(Parameswaran, Stallaert, & Whinston, 2001) are the security requirements consistent with the
not also applicable in the security arena. Although enterprise-wide security requirements and as-
these two DSS methodologies are used in product surances. For example, a financial enterprise
selection, their applicability in the selection of system needs a service for calculating tax offset.
secure software product is limited. Many third-party Web services may offer this
Most research works in the area of Web ser- functionality with various security functions
vices security often focus on how to make Web such as confidentiality of data, integrity of the
services secure. Some papers propose various calculated data, and so forth. The managers must
security metrics models as reported in Berinato determine what type of security the identified
(2005) and Payne (2002). Khan (2006) proposes functionality will have. Security requirements
a scheme for assessing security properties of related to the functionality may be determined
software components. The assessment scheme based on the threats, vulnerabilities, and risks
provides a numeric score to the candidate soft- associated with the functionality.
1115
Figure 1. The environment of security decision

tHE ArcHItEctUrE
support system (SDSS)
A DSS can consist of various types of components
based on the domain of application and the type
of system. Three fundamental components of DSS
are identified by Sprague and Carlson (1982): the
database management system, the model-based
management system, and the user interface. Haag,
Cummings, McCubbrey, Pinsonneault, and Dono-
van (2000) decompose these three components
into more detail. Consistent with these, Figure 2
depicts three major components of the proposed
The selection of a list of candidate Web ser-
SDSS. The identified security requirements and
vices is based on the identified functionality. At
the security properties supported by each candi-
this stage the conformity of the defined security
date Web service are the main input to the system.
requirements is not checked. We select only those
These data are structured and mapped into the
Web services which can provide the desired func-
same format by the system. The security profiles
tionality such as calculating tax offset. This also
are stored in the knowledge base. The output of the
involves gathering information regarding the
system is a rating of the candidate Web services.
security properties of the candidate Web services.
We call it security profiling. In this activity, se-
curity properties of each candidate Web service
are collected from the published security claim, DEFINING sEcUrIty crItErIA
available user’s guide, and from enquiries (Khan,
This component of SDSS defines the criteria for
2006). This profiling process enlists all security
the security requirements in a structured form
properties supporting the claimed security func-
based on the information provided by the IS
tion of each Web service.
managers. The elements of security criteria are
defined according to the ISO/IEC Standard 15408,
the Common Criteria (CC) for Information Tech-
nology Security Evaluation (Common Criteria,
Figure 2. A decision support system for selecting secure Web services
1116
1999). The CC provides evaluation measures of The example in Figure 3 illustrates the secu-
security by means of a common set of require- rity requirements of the functionality calculate
ments of the security functions of a system, and a tax. The functionality has two security require-
set of evaluation measures. The entire approach is ments: user data protection and authentication,
quantitative and it describes the security behavior as shown in level-2 nodes. Each of the two re-
of functions expected of an information system. quirements has equal importance as shown with
CC gives a comprehensive catalogue of high the value 50. The user data protection has one
level security requirements and assurances for security property as shown at level-2 node:
information systems products. CC consists of 11 encrypted(amount, ID). This property is further
classes for generic grouping of similar types of decomposed into two attributes: the key length
security requirements: security audit, communica- should be 128-bit and encryption algorithm should
tion, cryptographic support, user data protection, be RSA. The requirement authentication has a
identification and authentication, security man- security attribute: digital_signature(service_pro-
agement, privacy, protection of system security vider). This property has one attribute: algorithm
functions, resource utilization, system access, and used which should be RSA. This information will
trusted path and channels. be stored in the knowledge base.
The criteria are formulated in a tree structure. In this example, several key issues are ad-
The idea is to decompose a complex decision mak- dressed in order to check the conformity of user
ing component into simpler subcomponents that data protection and authentication. For instance,
are easily quantifiable and comparable in terms of if the user data protection is achieved by means of
values. In other words, security requirements are encrypted data, then next issues include what is
modeled into smaller subcomponents using a tree the length of the key and which algorithm is used
structure. The root of the tree (level 0) is the name in the encryption. These are the specific criteria
of the functionality. Each node at level-1 represents to check whether a Web service could provide
a CC security class such as user data protection, the required functionality with the desired level
authentication, security audit, and so forth. Level-2 of confidentiality of tax data.
and lower level nodes represent the decomposed
security attributes of the requirements. The value
associated with the each level-1 node signifies the stOrING sEcUrIty PrOFILE
relative importance of the node compared to other
nodes at the same level. All values irrespective of the The security information about each Web ser-
total number of nodes at level-1 must be summed vice gathered in the preprocess are entered into
up to 100. Figure 3 illustrates an example. the system. The format of data is the same as
Figure 3. An example of security criteria
1117
shown in Figure 3. SDSS provides a template in calculated on the basis of the presence or absence
order to capture the security requirements of the of certain security properties in the Web services.
candidate services. Let us consider a scenario. For example, if a node with a value 40 at level-1
Assume we have selected three candidate Web has full compliance with the security criteria, the
services called A, B, and C. All these services point will be 40 out of 100.
provide the same functionality, calculate tax, but IS managers can evaluate the rating of the
with varying security properties. Web service A candidate services and make a decision regard-
supports user data protection with encryption using which Web service will be selected. They can
ing a key length of 256-bit, but does not provide also modify or adjust the security requirements
authentication. Service B provides authentication differently and get a different rating of the same
with digital signature using RSA algorithm, but set of Web services. However, it is also possible
does not support user data protection. All of the to add newly identified security properties with
data are entered into the system. the profile of a candidate Web service and process
Logic programming (Baral & Gelfond, 1993) the properties for a new rating.
can be used as a formal reasoning tool to character-
ize the security requirements and security profiles
of the candidate Web services. The simple struc- FUtUrE trENDs
ture of the logic program allows us to represent a
complicated form of security knowledge and its We plan to implement the proposed approach as a
properties, and yet it is based on mathematical prototype in order to test its applicability. We are
logic (Das, 1992). In logic programming, security currently evaluating some existing tools such as
properties can be expressed in symbolic notations smodels and lparse (Syrjanen, 2000) which could
such as encrypted, digital_signature. Some of be utilized as the supporting components of the
the properties are identical with those defined in SDSS in order to facilitate the inference mecha-
BAN logic (Burrows, Abadi, & Needham, 1989). nisms. The approach can be further expanded
to select Web services with other nonfunctional
properties such as usability, maintainability, reus-
GENErAtING ALtErNAtIVEs ability, and so forth. Further research includes the
automatic verification of the security profile of
This is the heart of the SDSS. This component Web services with their implementation. The more
produces a comparative rating of all candidate Web challenging research could be directed to automate
services based on the conformity of their security the significant activities of the preprocess, such
profiles with the enterprise-wide security criteria. as automatic gathering of security profiles about
Inference rules are used to calculate the deviation the Web services.
between the security criteria and the profiles. This
component uses a rule-based approach in order
to make a comparative rating of the candidate cONcLUsION
services. Security properties are reasoned about
with the inference rules of logic programming. The article has presented a framework for a se-
The inference rules are applied to check whether curity decision support system for selecting Web
the security profile of a Web service is matched services with appropriate security assurances.
with the security requirements of the enterprise The article argues that the selection process of
system. The rating of a candidate service is based appropriate Web services for the enterprise ap-
on a value ranging from 0 to 100. The points are plication needs an automatic tool support such as
1118
SDSS. The approach could be automated to aid Common Criteria. (1999). Common criteria for
enterprise decision support systems. This can be information technology security evaluation (ISO/
used in strategic planning of information systems IEC 15408). NIST. Retrieved December 6, 2007,
as well. The main contribution of this article is from http://csrc.nist.gov/cc/
a framework for the assessment of Web services
Das, S. K. (1992). Deductive databases and logic
security properties on which further work could
programming. Addison-Wesley.
be initiated. The evaluation method presented
here is considered flexible enough, as security Finlay, P. N. (1994). Introducing decision support
requirements can be altered by the IS managers systems. Oxford, UK, Cambridge, MA: NCC
as they see appropriate for their needs. Blackwell; Blackwell Publishers.
Haag, S., Cummings, M., McCubbrey, D., Pinson-
neault, A., & Donovan, R. (2000). Management
rEFErENcEs
information systems: For the information age,
Alter, S. L. (1980). Decision support systems: Cur- (pp. 136-140). McGraw-Hill Ryerson Limited.
rent practice and continuing challenges. Reading, Keeney, R. L., & Raiffa, H. (1976). Decisions
MA: Addison-Wesley Pub. with multiple objectives. New York: John Wiley
Baral, C., & Gelfond, M. (1993). Representing and Sons.
concurrent actions in extended logic programming. Khan, K. (2007). Selecting Web services with
In Proceedings of the 13th International Joint security compliances: A managerial perspective.
Conference on Artificial Intelligence, Chambery, To appear in Proceedings of the Pacific Asia Con-
France (pp. 866-871). ference on Information Systems (PACIS).
Berinato, S. (2005, July). A few good metrics. Khan, K., & Han, J. (2006, April). Assessing
CIO-Asia Magazine. Online version http://www. security properties of software components: A
csoonline.com/read/070105/metrics.html software engineer’s perspective. In Proceedings
Besharati, B., Azram, S., & Kannan, P. K. (2005). of the Australian Software Engineering Confer-
A decision support system for product design ence, IEEE Computer Society.
selection: A generalized purchase modeling ap- Parameswaran, M., Stallaert, J., & Whinston, A.
proach. Decision Support Systems, 42, 333–350. B. (2001, August). A market-based allocation
doi:10.1016/j.dss.2005.01.002 mechanism for the DiffServ framework. Decision
Burrows, M., Abadi, M., & Needham, R. (1989, Support Systems, 31(3), 351–356. doi:10.1016/
December). A logic of authentication. ACM S0167-9236(00)00143-3
Operating Systems Review, 23(5), 1-13. A fuller Payne, S. (2002). A guide to security metrics.
version was published as DEC System Research SANS Institute.
Centre Report Number 39, Palo Alto, California,
February. Power, D. J. (1997). What is a DSS? The Online
Executive Jo971021.htmlurnal for Data-Intensive
Choi, H. R., Kim, H. S., Park, Y. J., & Whinston, Decision Support, 1(3). Online version http://
A. B. (2004, March). An agent for selecting www.taborcommunications.com/dsstar/97/1021/
optimal order set in EC marketplace. Decision
Support Systems, 36(4), 371–383. doi:10.1016/
S0167-9236(03)00027-7
1119
Power, D. J. (2004, February). Specifying an Security Criteria: A security criteria is a rule

expanded framework for classifying and describ- with a set of security properties that can be used to
ing decision support systems. Communications of assess a security function or security objective. A
the Association for Information Systems, 13(13), security criteria tests whether a security function
158–166. has desired security properties.
Security Function: A security function is the
Power, D. J. (2007). A brief history of decision
implementation of a security policy as well as a se-
support systems. DSSResources.COM. Retrieved
curity objective. It enforces the security policy and
December 6, 2007, from http://DSSResources.
provides required capabilities. Security functions
COM/history/dsshistory.html
are defined to withstand certain security threats,
Saaty, T. L. (1980). The analytical hierarchy vulnerabilities, and risks. A security function usu-
process. New York: McGraw-Hill. ally consists of one or more principals, resources,
security properties, and security operations.
Sprague, R. H., & Carlson, E. D. (1982). Building
Security Objective: A security objective is
effective decision support systems. Englewood
an abstract representation of a security goal. A
Cliffs, NJ: Prentice-Hall.
security objective defines a desired security state
Swanson, M., Bartol, N., Sabato, J., Hash, J., & of an entity or data of the system. It represents
Graffo, L. (2003, July). Security metrics guide for the main goal of a security policy.
information technology systems [Special Publica- Security Profiling: Security profiling is the
tion 800-55]. National Institute of Standard and security characterization of an entity, a service,
Technology (NIST), (p. 99). or a component in terms of security objectives as
well as security properties. It spells out the actual
Syrjanen, T. (2000). Lparse 1.0 user’s manual.
implemented security characteristics of an entity.
University of Helsinki.
Security Property: A security property is an
Turban, E. (1995). Decision support and expert implementation element used in a security func-
systems: Management support systems. Engle- tion. A set of security properties can form a security
wood Cliffs, NJ: Prentice Hall. function. A security property is an element at the
lowest level of the implementation.
Web Services: A Web service is a platform-
independent and self-contained software with
KEy tErMs AND DEFINItIONs defined functionality that can be available over
the Internet. It provides a standard way of integrat-
Security Class: A security class represents
ing mechanisms with enterprise applications over
a generic grouping of similar types of security
the net. A Web service can perform one or more
objectives that share a common focus while dif-
functionalities for the complex application system.
fering in coverage of security functions as well
as security properties.
This work was previously published in Encyclopedia of Decision Making and Decision Support Technologies, edited by Fred-
eric Adam and Patrick Humphreys, pp. 211-217, copyright 2008 by Information Science Reference (an imprint of IGI Global).
1120
1121
Chapter 4.16
ERP Systems Supporting Lean
Manufacturing in SMEs
Pritish Halgeri
Kansas State University, USA
Roger McHaney
Z. J. Pei
AbstrAct and embrace rapid change forced many SMEs to

rethink their production approaches, particularly in
Small and medium enterprises (SMEs), more regard to where they stand in relation to these two
than ever, are being forced to compete in a global methodologies. Over time, ERP vendors recognized
economy with increasingly complex challenges. the power and advantages of Lean manufacturing
This new economy has forced SMEs to become and developed ways to incorporate Lean-related
more responsive and agile in operational, tactical and features into their software. The main objective
strategic areas while requiring thoughtful integra- of this chapter is to explore how ERP and Lean
tion between business functions and manufacturing/ methodologies can coexist in SMEs. The chapter
production/ service operations. Enterprise Resource discusses misconceptions about the fit between ERP
Planning (ERP) and Lean manufacturing are two and Lean then summarizes differences and syner-
production control methodologies that have been gies between the two methodologies. The chapter
implemented in various ways. In early incarnations, emphasizes how linking ERP and Lean methods
ERP systems were considered a hindrance to Lean can lead to competitive advantage then explores
manufacturing efforts and were criticized for en- key Lean toolsets available in leading ERP systems
couraging large inventories and slower production. used by SMEs. Further focus is provided with ad-
The explosive growth of e-business methodolo- ditional insight on several leading ERP vendors
gies and the resulting pressure to become nimble offering Lean-enabled software modules. These
include Oracle, TTW WinMan and Pelion Systems.
DOI: 10.4018/978-1-60566-892-5.ch005
ERP Systems Supporting Lean Manufacturing in SMEs
INtrODUctION suggests this can be hard to overcome and for

some time, that appeared to be the case.
Small and medium enterprises (SMEs), more than In recent years, SMEs involved in the business-
ever, are being forced to compete in a global econ- to-business (B2B) market have worked hard to
omy with increasingly complex challenges. This develop delivery performance capability com-
new economy has forced SMEs to become more patible with larger corporate customers. In many
responsive and agile in operational, tactical and cases this means that the SME is required to
strategic areas while requiring thoughtful integra- interface with their clients’ ERP systems. Larger
tion between business functions and manufacturing/ enterprises rely on big ERP system vendors such
production/service operations. When faced with as SAP, ORACLE, and others (Rashid, Hossain &
similar pressures, larger firms migrated to expen- Patrick, 2002). The implementation cost of these
sive ERP systems. As early as 1999, researchers systems is high and installation complex making
(Gable & Steward, 1999) suggested SMEs would it difficult for SMEs to follow suit. In response,
follow suit and suggested reasons motivating this midrange and less complex systems have been
phenomenon. First, the larger enterprise market developed both by the large ERP vendors and by
for ERP systems was becoming saturated and ERP smaller software companies. In order to continue
vendors were hungry for new markets. Secondly, taking advantage of being, smaller nimble compa-
these larger firms were pushing ERP vendors to nies, and satisfy the needs of the larger corporate
create software to leverage inexpensive Internet partners, SMEs may need to use their ERP software
technologies that would promote closer integration in conjunction with other proven systems and
with their SME partners along the supply chain methodologies such as Lean planning and control
to obtain a variety of efficiency-based benefits. tools. SMEs may need to combine capabilities
Thirdly, SMEs made up a large portion of regional to continue using other concepts such as just-in-
economies and represented a high percent of overall time (JIT) and optimized production technology
manufacturing and service firms. And finally, ERP (Cheng & Podolsky, 1993; Deep, et al. 2008; Koh
packages designed for SMEs had become more & Simpson, 2005). This article specifically looks
sophisticated, cost efficient and upwardly scalable at progress made in integrating Lean production
for growth oriented firms. methodologies with ERP systems.
In spite of these obvious incentives many SMEs
were slow to adopt ERP technologies. According
to Aladwani (2001) two fundamental sources bAcKGrOUND
of resistance to innovations like an ERP exist:
perceived risk and habit. According to Aladwani competing Philosophies
(2001), perceived risk refers to one’s perception
of the risk associated with the decision to adopt Over the past couple decades, philosophies related
the innovation, i.e. the decision to accept an ERP to the most effective manner to run manufactur-
system and habit refers to current practices that ing operations have been debated and evolved
one is routinely doing. Koh & Simpson (2005) greatly (Nakashima, 2000). Added to the mix
suggest this is pronounced in SMEs due to has been increased competition and expectations
widespread informal culture and a disregard for for rapid production changes and retooling. This
formalizing business processes. Often in SMEs, has intensified the need for more efficient and
a worker wears many hats and as a result, opera- cost effective manufacturing and put pressure on
tions are conducted on the fly and without formal managers and production engineers to develop
procedures or documentation. Aladwani (2001) new and better solutions.
1122
The methodology debate has not been with- technologies and the resultant rise of e-business
out controversy and opposing philosophies have (Kent, 2002) have forced most SMEs to revisit
emerged as different means to address the same their production control methodologies and re-
basic problem (Spearman & Zazanis, 1992). evaluate where they stand in relation to ERP use
Nowhere is this situation more apparent than and Lean manufacturing implementation. As at-
where manufacturers are torn between two camps titudes have changed and information exchange
concerning production control (Piszczalski, 2000). increased, Lean practitioners have begun to
One embraces Enterprise Resource Planning adopt supportive ERP software to facilitate their
systems (ERP) where for the past two decades, advanced Lean manufacturing initiatives and al-
organizations have spent billions of dollars and low the rich compilation of organizational data
countless hours installing enterprise wide systems and production history take them to a new level
(Bradford, Mayfield & Toney, 2001). The other of operational excellence (Bragg, 2004).
camp is Lean manufacturing (Piszczalski, 2000),
pioneered in Japan by Toyota Motor Corporation Lean Enabled ErP Development
but now embraced in the U.S. by literally thousands
of firms, particularly SMEs (Bartholomew, 1999). The area of Lean enabled ERP software has be-
ERP and Lean have emerged from funda- come the subject of many recent and important,
mentally different approaches to production. As ongoing developments. However, very few re-
a result, misconceptions that Lean and ERP ap- searchers have examined this area to discuss the
proaches do not mix well have emerged (Steger- availability of Lean enabled ERP systems (Halgeri,
Jensen & Hvolby, 2008). In some cases, these et al, 2008). Additionally, not many researchers
misconceptions have gone as far to argue that describe the Lean toolset features offered by ERP
ERP systems are actually the antithesis of Lean vendors. This article will attempt to rectify that
manufacturing and are largely responsible for by discussing the differences between ERP and
a variety of inefficiencies (Nakashima, 2000). Lean methodologies and then examining available
Others say ERP is a waste of money and time in Lean toolsets provided by several ERP vendors.
no uncertain terms. For instance, in 1999, Bar- The article will go on to provide a list of ERP
tholomew reported opponents of ERP as saying: vendors offering support in Lean manufacturing
We are staying away from ERP because it doesn’t then provide additional details on several selected
work. To do both ERP and Lean jeopardizes the vendors together with the expected benefits of
success rate of either (Bartholomew, 1999). their software solutions.
This opinion is one side of many arguments
both pro and con. Others have said: only by us- Enterprise resource Planning (ErP)
ing computer systems (ERP) manufacturers can
possibly get their arms around the Herculean ERP is defined as a method for the effective plan-
task of recognizing the multitude of constraints ning and controlling of all the resources needed
and issues that inevitably impact operations and to take, make, ship and account for customer
planning (Piszczalski, 2000). To them, it is obvi- orders in a manufacturing, distribution or service
ous that ERP systems have an important purpose company (Miller, 2002). ERP software attempts
in gathering enterprise data that can and should to integrate all departments and application
be used in conjunction with Lean methodologies modules into one computer system to serve each
(Bradford, Mayfield & Toney, 2001). department’s needs from a central repository.
Added to competitive pressures, the explosive ERP effectively eliminates standalone computer
growth of information and telecommunications systems for each functional business silo or soft-
1123
ware module (e.g. manufacturing/production, software without changing the business model to
warehousing, accounting, marketing, logistics, match encoded best practices can result in fewer
human resources, finance, etc.) and replaces them benefits, and may even decrease efficiency and
with an integrated program composed of various productivity. Additionally, employees must be
modules representative of required components trained to use both new software and new busi-
of an organization’s business. The primary differ- ness processes in the conduct of their work tasks.
ence is that these components are now linked with Another disadvantage relates to cost factors. ERP
a common database therefore giving all entities systems can be expensive to install and run. Hid-
within the organization the ability to share and den costs also exist. For example, training, data
view desired information while eliminating data conversion, testing, and updates all add to the price
entry and update redundancies, and effectively tag. Of course, these costs are generally part of
enhancing communication with easy access to any organizational information system and should
work flows. (Koch, 2008). Additionally, ERP be viewed within that perspective.
solidly integrates all necessary business functions,
such as production planning, accounting, purchas- Lean Manufacturing
ing, inventory control, marketing, sales, finance,
and human resources, into a single system with Lean manufacturing is a term given to a family
a shared database (Li, Liao & Lei, 2006) which of related methodologies that seek to streamline
enables a variety of automatic data collection not production processes. Sometimes Lean is re-
previously possible. ferred to as Just-in-Time (JIT) manufacturing,
As a direct result of ERP systems’ abilities kanban system, Toyota Production System, or
to collect data and improve the ease of informa- flow manufacturing. A primary feature of Lean
tion sharing, several other advantages generally manufacturing how it uses demand to pull items
result from its use. For example, ERP systems into inventory and through the manufacturing
consistently improve SMEs’ productivity, as the process. Lean attempts to sequence these items
systems eliminate the need for multiple entries of on flow lines to maximize resource utilization
identical data, reduce the possibility of errors and (Kent, 2002). A key goal of Lean manufacturing
inconsistencies, and reduce time spent on needless is the continuous reduction and even eventual
phone calls and inquiries to other departments. elimination of all waste in the production process
Other advantages provided to SMEs using ERP (Mekong Capital, 2004; Strategosinc, 2008).
systems include improved tracking and forecast- In general, Lean manufacturing seeks to mini-
ing abilities, better customer service, ability to mize all the resources (including time) used in
standardize manufacturing processes more easily, various enterprise activities. A key aspect of Lean
and in some instances, cost savings after initial is to identify and reduce or eliminate non-value-
start-up costs are recouped (Jacobs, 2007). added activities in design, production, supply chain
While ERP systems possess numerous advan- management, and streamline customer interac-
tages, key disadvantages also exist. For instance, tion. Lean principles and practices aim to reduce
ERP systems require extensive installation and cost through the relentless removal of waste and
configuration processes that may take from three through the simplification of all manufacturing
months to a year or more depending on the size and and support processes (Miller, 2002) consistent
complexity of the organization. For some SMEs with continuous quality improvement techniques.
this may be less of an issue since less customiza-
tion may be required. ERP systems often demand
reengineering business processes. Installing the
1124
Key Principles of Lean Manufacturing (8) Continuous Flow: Lean manufacturing

seeks to implement continuous production
Several key principles lay behind the develop- flows free of any interruptions, bottlenecks,
ment and use of Lean manufacturing techniques. detours, backflows, unwanted changes or
Among these are: waiting (Mekong Capital, 2004)
(1) Elimination of Waste: Lean thinking offers Lean Manufacturing Shortcomings

a definition of value and therefore allows an
organization to determine what activities and Lean manufacturing is not without shortcom-
resources are needed to create and maintain ings. Key features in Lean can only be used to
this value (Poppendieck, 2002). their fullest potential in situations where a stable
(2) Pull Production: In its most basic form, master schedule exists. This means that current
downstream activities inform upstream processing must match with capacity. Unexpected
activities when replenishment is required. changes or an influx of unexpected orders can be
(3) Value Stream Identification: To deliver difficult to accommodate. Further, changes in lead
products as specified by customers, an as- times can result in problematic changes. Many
sessment of required actions to move through manufacturers expecting these situations may
purchasing, production, distribution and not use Lean methodologies in the first place.
Another potential shortcoming is the inability to
other parts of the value stream are necessary
share and communication electronic transaction
(Burton & Boeder, 2003; Paez, et. al. 2004)
data with business operations and other parts
(4) Standardization of Processes: Lean manu-
of the enterprise. In Lean, information flow is
facturing seeks to develop standardized work
minimized (Nagendra & Das,1999).
processes to eliminate variation in the way
individual workers may perform their tasks.
Lean may use a combination of training
ErP AND LEAN: cAN
techniques and process design to ensure this
tHEy cOEXIst?
is possible (Mekong Capital, 2004).
(5) Quality at the Center: At the center of Lean
ErP and Lean compared
manufacturing is a commitment to quality
improvement. Lean encourages the elimina-
Although both ERP and Lean manufacturing seek
tion of defects at the source and strives to
to provide a similar outcome to an organization,
ensure quality inspection is completed by the conflict between the methodologies can be
workers during in-line production process attributed to implementation and philosophical
(Poppendieck, 2002). difference in general (Bartholomew, 2003). One
(6) People Add Value: Lean manufacturing of these differences involves the way goods and
values the human component and seeks services are viewed in the two methodologies. For
to add value to processes through human instance, Lean promotes a “pull” environment as
interaction (Arc Strategies, 2007). the operative principle whereby goods or services
(7) Continuous Improvement: Lean manufac- are not purchased or produced until demand exists.
turing is never content with the status quo. Meanwhile in traditional ERP systems, the primary
It seeks to improve and enhance efficiency focus is on the “push” principle where forecasts
and production practices. So Improvement are developed from historic data and orders, and
is never truly complete (Turbide, 2005). then goods are produced to meet expected demand
(Nakashima, 2000).
1125
While not sounding too different, the basic use the actual end customer demand to drive the
philosophy between these two approaches is manufacturing process as much as possible. Pull
significant. First, in a push system, a production systems attempt to match the rate of production
job begins on a start date computed by subtract- for each product to the rate of customer consump-
ing an established lead time from the date the tion (Steger-Jensen & Hvolby, 2008) as it occurs.
goods are required, usually for shipping or for This has the net effect of reducing inventories and
assembly (Spearman & Zazanis, 1992). A push holding down costs throughout the supply chain.
system schedules the release of work based on A pull system is characterized by the practice of
demand. The key factor here is access to historic downstream work centers pulling stock from pre-
data that aids in the development of forecasts for vious operations (Spearman & Zazanis, 1992). In
items to be produced. Release time has to be fixed other words, nothing is produced by the upstream
and cannot be modified easily for unexpected supplier until the downstream customer signals a
changes in the manufacturing system or product need (Kent, 2002). Work is coordinated by using
being manufactured. Figure 1 represents a general an information flow up the supply chain, having
push production system. originated with the end consumer of the product
In a push system, information from the master (Spearman & Zazanis, 1992). Figure 2 shows a
production schedule (MPS) flows downstream typical pull system.
toward the finished goods inventory and is com- Information flows from the finished goods
puter generated (Kent, 2002) based on history and inventory (the customer) upstream towards the
current orders. The MPS becomes the plan used raw material inventory and in many SMEs can be
by a company for production. The MPS consid- conveyed visually by the use of a kanban (Kent,
ers all work currently in-house and develops an 2002). Many systems are now completely run
aggregate plan based on forecasts for individual by automated computer-driven systems. Other
parts, actual orders on hand, expectations for or- differences between traditional ERP systems and
ders, and available capacity (Lee & Adams, 1986). Lean manufacturing are summarized in Table 1.
A pull system takes the opposite approach and
at its center is inherent flexibility. Pull systems
Figure 1. Typical push system (Halgeri, 2008)
Figure 2. Typical pull system (Halgeri, 2008)
1126
Table 1. Differences between ERP systems and lean manufacturing
Aspect ERP LEAN Reference

Emphasis Planning Continual Improvement of Produc- Bradford, Mayfield & Toney,
tion Process 2001
Production Plans Combination of Actual Sales and Based solely on actual orders from Nakashima, 2000
Forecasted Sales Projected from internal, downstream processes or
Historic Data external customers
Transaction Creates non value-added transactions Seeks to eliminate all waste including Bartholomew, 1999
because every event and activity in movement, unnecessary transactions,
entire business is tracked (although and materials. Seeks to speed and
most are done automatically) smooth production
Traditional Approach Top-down Bottom-up (Nakashima, 2000)
Time Horizons As short as a few weeks but as long Based on daily production capacity (Nakashima, 2000)
as a year or more. Average around and actual orders received
12-week mark.
Basic Focus Forward-looking planning, commu- Cost reduction and process improve- Bartholomew, 2003
nication, and scheduling tool ment methodology
Platform Computer dependent Shop Floor oriented (often with (Piszczalski, 2000)
computerization
Production Concepts Loaded machine work centers Balanced production lines with (Nakashima, 2000)
synchronized Takt and cycle times
Information Philosophy More is better. For example, more Less is best. For example less vari- (Piszczalski, 2000)
information, more flexibility, more ability, less material, less movement,
functions and more features are less floor space are desirable
desirable
Product Movement Product moves in batches with speci- Each operation is completed on a (Nakashima, 2000)
fied operations being performed on single unit with specific unit moved
the complete batch before moving to next operation in a continuous flow
on to the next operation
Advantages of ErP over Lean their immediate point of interaction. This can
inhibit the design of better products.
ERP can provide advantages over Lean. For in- Lean can create the low cost, efficient opera-
stance, researchers have claimed ERP, in general, tions needed for survival. However, that is not
is more applicable to more manufacturing firms the only goal of most companies where long term
than Lean (Spearman, Woodruff & Hopp 1990) survival is also an important consideration. Thus,
and that Lean should be used in high volume in many SMEs, plant software adoption and the
production environments with relatively few part use of historic data is expected to rise. And of
types (Bonvik, Couch & Gershwin, 1997) but not course, the plant operations that can better respond,
other environments. innovate, and share information will find greater
In addition to a number of functionalities, opportunities for cost savings, future goods, and
software can extend visibility into the plant. customers (Greene, 2004).
Kanban management systems can communicate Software can bring capabilities to the plant
current demand to suppliers, but generally fails environment that are difficult or even impossible
to illuminate quality or other shop floor issues. to achieve by any other mechanism. Of course, not
Lean manufacturing’s line-of-sight management every production operation will require all of the
can also fall short since it typically also does not capabilities available in software systems, but for
help internal departments serve customers beyond those needing to achieve specific goals, certain
1127
features can be crucial. It is possible SMEs not Potential for Linking ErP and Lean
specifically interested in implementing Lean may
still acquire software tools typically found in ERP Some academics and industry analysts have im-
or other enterprise systems for use in their envi- plied that ERP is old school and Lean manufactur-
ronments. Several examples of these features are: ing has replaced it. Others, of course, disagree,
saying those claims are akin to saying the auto-
• Track and Trace: Many industries re- mobile chassis is obsolete because a new engine
quire full product histories that track all has been invented (Miller, 2002). It is important
materials back to their sources for qual- to make a distinction between an ERP and Lean
ity and other reasons. Required or not, a manufacturing. ERP can be thought of as a large
track and trace feature makes great busi- scale software system for linking all parts of the
ness sense. Not only can warranty costs be firm and encouraging best practice business pro-
reduced but brand image can be protected. cesses. Lean manufacturing is one of these best
Additionally, recall costs can be better con- practices mostly focused in the manufacturing
trolled and supply problems can be identi- area. At least, that’s one productive way of view-
fied and suppliers rated and ranked. Most ing the synergy that may emerge from a marriage
ERP software systems include this feature. of the two methodologies.
• Performance Dashboard: Managers need As an example of how the broader set of
access to performance data and a perfor- organizational data can benefit Lean manufac-
mance dashboard provides access to in- turing, consider a shortcoming of many Lean
formation as it is acquired. Timely and initiatives. Often, these initiatives make rapid
accurate views of data are essential to im- progress then the rate of improvement plateaus.
proving business performance. The difficulty of managing pull operations in an
• Work Instruction Tools: Lean manufac- environment where demands and product mix
turing relies on work standardization. In widely fluctuate is partly responsible for this
many new software systems developed to plateau (Bragg, 2004). Another difficulty might
support Lean and in many ERP systems, be related to incomplete data. Imagine a 30%
best practices down to the work instruction reduction in process cycle time resulting from
level are maintained. An easy way to en- a Lean kaizen event that does not get formally
sure quality and efficiency is through the communicated to marketing, procurement,
maintenance of up-to-date work instruc- material planning, customer order management
tions. This functionality often comes from or the capacity planning functions (Metzger,
specialized vendors and ERP software 2008). What is perfectly obvious in manufac-
providers. turing may not be communicated up the supply
• Resource Allocation Tools: Production chain to other business operations. Visual pull
systems with responsibility to produce systems developed for a local process must be
multiple products often face resource al- communicated and linked through ERP to sus-
location problems. This may include work tain the Lean program results across the SME
fixtures, capital equipment, or even skilled (Metzger, 2008). This is especially important
workers. Finite capacity scheduling sys- when the SME maintains separate production
tems and other resource allocation tools and business facilities.
have been developed to aid with these Having access to additional business data may
situations. lead to understanding where bottlenecks may
still occur, or what operations may be further
1128
optimized. Without historic data combined with by 50%, while sales volumes increased by 35%
analytic tools, this can become increasingly dif- (Nakashima, 2000). These improvements are by
ficult or even impossible. The data associated no means insignificant.
with an ERP can help identify these constraints
to optimize demand flow manufacturing (Bragg, ErP Vendors Offering Lean tools
2004).
Other challenges to Lean include trying to Many ERP vendors have introduced “enablers”
modernize manual kanban methods that use that incorporate Lean manufacturing best prac-
printed cards, which work well within SMEs tices into their systems. These enablers include a
with contained environments or “line of sight” variety of new modules, database items, toolsets,
manufacturing facilities. Major difficulties can and business process modifications that create new
result when these systems are extended across functionality in existing software (Nakashima,
the plant and to suppliers, since breakdowns in 2000). The following sections describe these
manual kanbans get lost at the local level (Bragg, features in more detail (Halgeri, 2008):
2004). Typical approaches for Lean used by most
SMEs today do not provide an optimal return Toolset # 1: Just-in-Time
on investment for companies (Steger-Jensen &
Hvolby, 2008) and financial data regarding each A key difference between ERP implementation
part of the operation may not be collected. and Lean manufacturing has been the approach
Lean enabled modules offered by ERP suppli- to production. Since ERP has traditionally used a
ers and specialist suppliers may help overcome push production system, adopting a philosophy of
many of these challenges (Bragg, 2004). In fact, pull production has been a challenge. Many ERP
ERP vendors are starting to offer new solutions vendors have begun to add a JIT toolset to their
aimed to bridge the gap between the shop floor best practices procedures as a step in this direc-
and ERP (Nakashima, 2000; Bradford, Mayfield, tion. A primary functionality of this tool extends
& Toney, 2001). These solutions may be offered in the pull concept through the entire supply chain
the form of Lean modules or add-on components from suppliers to customers. Of course, incorpo-
(Bradford, Mayfield & Toney, 2001). rating just-in-time features encourages flexible,
Benefits gained by implementing Lean-en- small lot deliveries of parts and materials without
abled ERP modules can be illustrated by the fol- the need for traditional purchase orders and this
lowing example provided by Nakashima (2000). can be difficult to implement in an ERP system
Cerberus, a New Jersey-based manufacturer of (Nakashima, 2000).
commercial fire detection systems, and division One vendor, Oracle, and its Flow Manufactur-
of Siemens, implemented Lean techniques and ing software includes multiple component replen-
American Software’s Flow Manufacturing ap- ishment routines that support multiple component
plication to achieve significant increases in both demand patterns. This flexible approach to JIT
flexibility and productivity. Before implementing procurement helps eliminate stock-outs. The
Flow Manufacturing application, a new model software automates material replenishment pro-
would be produced every 15-30 days. After imple- cesses through the creation of a self-regulating pull
menting Flow Manufacturing, multiple models production system. Oracle also offers Feeder Line
come off the production line every 20 seconds Synchronization processes that can create specific
(Nakashima, 2000). Productivity increased by JIT schedules for complex, dependent demand to
as much as 15-20%, floor space was reduced by ensure parts are delivered when needed. This is
25-30%, finished goods inventory was reduced ideal for products that must be built specifically
1129
for customer orders, or are highly customized information management (from order taking to
or variable. Using additional features, required delivery), as well as material transformation (from
components and parts can be ordered to ensure incoming raw materials to finished goods) are all
delivery at the perfect time to become available key to conducting a value stream analysis. VSA
during production. Stable, predictable demand demonstrates process steps for elimination, modi-
can use kanban management processes but more fication or improvement both with and without
complex demand can access both ERP data and investment. A Value Stream Mapping (VSM) tool
use JIT concepts for delivery (Oracle, 2006). can be implemented to further improve this analy-
sis by graphically illustrating current and desired
Toolset # 2: Convertors to Transform processes in the value stream. This makes it easier
Multi-Level Bills-of-Material into to understand and implement recommendations
Flat Bills with Event Sequencers derived from earlier or ongoing Value Stream
Analysis activities (McManus & Richard, 2002).
Many ERP vendors have begun to offer function- Several vendors offer both VSA and VSM
ality intended to eliminate traditional MRP-based support in their ERP software systems. First,
bills-of-material that simulate the manufacturing Pelion’s ERP system is noted for its strengths
process and establish start and due dates for each in VSM. Their EASYVSM workbench software
department. These time-phased and shop floor adapts to specific production environments to re-
control routings are converted into flat bills without duce manufacturing cycle times and to streamline
sub-assemblies or parent assembly product rout- work processes. Additionally, non-value added
ing. These converters transform multilevel bills- work can be identified and eliminated. This helps
of-material into flat bills with event sequencers, increase customer responsiveness and satisfaction
emulating processes used in Lean manufacturing (Pelion Systems, 2008). The following features
(Nakashima, 2000). of EASYVSM have been documented as mecha-
nisms to improve product and material flows
Toolset # 3: Analysis and Mapping Tools (QSG Team, 2006):
Lean manufacturing initiatives often begin with

• Current and future state maps
an analysis phase used to identify potential
• Collaborative, enterprise-wide value
and existing production problems. The goal is
stream maps
to determine “quick wins” to immediately and
• VSM presentation material
dramatically improve production performance
• Visual communication boards in multiple
(Bragg, 2004). These quick wins may come about
because of production practices or when Lean locations
principles are used to examine other business • Cycle time organization features
practices (McManus & Richard, 2002). Several • Features to manage changeover times, pro-
methods are used to accomplish this including duction volumes, information flows, staff-
Value Stream Analysis (VSA). ing requirements and inventory strategies
VSA is used to ensure the product specified by • Calculation tools for value-added and non-
the customer is delivered in a cost effective way value added times
and value is added throughout the supply chain • Visualization for high leverage projects,
(Paez, Dewees, Genaidy, Tuncel, Karwowski, kaizen events and operating programs
& Zurada, 2004). Womack & Jones (1996) sug-
gest several business methodologies including: Oracle’s Flow Manufacturing software also
problem solving (from design to product launch), offers VSA and VSM support. Flow Manufac-
1130
turing offers Graphical Line Designer to aid in changes can impact the work process as much as
visualization of current value streams and to the bill-of materials (Nakashima, B. 2000). Many
create maps which can be redesigned into bal- ERP systems have begun to implement this feature.
anced line operations and in ways that eliminate
waste and redundancy. Graphical Line Designer Toolset # 6: Kanban Control
defines standard processes and associates them and Management
with configurable models, product families, and
networks or processes. Each individual process Simple flow manufacturing, particularly in SMEs,
is linked to a set of primary and feeder processes often will not require enterprise software. Many
and can be identified as rework where relevant. successful manufacturers use kanban tags instead
Overall, users are able to focus on events important of work orders, however, this works best only in
to quality improvement initiatives (Oracle, 2006). situations where production is steady and product
demand stable with few changes (Turbide, 2005).
Toolset # 4: Demand Smoothing As production becomes more complex with greater
variation, changes, and customers, specialty
The focus of demand smoothing tools are to ac- kanban will be required. Special situations such
cumulate forecasts and customer demands and then as promotional peaks, seasonal demand, long
provide graphical analyses of requirements for lead-time supplies, and highly customized orders
daily production. This functionality is a natural fit can lead to difficulties in the kanban environment
for ERP systems since forecasting is often based on (Bragg, 2004). As demand changes, manual kan-
historical data. Most demand smoothing tools also bans can become more difficult to manage thus
consider important resources needed to complete changes harder to track (Garwood, 2002).
the recommended production (Nakashima, 2000). According to Bragg (2004) another difficulty
An ERP package that offers demand smooth- associated with manual kanban relates to loss. He
ing is Manugistics. It specializes in support for reports that up to 1% of manual kanbans gets lost
dynamic pricing techniques. Additionally, it can every day (Bragg, 2004). Manual kanbans are also
be integrated with production planning software hard to track and often labor intensive (Bragg,
modules. Manugistics smoothes demand through 2004). Simplicity can come at a cost.
dynamic pricing. So if demand exceeds produc- ERP software vendors have created support
tion capabilities, pricing is raised and vice versa. modules that address many kanban control and
This helps manufacturers proactively stabilize management problems. A promising solution has
demand so production rates more closely mirror been to offer electronic kanbans that continue to
sales (Bragg, 2004). Other ERP vendors offering allow plant floor operations to capitalize on the
demand smoothing include Oracle, SAP, and IFS. advantages of kanban but at the same time provide
access to a greater range of organizational data
Toolset # 5: Engineering and intelligence to optimize the use of resources.
Change Orders This enables smarter scheduling together with the
benefits of kanban. A much larger segment of the
Engineering change orders can be a huge dis- manufacturing community can benefit (Turbide,
ruption to workflow. Workflow or other similar 2005). Additionally, ERP systems can provide
technology to communicate engineering changes additional analysis tools which can help recalcu-
to the production line can be implemented using an late the size and number of kanban bins required.
ERP system to more easily and more immediately This can be done on-the-fly and used to alert key
signal the factory floor. This can be important since personnel of expected changes. Manual kanbans
1131
can then be changed. Of course, if electronic In a configure-to-order, products are assembled

kanbans are being used, then the changes can be from modules based on pre-constructed compo-
done automatically (Nakashima, 2000). nents that are then finished and configured to meet
Pelion’s ERP software offers a variety of customer demand (Bradford, Mayfield & Toney,
functions to extract updated requirements from 2001). A line design tool helps to synchronize
the organizational data repositories and uses this related work activities and ensure that raw mate-
information to send out new kanban levels to the rial consumption is considered and planned in a
plant floor. This can be done on a continual or way to reduce or eliminate queue time, inventory,
at least regular basis to avoid production over- and work-in-process. This helps support continu-
or-under estimates. These tools allow for proper ous improvement efforts by ensuring the facility
sizing and timing of material deliveries together has a continually balanced manufacturing line
with electronic release signals. This ensures ven- (Bradford, Mayfield & Toney, 2001).
dors provide timely delivery of needed supplies Oracle offers a tool called Sequencing Rules
and parts (Garwood, 2002). to perform these functions. Oracle uses ILOG’s
Oracle also provides kanban functionality in its optimization engine in conjunction with enterprise
Flow Manufacturing software. Oracle’s approach data to develop optimal production sequences.
has been to support all kanban transactions with ILOG is a third-party software vendor specializ-
its Mobile Supply Chain Applications (MSCA) ing in resource optimization, resource allocation
module. Kanban replenishment becomes more and network management among other areas.
sophisticated with access to enterprise data. Oracle’s Sequencing Rules toolset offers a va-
Therefore kanban can be based upon anticipated riety of features relating to Lean (Bragg, 2004).
volumes based on historical data and the optimal Among these are:
number of kanbans and their sizes can be calculated
based on anticipated and actual demand. This also • Groupings: Allows products with com-
protects against errors that may be attributable to mon attributes to be placed together
varying demands (Oracle, 2006). • Spacing between Products: Ensures
hard-to-manufacture variants not bunched
Toolset # 7: Line Design / Sequencing in production sequence
• Required / Disallowed Transitions:
In multiple product production lines, minimizing Transitions are monitored and tracked so
changeovers can become a difficult and tricky changeover-related costs can be reduced.
problem. In production lines with the capability to
simultaneously produce multiple parts, it becomes In another ERP package from SYSPRO (ARC
necessary to not only minimize changeovers, but Strategies, 2007), a “rules based” Product Con-
also sequence them in a logical order that maxi- figurator software module has been developed to
mizes overall production (Bragg, 2004). specifically address the needs of assemble-to-order
Some ERP vendors now offer Line Design or engineer-to-order SMEs. The main premise is
and Sequencing toolsets adapted from Lean that products comprised of varying combinations
manufacturing. These software tools are designed of predefined components, subassemblies, and
so flow logic can be used to maximize specific operations can be configured from within the
customer-ordered variations of offered product Sales Order, Quotation/Estimating and/or Work-
mixes and the key components required to meet In-Process modules. This gives a salesperson or
the demand. This is particularly true when prod- estimator the ability to immediately configure a
ucts are configure-to-order (Nakashima, 2000). product based on a client’s answer to questions.
1132
The Configurator software can be used to select generally achieved by using visual Kanban cards
a number of pre-finished products, to create a sales that can be tags, labels, containers, or electronic
kit, or a standard bill of material to request the signals (Turbide, 2005).
manufacture of a custom product (CTS, 2008). One of the available ERP systems offering
Backflush is the deduction from inventory support in Orderless Manufacturing is TTW’s
records of the component parts used in an as- WinMan software. It uses an empty container as
sembly or subassembly by exploding the bill of a signal to trigger internal replenishment. These
materials by the production count of assemblies empty containers further indicate completion of
produced (Cox and Blackstone, 2008). In other a finished product and backflushing of the com-
words, inventory can be backflushed to remove ponent material. In order to help complete the
raw material from inventory, instead adding to transaction, TTW has designed a set of internal
finished goods inventory or to inventories of Kanban cards which can be bar code scanned.
stocked sub assemblies. In Lean manufacturing, (WinMan, 2003). Oracle Flow Manufacturing
backflushing should be a routine transaction for records completions of assemblies without having
material issues. Activity reporting and inventory to create work orders (Oracle, 2006).
level updates can be replaced by the performance
of all inventory transactions upon completion of a ErP Vendors Offering Lean support
single unit (Nakashima, 2000). In some plants, the
replenishment signal for a bin is generated from a As is evident from the list of Lean-enabled features
backflush, since this saves the waste of creating now being supported in ERP systems, a number
Kanban cards and scanning them. of vendors have begun to enter the marketplace.
Oracle Flow Manufacturing backflushes all the Table 2 provides a list of some of these vendors
components and performs resource and overhead and their primary Lean-enabled software mod-
transactions upon recording the assembly comple- ules. Three of these vendors are then explored
tion. Oracle Flow Manufacturing allows scrap- in more detail.
ping assemblies and returning from scrap at any
operation using either scheduled or unscheduled Vendor #1: Oracle
flow schedules. A scrap transaction will cause all
the components through the scrap operation to be Oracle is considered a leader in Lean-manufac-
backflushed (Oracle, 2006). turing implementation within its ERP system.
According its research, Oracle had over 100
Toolset # 8: Orderless companies using its Flow Manufacturing module,
Flow Manufacturing and as many as four times that amount used its
Kanban software (Bragg, 2004). Oracle Flow
Traditional batch production is controlled through Manufacturing has been implemented as part
work orders which, by Lean standards, include too of the Oracle E-Business Suite (Oracle, 2006).
much waste (Turbide, 2005). Kanban-based flow Oracle considers Flow Manufacturing as crucial
manufacturing is conducted without work orders to its e-commerce strategy. Because of this, Flow
– and therefore without the waste associated with Manufacturing has been promoted as effective in
work orders. Flow manufacturing is characterized the reduction of product cycle times, inventories,
by production lines and/or cells in which work and process complexity. In addition, Oracle claims
is pulled or moves piece-by-piece through the the software will simplify production, and help
process and not in batches (Turbide, 2005). This meet production demand at affordable prices
pull based work flow in the production line is (Kent, 2002). Of course, none of these features
1133
Table 2. Lean enabled ERP systems
Lean-Enabled Software Vendor Website

Alliance/MFG Exact Software www.alliancemfg.com
American Software Enter- American Software www.amsoftware.com
prise Version 3
Demand Point Pelion Systems www.pelionsystems.com
E2 Shoptech www.shoptech.com
e-Intelliprise American Software, Inc www.amsoftware.com/marketing/intelliprise-home.asp
Fourth Shift Softbrands Manufacturing www.fourthshift.com
Solutions
Global Shop Global Shop Solutions www.globalshopsolutions.com/product.htm
IFS Applications IFS www.ifsworld.com
Infor ERP VISUAL Infor Global Solutions, Inc. www.infor.com
Made2Manage Consona www.made2manage.com
MFG/PRO QAD www.qad.com
Manugistics JAD Software Group, Inc. www.jda.com
MISys MISys www.misysinc.com/mi2kover.htm
Oracle E –Business Suite Oracle www.oracle.com
PeopleSoft Enterprise Oracle www.oracle.com/applications/peoplesoft-enterprise.html
Seradex ERP Solutions Seradex www.seradex.com/ERP/Lean_Manufacturing_ERP.php
(Seradex, 2007)
Sage ERP X3 Adonix www.adonix.com
SYSPRO Enterprise SYSPRO www.syspro.com
Ultriva Ultriva (ebots) www.ultriva.com
Vista Epicor www.epicor.com
WinMan TTW www.winmanusa.com
xApp SAP www.sap.com/solutions
would be possible without the synergy offered systems (Lee & Adam, 1986; WinMan, 2006).
by coupling Lean with the wide array of ERP Table 3 provides a list of several Lean-enabled
enterprise data. tools offered by Oracle (Oracle, 2006):
Additionally, Oracle’s software allows users to
create simulations of expected Lean environment Vendor #2: TTW
changes. Several simulation tools allow experi-
mentation with line balancing, optimized product TTW’s Windows ERP product, WinMan is par-
flows, and Heijunka sequencing (Wheatley, 2007). ticularly suited for use by SMEs. This product
Oracle’s software not only supports kanban from features an integrated manufacturing management
assembly to the supplier base, it also includes system designed specifically for the needs of
tools to provide users with the capability to view small to midsized enterprises. Since it manages
historic demand and optimize the size and number both manufacturing and business operations,
of system kanban cards (Wheatley, 2007). Various including bill of materials, purchasing, material
researchers and practitioners have reported a great requirements planning (MRP), inventory planning
deal of success with Oracle’s Lean enabled ERP and control, and master production scheduling
1134
Table 3. Lean-enabled tools from Oracle
Feature Description
Value Stream Mapping Identifies opportunities for improvement
Value Stream Analysis Visualizes opportunities for improvement
Line Design and Balancing Supports mixed model production of standard or configured products
Just-in-Time Procurement Pull-based, kanban replenishment chain supported to improve inventory turns. Also, supports
synchronized component replenishment for configured or build-to-order components.
Electronic Work Methods Lean manufacturing execution workstation supported for operators thus enabling move
toward paperless shop floor.
Backflushing Scrap transaction generated automatically causing all components to be backflushed
Kanban Supports kanban transactions with Mobile Supply Chain Applications (MSCA) module
Orderless Flow Manufacturing Allows recording of completions of assemblies without having to create work orders
Sequencing and Scheduling Capability Produces directly to customer order
(MPS), it can help control all aspects of an SME of related tools. Table 4 illustrates (CTS, 2008;
(Global Shop Solution, 2008). WinMan, 2006).
TTW reports several success stories related
to WinMan implementations including Lantech Vendor #3: Pelion Systems
being about to grow its share of the stretch-wrap
machinery market from 35 percent to 50 percent. Unlike other ERP vendors, Pelion Systems’ suite
Capricorn Cars Parts increased its inventory of products is developed as third party software to
turnover by 188% and increased its parts port- augment existing ERP applications. The primary
folio from 1,800 to 15,000 while experiencing objective of Pelion’s software is to reduce exces-
a marginal increase in overhead costs (WinMan, sive lead time and lot size obstacles through the
Athena Controls 2008). use of Lean manufacturing practices (Garwood,
Since many SMEs utilize the benefits offered 2002). Pelion calls its primary software Demand
with Lean practices, WinMan provides a number Flow Technology. The approach is to combine a
Table 4. Lean-enabled tools from TTW’s WinMan
Feature Description
Support for Just-in-Time External Kanban generates bar coded kanban cards for pre-selected suppliers and allows
manufacturers to pull inventory on demand from the shop floor
Demand Driven Manufacturing Eliminates non-value added activities and reduces inventory by using pull manufactur-
ing techniques
Push/Pull Flexibility Ability to process back to back orders in both purchasing and manufacturing and simulates
the pull concept while still utilizing traditional push purchase orders and manufacture
orders
Line Design and Sequencing. Product Configuration workbench allows selection of various options on the fly dur-
ing the sales order process. Guided by pre-selected logic. May offer inclusion and/or
exclusion rules.
Orderless Flow Manufacturing Uses an empty container as a signal to trigger internal replenishment. Empty containers
may signal completion of finished product and trigger backflushing of the component
material
Backflushing Backflushing is deduction from inventory records of component parts used in an as-
sembly or subassembly
1135
kanban execution system with value stream map- FUtUrE trENDs

ping capability. Full integration occurs within the
line design and balancing tools inside Pelion’s The near saturation of ERP sales in large organi-
Lean module, Collaborative Flow Manufacturing zations has encouraged software vendors to seek
(CFM). This further offers support for kanban size additional venues for development and growth.
determination in flow lines. Pelion provides func- SMEs provide a natural fit for a new focus of ERP
tionality that gives users the capability to work on software development. A problem in marketing
four time horizons simultaneously (Bragg, 2004). software within this arena has been the widespread
For instance a production planner can use an annual focus on Lean manufacturing used by many SMEs.
time horizon to develop a plant layout, a quarterly This problem becomes complicated since most
horizon to implement new product introductions ERP software is based on push methodologies and
together with expected kanban sizes and takt time, Lean practice revolves around pull methodologies.
monthly horizons for work schedules and other Software vendors have addressed this through a
plans, and finally daily horizons to monitor and variety of new software modules and ERP add-ons.
manage work output (Bragg, 2004). This in turn has encouraged additional manu-
Bragg (2004) reports a number of success sto- facturers to adopt Lean practices and experience
ries associated with Pelion software. For instance, the synergy offered by using Lean-enabled ERP
Husqvarna reduced its inventory by more than a software. However, manufacturers face a variety
million dollars and was able to improve order fulfill- of issues when making the transition to Lean
ment within a predetermined time frame from less (Michel, 2002).
than sixty percent to more than ninety-five percent. Although the primary question, can ERP soft-
Brooks Automation reported a twenty million dol- ware be used to support Lean has been answered
lar inventory reduction. Nissan Forklift reported a by a variety of organizations with their software
twenty percent reduction in final assembly direct solutions (See table), many Lean purists believe
labor, a forty-seven percent reduction in inventory only visual signals and shop floor implementation
and nearly five and half million in annual savings. provide a true implementation of this philosophy
Table 5 provides a look at the Lean-enabled toolsets and the added functionality of ERP flies in the
provided by Pelion Systems. face of basic Lean premises. Others believe ERP
systems provide additional transactional founda-
tions and historical data collection and analyses
Table 5. Lean-enabled toolsets from Pelion Systems
Feature Description
Value Stream EasyVSM workbench determines key breakthrough improvement opportunities, helps distinguish between value-
Analysis added and non value-added activities.
Value Stream Map- Delivers visual roadmap of Kaizen and other improvements, increases visibility and aids in communication
ping
Lean Engineering: Defines flow as a means of driving Lean process layouts and aids in continuous improvement effects
Lean Material Flow Aids in material flow design, strategic inventory decisions, stock out reductions, pull signal methods in all parts of
production, inventory turnover. Supports rapid analysis of dynamic material requirements
Line Design and Matches product mix with demand volume (both actual and forecast) with factory resources. Provides alerts to
Sequencing potential resource allocation bottlenecks
Kanban Control Enables real-time coordination on dynamic demand-pull requirements; tracks fulfillment performance and ensures
and Management production goals met in effective, timely fashion
1136
that can further improve Lean practices. Added possible. More ERP vendors will add Lean tools
then are benefits associated with having business and continue to improve those already in use.
software integrated and tied to the same database.
ERP vendors are competing with best-of-breed
software vendors to understand and offer solutions cONcLUsION
to support best practice implementations of Lean
production (Michel, 2002). A Lean-enabled ERP Overall, this research suggests Lean manufactur-
implementation must necessarily include a variety ing has been successfully integrated with ERP
of new modules, procedures, practices and toolsets software as a best-of-breed approach useful
to add new functions to existing ERP software to many SMEs and other manufacturers. With
(Nakashima, 2000). These modules often include: global competition heating up, small and medium
value stream analysis, value stream mapping, enterprises (SMEs), more than ever, must have
lean engineering, lean material flow, line design access to superior software systems and tools.
and sequencing, backwash capability, kanban This article has described how ERP (Enterprise
management, and others. Resource Planning) and Lean manufacturing
A variety of software packages and functional- have been implemented in various ways. In early
ity have emerged. Table 6 adapted from Halgeri incarnations, ERP systems were considered a
(2008) provides a glimpse at these cross refer- hindrance to Lean manufacturing efforts and were
enced with the three vendor solutions this article criticized for encouraging large inventories and
explored in more depth. slower production. In more recent years, it has
Certainly the future of ERP systems will in- become apparent that SMEs must seek the best
clude more Lean-enabled tools and continue to of both worlds and use the benefits of ERP and
capitalize on the synergy derived from the using retain the nimbleness provided by Lean. Linking
best Lean practices within an environment of ERP and Lean methods can lead to competitive
automatic data collection and access. SMEs will advantage as demonstrated in a number of soft-
continue to derive benefits from appropriate scaled ware solutions offered by vendors such as Oracle
versions of ERP systems without losing their edge TTW, and Pelion. These tools represent the future
in terms of nimbleness and ability to react to their of SMEs and their use of ERP systems.
customers’ needs. Larger organizations will gain
greater access to the methods developed in smaller
organizations and will find closer links and bet-
ter communication along the entire supply chain
Table 6. Software comparison
Oracle Flow Manufacturing TTW’s Winman Pelion systems Demad Flow

Lean Initiative (Turbide, 2005) (Wheatley, 2007) (Bragg, 2004; Pelion, 2008)
Analysis Tools √ - √
Mapping Tools √ - √
JIT Procurement Support √ √ √
Kanban Control √ √ √
Sequencing √ √ √
Demand Smoothing √ √ √
1137
rEFErENcEs Deep, A., Dani, S., & Burns, N. (2008). Investi-

gating factors affecting ERP selection in made-
Aladwani, A. M. (2001). Change management to-order SME sector. Journal of Manufacturing
strategies for successful ERP implementation. Technology Management, 19(4), 430–446.
Business Process Management Journal, 7(3), doi:10.1108/17410380810869905
266–275. doi:10.1108/14637150110392764
Gable, G., & Stewart, G. (1999). SAP R/3 imple-
Bartholomew, D. (1999). Lean vs. ERP. Industry mentation issues for small to medium enterprises.
Week, 248, 1–6. In W.D. Haseman & D.L. Nazareth (Eds.), Pro-
Bartholomew, D. (2003). ERP: Learning to be ceedings of the 5th Americas Conference on In-
Lean. Industry Week. Retrieved July 19, 2008, formation Systems (pp. 779-781), Milwaukee, WI.
from http://www.industryweek.com/ReadArticle. Garwood, D. (2002). ERP or flow manufacturing?
aspx?Article ID=2289. Collaboration, not separation. R.D. Garwood,
Bonvik, A. M., Couch, C. E., & Gershwin, S. Inc. Retrieved July 22, 2008 from http://www.
B. (1997). A comparison of production-line rdgarwood.com/archive/hot56.asp.
control mechanisms. International Journal Global Shop Solutions. (2008). Global solutions
of Production Research, 35(3), 789–804. products. Retrieved August 30, 2008 from http://
doi:10.1080/002075497195713 www.globalshopsolutions.com/erp-software/
Bradford, M., Mayfield, T., & Toney, C. (2001) default.asp.
Does ERP fit in a Lean world? Strategic Finance, Greene, A. (2004). Toyota production systems:
May, 28-34. Lean goes mainstream. Managing Automation
Bragg, S. (2004) Software solutions taking Lean (April). Retrieved July 20, 2008, from http://www.
manufacturing to the next level. Retrieved July managingautomation.com/maonline/magazine/
20, 2008, from http://www.oracle.com/lean/ read/view/Toyota_Production_Systems__Lean_
arc_leanmfg.pdf Goes_Mainstream_3874.
Burton, T. T., & Boeder, S. M. (2003). The Lean Halgeri, P., Pei, Z. J., Iyer, K. S., Bishop, K., &
extended enterprise: Moving beyond the four walls Shehadeh, A. (2008). ERP systems supporting
to value stream excellence. Fort Lauderdale, FL: Lean manufacturing: A literature review. 2008
J. Ross Publishing. International Manufacturing Science & Engineer-
ing Conference (MSEC), Evanston, IL, USA.
Cheng, T. C. E., & Podolsky, S. (1993). Just-in-
time manufacturing: An introduction (2nd ed.). Jacobs, R. (2007). Enterprise resource planning
London: Chapman & Hall. (ERP) - A brief history. Journal of Operations
Management, 25(2), 357–363. doi:10.1016/j.
Cox, J., & Blackstone, J. (Eds.). (2008). APICS jom.2006.11.005
dictionary (12th ed.). Chicago, IL: APICS Educa-
tional Society for Resource Manage. Kent, J. F. (2002). An examination of traditional
ERP and Lean manufacturing production con-
CTS. (2008). Manufacturing software reviews. trol methods with a view of flow manufacturing
Retrieved July 20, 2008 from http://www.cts- software as an alternative. MS thesis, University
guides.com/manufacturing.asp of Oregon.
1138
Koch, C. (2008). ABC: An introduction to ERP. Miller, G. J. (2002). Lean and ERP: Can they
CIO. Retrieved March 18, 2008, from http:// co-exist? Retrieved July 19, 2008 from http://
www.cio.com/article/40323/ABC_An_Introduc- facilitatorgroup.net/pdf/LeanERPCoExist.pdf
tion_to_ERP/1
Nagendra, P. B., & Das, S. K. (1999). MRP/SFX:
Koh, L., & Simpson, M. (2005). Change and A kanban-oriented shop floor extension to MRP.
uncertainty in SME manufacturing environ- Production Planning and Control, 10(3), 207–218.
ments using ERP. Journal of Manufacturing doi:10.1080/095372899233172
Technology Management, 16(6), 629–653.
Nakashima, B. (2000). Lean and ERP: Friend or
doi:10.1108/17410380510609483
foe? Advanced Manufacturing (September) Re-
Lee, T. S., & Adam, E. E. Jr. (1986). Forecasting trieved July 20, 2008 from http://www.advanced-
error evaluation in material requirements planning manufacturing.com/index.php?option=com_stati
(MRP) production-inventory systems. Manage- cxt&staticfile=informationtech.htm&Itemid=44
ment Science, 32(9), 1186–1205. doi:10.1287/
Oracle (2006). Oracle Flow Manufacturing
mnsc.32.9.1186
datasheet. Retrieved July 19, 2008 from ttp://
Li, Y., Liao, X. W., & Lei, H. Z. (2006). A knowl- www.oracle.com/applications/manufacturing/
edge management system for ERP implementa- flow-manufacturing-data-sheet.pdf.
tion. Systems Research and Behavioral Science,
Paez, O., Dewees, J., Genaidy, A., Tuncel, S.,
23(2), 157–168. doi:10.1002/sres.751
Karwowski, W., & Zurada, J. (2004). The Lean
McManus, H. L., & Richard, L. M. (2002). Value manufacturing enterprise: An emerging socio-
Stream analysis and mapping for product develop- technological system integration. Human Fac-
ment. In Proceedings of 23rd ICAS Congress (pp. tors and Ergonomics in Manufacturing, 14(3),
6103.1-6103.10). Toronto, Canada. 285–306. doi:10.1002/hfm.10067
Mekong Capital. (2004). Introduction to Lean Pelion Systems. (2008). Pelion Systems Solu-
manufacturing for Vietnam. Retrieved August tions. Retrieved July 22, 2008 from http://www.
30, 2008 from http://www.mekongcapital.com/ pelionsystems.com/solutions.asp
Introduction%20to%20Lean%20Manufactur-
Piszczalski, M. (2000). Lean vs. information sys-
ing%20-%20English.pdf
tems. Automotive Manufacturing & Production,
Metzger, B. (2008). Linking Lean and ERP systems 112(8), 26–28.
together for sustained advantage [White paper
Poppendieck, M. (2002) Principles of Lean think-
of TriMin Systems, Inc.]. Retrieved August 30,
ing (pp. 1-7). Poppendieck LLC.
2008 from http://www.triminmfg.com/images/
KnowledgeBase/Metzger.pdf Rashid, M. A., Hossain, L., & Patrick, J. D. (2002).
The evolution of ERP systems: A historical per-
Michel, R. (2002). Multiple paths to Lean: De-
spective. Hershey, PA: Idea Group.
tector Electronics, Norlen turn to specialized
Lean manufacturing solutions. Manufacturing Seradex. (2007). Lean Manufacturing - Seradex
Business Technology. Retrieved July 20, 2008 ERP Solutions. Retrieved September 3, 2008 from
from http://www.mbtmag.com/article/CA254538. http://www.seradex.com/ERP/Lean_Manufactur-
html?q=Lean+ERP+software ing_ERP.php
1139
Spearman, M. L., Hopp, W. J., & Woodruff, D. Turbide, D. A. (2005). Five ways ERP can help
L. (1999). A hierarchical control architecture for you implement Lean. EPICOR Software. Retrieved
constant work-in-process (CONWIP). Journal July 16, 2008 from http://whitepapers.zdnet.com/
of Manufacturing and Operations Management, abstract.aspx?docid=351964.
2(3), 147–171.
Wheatley, M. (2007). ERP is needed to sustain the
Spearman, M. L., & Zazanis, M. A. (1992). Push gains of Lean programs. Manufacturing Business
and pull production systems: Issues and com- Technology. Retrieved July 18, 2008 from http://
parisons. Operations Research, 40(3), 521–532. www.mbtmag.com/article/CA6450623.html
doi:10.1287/opre.40.3.521
WinMan. (2003). WinMan and Lean systems - A
Steger-Jensen, K., & Hvolby, H. (2008). Review white paper on integrating WinMan with Lean
of an ERP System Supporting Lean Manufactur- systems. Retrieved August 20, 2008 from http://
ing. In T. Koch (Ed.), International Federation www.winmanusa.com/PDF/WinMan_Lean_Sys-
for Information Processing (IFIP), Volume 257, tems.pdf
Lean Business Systems and Beyond (pp. 67-74).
WinMan. (2006). Athena Controls. Retrieved
Boston, MA: Springer.
August 20, 2008 from http://www.winmanusa.
Strategies, A. R. C. (2007). The when, why and com/success.asp
how of ERP support for Lean. SYSPRO, 1-22. Re-
Womack, J. P., & Jones, D. T. (1996). Beyond
trieved July 18, 2008 from http://www.syspro.com
Toyota: How to root out waste and pursue per-
Strategosinc (2008). Origins & history Lean fection. Harvard Business Review, (September-
Manufacturing. Retrieved July 18, 2008 from October): 140–158.
http://www.strategosinc.com/just_in_time.htm
Team, Q. S. G. (2006). Manufacturing can ac-
tively manage their value streams with Pelion’s
next generation EasyVSM tool. Retrieved July 20,
2008 from http://www.qsoftguide.com/cm/index.
php?blog=2&p=254&more=1&c=1&tb=1&pb=1
1140
1141
Chapter 4.17
Specifying Software Models
with Organizational Styles
Manuel Kolp
Université Catholique de Louvain Place des Doyens, Belguim
Yves Wautelet
Université Catholique de Louvain Place des Doyens, Belguim
Stéphane Faulkner
University of Namur Rempart de la Vierge, Belgium
Organizational Modeling is concerned with Analyzing the organizational and intentional con-
analyzing and understanding the organizational text within which a software system will eventu-
context within which a software system will ally operate has been recognized as an important
eventually function. This chapter proposes orga- element of the organizational modeling process
nizational patterns motivated by organizational also called early requirements engineering (see
theories intended to facilitate the construction of e.g., (Anton 1996, Dardenne, van Lamsweerde
organizational models. These patterns are defined & Fickas 1993, Yu 1995)). Such models are
from real world organizational settings, modeled founded on primitive concepts such as those
in i* and formalized using the Formal Tropos of actor and goal. This chapter focuses on the
language. Additionally, the chapter evaluates the definition of a set of organizational patterns that
proposed patterns using desirable qualities such as can be used as building blocks for constructing
coordinability and predictability. The research is such models. Our proposal is based on concepts
conducted in the context of Tropos, a comprehen- adopted from organization theory and strategic
sive software system development methodology. alliances literature. Throughout the paper, we
use i* (Yu 1995) as the modeling framework in
DOI: 10.4018/978-1-60566-146-9.ch006 terms of which the proposed patterns are presented
Specifying Software Models with Organizational Styles
and accounted for. The research reported in this structure of an organization. Section 3 details two
paper is being conducted within the context of organizational patterns – the structure-in-5 and
the Tropos project (Giorgini, Kolp, Mylopoulos the joint venture – based on real world examples
& Pistore 2004, Giorgini, Kolp, Mylopoulos of organizations. These patterns are modeled in
& Castro 2005), whose aim is to construct and terms of social and intentional concepts using the
validate a software development methodology for i* framework and the Formal Tropos specification
agent-based software systems. The methodology language. Section 4 identifies a set of desirable
adopts ideas from multi-agent system technolo- non-functional requirements for evaluating these
gies, mostly to define the implementation phase patterns and presents a framework to select a pat-
of our methodology. It also adopts ideas from tern with respect to these identified requirements.
Requirements Engineering, where actors and goals Section 5 overviews the Tropos methodology.
have been used heavily for early requirements Finally, Section 6 summarizes the contributions
analysis. The project is founded on that actors of the chapter and overviews related work.
and goals are used as fundamental concepts for
modeling and analysis during all phases of soft-
ware development, not just early requirements, or strUctUrING OrGANIZAtIONs
implementation. More details about Tropos can
be found in (Giorgini et al. 2005). The present Organizational structures are primarily studied by
work continues the research in progress about Organization Theory (e.g., (Mintzberg 1992, Scott
social abstractions for the Tropos methodology. In 1998, Yoshino & Rangan 1995)), that describes
(Kolp, Giorgini & Mylopoulos 2002a), we have the structure and design of an organization and
detailed a social ontology for Tropos to consider Strategic Alliances (e.g., (Dussauge & Garrette
information systems as social structures all along 1999, Gomes-Casseres 1996, Morabito, Sack &
the development life cycle. In (Giorgini, Kolp & Bhate 1999, Segil 1996)), that model the strate-
Mylopoulos 2002, Kolp, Giorgini & Mylopoulos gic collaborations of independent organizational
2002b, Kolp, Giorgini & Mylopoulos 2006), we stakeholders who have agreed to pursue a set of
have described how to use this Tropos social ontol- agreed upon business goals.
ogy to design multi-agent systems architectures, Both disciplines aim to identify and study
notably for e-business applications (Kolp, Do & organizational patterns that describe a system
Faulkner 2004). As a matter of fact, multi-agent at a macroscopic level in terms of a manageable
systems can be considered structured societies of number of subsystems, components and modules
coordinated autonomous agents. In the present inter-related through dependencies.
paper, which is a extended and revised version of In this chapter, we are interested to identify,
(Kolp, Giorgini & Mylopoulos 2003), we empha- formalize and apply, for organizational modeling,
size the use of organizational patterns based on patterns that have been already well-understood
organization theory an strategic alliances for early and precisely defined in organizational theories.
requirements analysis, with the concern of model- Our purpose is not to categorize them exhaustively
ing the organizational setting for a system-to-be nor to study them on a managerial point of view.
in terms of abstractions that could better match The following sections will thus only insist on
its operational environment (e.g., an enterprise, patterns that have been found, due to their nature,
a corporate alliance, . . .) interesting candidates also considering the fact
The paper is organized as follows. Section 2 that they have been studied in great detail in the
describes organizational and strategic alliance organizational literature and presented as fully
theories, focusing on the internal and external formed patterns.
1142
Organization theory standardizing work processes, outputs, and skills.

It is also in charge of applying analytical proce-
“An organization is a consciously coordinated dures to adapt the organization to its operational
social entity, with a relatively identifiable bound- environment. The Support provides specialized
ary, that functions on a relatively continuous services, at various levels of the hierarchy, outside
basis to achieve a common goal or a set of goals” the basic operating workflow (e.g., legal counsel,
(Morabito et al. 1999). Organization theory is the R&D, payroll, cafeteria). We describe and model
discipline that studies both structure and design examples of structures-in-5 in Section 3.
in such social entities. Structure deals with the The pyramid pattern is the well-know hier-
descriptive aspects while design refers to the archical authority structure. Actors at lower levels
prescriptive aspects of a social entity. Organiza- depend on those at higher levels. The crucial
tion theory describes how practical organizations mechanism is the direct supervision from the Apex.
are actually structured, offers suggestions on how Managers and supervisors at intermediate levels
new ones can be constructed, and how old ones only route strategic decisions and authority from
can change to improve effectiveness. To this end, the Apex to the operating (low) level. They can
since Adam Smith, schools of organization theory coordinate behaviors or take decisions by their
have proposed models and patterns to try to find own, but only at a local level.
and formalize recurring organizational structures The chain of values merges, backward or
and behaviors. forward, several actors engaged in achieving or
In the following, we briefly present organiza- realizing related goals or tasks at different stages
tional patterns identified in Organization Theory. of a supply or production process. Participants
The structure-in-5 will be studied in detail in who act as intermediaries, add value at each step
Section 3. of the chain. For instance, for the domain of goods
The Structure-in-5. An organization can be distribution, providers are expected to supply
considered an aggregate of five substructures, as quality products, wholesalers are responsible for
proposed by Minztberg (Mintzberg 1992). At the ensuring their massive exposure, while retailers
base level sits the Operational Core which car- take care of the direct delivery to the consumers.
ries out the basic tasks and procedures directly The matrix proposes a multiple command
linked to the production of products and services structure: vertical and horizontal channels of in-
(acquisition of inputs, transformation of inputs into formation and authority operate simultaneously.
outputs, distribution of outputs). At the top lies the The principle of unity of command is set aside,
Strategic Apex which makes executive decisions and competing bases of authority are allowed to
ensuring that the organization fulfill its mission in jointly govern the workflow. The vertical lines
an effective way and defines the overall strategy are typically those of functional departments
of the organization in its environment. The Middle that operate as ”home bases” for all participants,
Line establishes a hierarchy of authority between the horizontal lines represents project groups or
the Strategic Apex and the Operational Core. It geographical arenas where managers combine and
consists of managers responsible for supervising coordinate the services of the functional specialists
and coordinating the activities of the Operational around particular projects or areas.
Core. The Technostructure and the Support are The bidding pattern involves competitivity
separated from the main line of authority and mechanisms, and actors behave as if they were
influence the operating core only indirectly. The taking part in an auction. An auctioneer actor runs
Technostructure serves the organization by mak- the show, advertises the auction issued by the
ing the work of others more effective, typically by auction issuer, receives bids from bidder actors
1143
and ensures communication and feedback with local dimension and interact directly with other
the auction issuer who is responsible for issuing partners to exchange resources, such as data and
the bidding. knowledge. However, the strategic operation and
coordination of such an organization, and its ac-
strategic Alliances tors on a global dimension, are only ensured by
the joint management actor in which the original
A strategic alliance links specific facets of two or actors possess equity participations. We describe
more organizations. At its core, this structure is a and model examples of joint ventures in Section 3.
trading partnership that enhances the effectiveness The arm’s-length pattern implies agree-
of the competitive strategies of the participant ments between independent and competitive,
organizations by providing for the mutually ben- but partner actors. Partners keep their autonomy
eficial trade of technologies, skills, or products and independence but act and put their resources
based upon them. An alliance can take a variety and knowledge together to accomplish precise
of forms, ranging from arm’s-length contracts to common goals. No authority is lost, or delegated
joint ventures, from multinational corporations from one collaborator to another.
to university spin-offs, from franchises to equity The hierarchical contracting pattern iden-
arrangements. Varied interpretations of the term tifies coordinating mechanisms that combine
exist, but a strategic alliance can be defined as arm’s-length agreement features with aspects of
possessing simultaneously the following three pyramidal authority. Coordination mechanisms
necessary and sufficient characteristics: developed for arm’s-length (independent) charac-
teristics involve a variety of negotiators, mediators
• The two or more organizations that unite to and observers at different levels handling condi-
pursue a set of agreed upon goals remain tional clauses to monitor and manage possible
independent subsequent to the formation contingencies, negotiate and resolve conflicts and
of the alliance. finally deliberate and take decisions. Hierarchi-
• The partner organizations share the ben- cal relationships, from the executive apex to the
efits of the alliances and control over the arm’s-length contractors restrict autonomy and
performance of assigned tasks. underlie a cooperative venture between the parties.
• The partner organizations contribute on a The co-optation pattern involves the incor-
continuing basis in one or more key stra- poration of representatives of external systems
tegic areas, e.g., technology, products, and into the decision-making or advisory structure
so forth. and behavior of an initiating organization. By
co-opting representatives of external systems,
In the following, we briefly present organiza- organizations are, in effect, trading confidentiality
tional patterns identified in Strategic Alliances. and authority for resource, knowledge assets and
The joint venture will be studied in details in support. The initiating system has to come to terms
Section 3. with the contractors for what is being done on its
The joint venture pattern involves agreement behalf; and each co-optated actor has to reconcile
between two or more intra-industry partners to and adjust its own views with the policy of the
obtain the benefits of larger scale, partial invest- system it has to communicate.
ment and lower maintenance costs. A specific joint
management actor coordinates tasks and manages
the sharing of resources between partner actors.
Each partner can manage and control itself on a
1144
MODELING OrGANIZAtIONAL The Direction – four directors responsible for

PAttErNs the main aspects of LDV Bates’s Global Strategy
(advertising campaigns, creative activities, ad-
We will define an organizational pattern as a meta- ministration, and finances) – forms the Strategic
class of organizational structures offering a set of Apex. The Middle Line, composed of the Cam-
design parameters to coordinate the assignment of paigns Management staff, is in charge of finding
organizational objectives and processes, thereby and coordinating advertising campaigns (market-
affecting how the organization itself functions. ing, sales, edition, graphics, budget, . . .). It is
Design parameters include, among others, goal supported in these tasks by the Administration
and task assignments, standardization, supervision and Accounts and IT and Documentation depart-
and control dependencies and strategy definitions. ments. The Administration and Accounts consti-
This section describes two of the organizational tutes the Technostructure handling administrative
patterns presented in Section 2: the structure-in-5 tasks and policy, paperwork, purchases and bud-
and the joint-venture. gets. The Support groups the IT and Documenta-
tion departments. It defines the IT policy of Agate,
structure-in-5 provides technical means required for the man-
agement of campaigns, and ensures services for
To detail and specify the structure-in-5 as an orga- system support as well as information retrieval
nizational pattern, this section presents two case (documentation resources). The Operational Core
studies: LDV Bates (Bates 2006) and GMT (GMT includes the Graphics and Edition staff in charge
2006). They will serve to propose a model and a of the creative and artistic aspects of realizing
semi-formal specification of the structure-in-5. campaign (texts, photographs, drawings, layout,
LDV Bates. Agate Ltd is an advertising agency design, logos).
located in Belgium that employs about fifty staff, Figure 1 models LDV Bates in structure-in-5
as detailed in Table 1. using the i* strategic dependency model. i* is a
Table 1. Organization of LDV Bates
Direction Edition
1 Campaigns Director 2 Editors IT
1 Creative Director 4 Copy writers 1 IT manager
1 Administrative Director 1 Network administrator
1 Finance Director Documentation 1 System administrator
1 Media librarian 1 Analyst
Campaigns Management 1 Resource librarian 1 Computer technician
2 Campaign managers 1 Knowledge worker
3 Campaign marketers
1 Editor in Chief Administration Accounts
1 Creative Manager 3 Direction assistants 1 Accountant manager
4 Manager Secretaries 1 Credit controller
Graphics 2 Receptionists 2 Accounts clerks
6 Graphic designers 2 Clerks/typists 2 Purchasing assistants
2 Photographers 1 Filing clerk
1145
Figure 1. LDV Bates as a Structure-in-5
modeling framework for organizational modeling clouds, hexagons and rectangles; respectively,
(Yu 1995), which offers goal-and actor-based and dependencies have the form depender →
notions such as actor, agent, role, position, goal, dependum → dependee.
softgoal, task, resource, belief and different kinds GMT is a company specialized in telecom
of social dependency between actors. Its strate- services in Belgium. Its lines of products and
gic dependency model describes the network of services range from phones & fax, conferencing,
social dependencies among actors. It is a graph, line solutions, internet & e-business, mobile solu-
where each node represents an actor and each tions, and voice & data management. As shown
link between two actors indicates that one actor in Figure 2, the structure of the commercial or-
depends on the other for some goal to be attained. ganization follows the structure-in-5. An Execu-
A dependency describes an “agreement” (called tive Committee constitutes the Strategic Apex. It
dependum) between two actors: the depender and is responsible for defining the general strategy
the dependee. The depender is the depending ac- of the organization. Five chief managers (fi-
tor, and the dependee, the actor who is depended nances, operations, divisions management, mar-
upon. The type of the dependency describes the keting, and R&D) apply the specific aspects of
nature of the agreement. Goal dependencies the general strategy in the area of their compe-
represent delegation of responsibility for fulfill- tence: Finances & Operations is in charge of
ing a goal; softgoal dependencies are similar to Budget and Sales Planning & Control, Divisions
goal dependencies, but their fulfillment cannot Management is responsible for Implementing
be defined precisely (for instance, the apprecia- Sales Strategy, and Marketing and R&D define
tion is subjective or fulfillment is obtained only Sales Policy and Technological Policy.
to a given extent); task dependencies are used in The Divisions Management groups managers
situations where the dependee is required to per- that coordinate all managerial aspects of product
form a given activity; and resource dependencies and service sales. It relies on Finance & Opera-
require the dependee to provide a resource to the tions for handling Planning and Control of prod-
depender. As shown in Figure 1, actors are repre- ucts and services, it depends on Marketing for
sented as circles; dependums – goals, softgoals, accurate Market Studies and on R&D for Tech-
tasks and resources – are represented as ovals, nological Awareness.
1146
Figure 2. GMT’s sales organization as a Structure-in-5
The Finances & Operations departments • The dependencies between the Strategic
constitute the technostructure in charge of man- Apex as depender and the Technostructure,
agement control (financial and quality audit) and Middle Line and Support as dependees
sales planning including scheduling and resource must be of type goal
management. • A softgoal dependency models the stra-
The Support involves the staff of Marketing and tegic dependence of the Technostructure,
R&D. Both departments jointly define and sup- Middle Line and Support on the Strategic
port the Sales Policy. The Marketing department Apex
coordinates Market Studies (customer position- • The relationships between the Middle Line
ment and segmentation, pricing, sales incentive, and Technostructure and Support must be
. . .) and provides the Operational Core with of goal dependencies
Documentation and Promotion services. The R&D • The Operational Core relies on the
staff is responsible for defining the technological Technostructure and Support through task
policy such as technological awareness services. and resource dependencies
It also assists Sales people and Consultants with • Only task dependencies are permitted be-
Expertise Support and Technology Training. tween the Middle Line (as depender or de-
Finally, the Operational Core groups the Sales pendee) and the Operational Core (as de-
people and Line consultants under the supervision pendee or depender).
and coordination of Divisions Managers. They
are in charge of selling products and services to To specify the formal properties of the pattern,
actual and potential customers. we use Formal Tropos (Fuxman, Liu, Mylopoulos,
Figure 3 abstracts the structures explored in the Roveri & Traverso 2004), which extends the primi-
case studies of Figures 1 and 2 as a Structure-in-5 tives of i* with a formal language comparable to
pattern composed of five actors. The case studies that of KAOS (Dardenne et al. 1993). Constraints
also suggested a number of constraints to supple- on i* specifications are thus formalized in a first-
ment the basic pattern: order linear-time temporal logic. Formal Tropos
1147
Figure 3. The structure-in-5 pattern
provides three basic types of metaclasses: actor, al. 2004) for more details). Special predicates can
dependency, and entity (Giorgini, Kolp & Mylo- appear in the temporal logic formulas: predicate
poulos 2002). The attributes of a Formal Tropos JustCreated(x) holds in a state if element x exists
class denote relationships among different objects in this state but not in the previous one; predicate
being modeled. Fulfilled(x) holds if x has been fulfilled; and predi-
cate JustFulfilled(x) holds if Fulfilled(x) holds in
Metaclasses this state, but not in the previous one.
Actor:= Actor name[attributes] [creation-prop- In the following, we only present some
erties] [invar-properties][actor-goal] specifications for the Strategic Management and
With subclasses: Operational Management dependencies.
Agent(with attributes occupies: Position, play:
Role) Actor StrategicApex
Position(with attributes cover: Role) Actor MiddleLine
Role Actor Support
Dependency:= Dependency name type mode Actor Technostructure
Depender name Dependee name [attributes] Actor OperationalCore
[creation-properties] [invar-properties] [fulfill- Dependency StrategicManagement
properties] Type SoftGoal
Entity:=Entity name [attribute] [creation-prop- Depender te: Technostructure, ml: MiddleLine,
erties][invar-properties] su: Support
Actor-Goal:= (Goal|Softgoal) name mode Dependee sa: StrategicApex
FulFillment(actor-fulfill-property) Invariant
Classes: Classes are instances of Metaclasses. ∀dep: Dependency (JustCreated(dep) →
Consistent(self, dep)) ∀ag: Actor - Goal
In Formal Tropos, constraints on the lifetime (JustCreated(ag) → Consistent(self, ag))
of the (meta)class instances are given in a first- Fulfillment
order linear-time temporal logic (see (Fuxman et
1148
∀dep: Dependency (dep.type = goal ∧ dep.de- Core) that implement (ImplementedBy) the Op-
pender = sa ∧ erationalManagaemnt goal.]
(dep.dependee = te dep.dependee = ml dep.de- In addition, the following structural (global)
pendee = su)) ∧ properties must be satisfied for the Structure-in-5
Fulfilled(self) → ♦Fulfilled(dep) pattern:
[Invariant properties specify, respectively, that • ∀inst1, inst2: StrategicApex → inst1 =

the strategic management softgoal must be consis- inst2
tent with any other dependency of the organiza- [There is a single instance of the
tion and with any other goal of the actors in the Strategic Apex (the same constraint
organization. The predicate Consistent depends also holds for the Middle Line, the
on the particular organization we are considering Technostructure, the Support and the
and it is specified in terms of goals’ properties to Operational Core)]
be satisfied. The fulfillment of the dependency • ∀sa: StrategicApex, te: Technostructure,
necessarily implies that the goal dependencies ml: MiddleLine,
between the Middle Line, the Technostructure, su: Support, dep: Dependency
and the Support as dependees, and the Strategic (dep.dependee = sa ∧ (dep.depender
Apex as depender have been achieved some time = te ∨ dep.depender = ml
in the past] ∨ dep.depender = su) → dep.type =
softgoal)
Dependency OperationalManagement Type Goal [Only softgoal dependencies are per-
Mode achieve mitted between the Strategic Apex as
Depender sa: StrategicApex dependee and the Technostructure,
Dependee ml: MiddleLine the Middle Line, and the Support as
Invariant dependers]
Consistent(self, StrategicM anagement) • ∀sa: StrategicApex, te: T echnostructure,
∃ c: Coordination (c.type = task ∧ c.dependee ml: M iddleLine,
= ml ∧ c.depender = OperationalCore ∧ su: Support, dep: Dependency:
ImplementedBy(self, c)) (dep.depender = sa ∧ (dep.dependee
Fulfillment = te ∨ dep.dependee =
∀ts: Technostructure, dep: Dependency (dep.type ml ∨ dep.dependee = su) → dep.type
= goal ∧ = goal)
dep.depender = ml ∧ dep.dependee = ts) ∧ [Only goal dependencies are permit-
Fulfilled(self)) → ♦Fulfilled(dep) ted between the Technostructure,
the Middle Line, and the Support as
[The fulfillment of the Operational manage- dependee, and the Stategic Apex as
ment goal implies that all goal dependencies depender]
between the Middle Line as depender and the • ∀su: Support, ml: M iddleLine, dep:
Technostructure as dependee have been achieved Dependency
some time in the past. Invariant properties speci- ((dep.dependee = su ∧ dep.depender
fies that Operational Management goal has to be = ml) → dep.type = goal)
consistent with Strategic Management softgoal [Only task dependencies are permitted
and that there exists a coordination task (a task between the Middle Agency and the
dependency between MiddleLine and Operational Operational Core]
1149
• ∀te: T echnostructure, oc: OperationalCore, Joint Venture

dep: Dependency
((dep.dependee = te ∧ dep.depender We describe here two alliances – Airbus (Dussauge
= oc) → & Garrette 1999) and a more detailed one, Carsid
(dep.type = task ∨ dep.type = re- (Wautelet, Kolp & Achbany 2006) – that will
source)) serve to model the joint venture structure as an
[Only resource or task dependencies are organizational pattern and propose a semi-formal
permitted between the Technostructure specification.
and the Operational Core (the same Airbus. The Airbus Industrie joint venture
constraint also holds for the Support)] coordinates collaborative activities between
• ∀a: Actor, ml: M iddleLine, European aeronautic manufacturers to built
(∃dep: Dependency(dep.depender = a and market airbus aircrafts. The joint venture
∧ dep.dependee = involves four partners: British Aerospace (UK),
ml) ∨ (dep.dependee = a ∧ dep.de- Aerospatiale (France), DASA (Daimler-Benz
pender = ml) → Aerospace, Germany) and CASA (Construcciones
((∃sa: StrategicApex(a = sa)) ∨ (∃su: Aeronauticas SA, Spain). Research, develop-
Support(a = su) ∨ ment and production tasks have been distributed
(∃te: T echnostructure(a = te)) ∨ (∃op: among the partners, avoiding any duplication.
OperationalCore Aerospatiale is mainly responsible for develop-
(a = op)) ing and manufacturing the cockpit of the aircraft
[No dependency is permitted between and for system integration. DASA develops and
an external actor and the Middle manufactures the fuselage, British Aerospace the
Agency (the same constraint also holds wings and CASA the tail unit. Final assembly is
for the Operational Core)] carried out in Toulouse (France) by Aerospatiale.
Unlike production, commercial and decisional
This specification can be used to establish that activities have not been split between partners.
a certain i* model does constitute an instance of All strategy, marketing, sales and after-sales op-
the Structure-in-5 pattern. For example, the i* erations are entrusted to the Airbus Industrie joint
model of Figure 1 can be shown to be such an venture, which is the only interface with external
instance, in which the actors are instances of the stakeholders such as customers. To buy an Airbus,
Structure-in-5 actor classes (e.g., Direction and or to maintain their fleet, customer airlines could
IT&Documentation are instances of the Strategic not approach one or other of the partner firms
Apex and the Support, respectively), dependen- directly, but has to deal with Airbus Industrie.
cies are instances of Structure-in-5 dependencies Airbus Industrie, which is a real manufacturing
classes (e.g., Agency Global Strategy is an instance company, defines the alliance’s product policy and
the Strategic Management), and all above global elaborates the specifications of each new model of
properties are enforced (e.g., since there are only aircraft to be launched. Airbus defends the point
two task dependencies between Campaigns of view and interests of the alliance as a whole,
Management and Graphics&Edition, the fourth even against the partner companies themselves
property holds). when the individual goals of the latter enter into
conflict with the collective goals of the alliance.
Figure 4 models the organization of the Air-
bus Industrie joint venture using the i* strategic
1150
Figure 4. The Airbus Industrie Joint Venture
dependency model. Airbus assumes two roles: Duferco (Italy), Usinor (France) – that also par-
Airbus Industrie and Airbus Joint Venture. tially owns Cockerill-Sambre (Belgium) through
Airbus Industrie deals with demands from the Arcelor group – and Sogepa (Belgium), a public
customers, Customer depends on it to receive investment company, representing the Walloon
airbus aircrafts or maintenance services. The Region Government. Usinor has also brought its
Airbus Joint Venture role ensures the interface for subsidiary Carlam in the alliance.
the four partners (CASA, Aerospatiale, British Roughly speaking, the aim of a steel manufac-
Aerospace and DASA) with Airbus Industrie turing company like CARSID is to extract iron
defining Airbus strategic policy, managing con- from the ore and to turn it into semi-finished steel
flicts between the four Airbus partners, defending products. Several steps compose the transforma-
the interests of the whole alliance and defining tion process, each step is generally assumed by a
new aircrafts specifications. Airbus Joint Venture specific metallurgic plant:
coordinates the four partners ensuring that each
of them assumes a specific task in the building of • Sintering Plant. Sintering is the prepara-
Airbus aircrafts: wings building for British Aero- tion of the iron ore for the blast furnace.
space, tail unit building for CASA, cockpit build- The minerals are crushed and calibrated to
ing and aircraft assembling for Aerospace and form a sinter charge.
fuselage building for DASA. Since Aerospatiale • Coking Plant. Coal is distilled (i.e., heat-
assumes two different tasks, it is modeled as two ed in an air-impoverished environment in
roles: Aerospatiale Manufacturing and Aerospa- order to prevent combustion) to produce
tiale Assembling. Aerospatiale Assembling de- coke.
pends on each of the four partners to receive the • Blast Furnace. Coke is used as a combus-
different parts of the planes. tion agent and as a reducing agent to re-
Carsid (Carolo-Sidérurgie) is a joint venture moves the oxygen from the sinter charge.
that has recently arisen from the global concentra- The coke and sinter charge are loaded
tion movement in the steel industry. The alliance, together into the blast furnace to produce
physically located in the steel basin of Charleroi in cast iron.
Belgium, has been formed by the steel companies
1151
• Steel Making Plant. Different steps (de- ments, rich gases from the different plants, go-
sulphuration, oxidation, steel adjustment, droon, naphtalin, etc.) to external industries such
cooling, . . .) are necessary to turn cast iron as vehicle (automobile, train, boat, . . .) manufac-
into steel slabs and billets. First, elements turers, foundries, gas companies, building com-
other that iron are remove to give molten panies. It is also in charge of the proper environ-
steel. Then supplementary elements (tita- ment policy, a strategic aspect for steelworks that
nium, niobium, vanadium, . . .) are added are polluting plants. Most important, Carsid has
to make a more robust alloy. Finally, the been set up with the help of the Walloon Region
result – finished steel – is solidified to pro- to guarantee job security for about 2000 workers
duce slabs and billets. in the basin of Charleroi. Indeed, the steel indus-
• Rolling Mill. The manufacture of semi-fin- try in general and the Walloon metallurgical basins
ished products involves a process known in particular are sectors in difficulty with high
as hot rolling. Hot-rolled products are of unemployment rates. As a corrolar, the joint ven-
two categories: flat (plates, coiled sheets, ture is committed to improve regional economy
sheeting, strips, . . .) produced from steel and maintain work in the region. Carsid has then
slabs and long (wire, bars, rails, beams, been contractually obliged to plan maintenance
girders, . . .) produced from steel billets. investment (e.g., blast furnace refection, renova-
tion of coke oven batteries, . . .) and develop
Figure 5 models the organization of the Carsid production plans involving regional sub contrac-
joint venture in i*. Carsid assumes two roles Carsid tors and suppliers. Since steelmaking is a hard
S.A. (”Société Anonyme” – the english equivalent and dangerous work sector, Carsid, like any
is ”Ltd”) and Carsid Joint Venture. other steelworks, is legally committed to respect,
Carsid S. A. is the legal and contractual inter- develop and promote accident prevention stan-
face of the joint venture. It handles the sales of dards.
steel semi-finished products (bars, plates, rails, The Carsid joint venture itself coordinates the
sheets, etc. but also slabs, billets) and co-products steel manufacturing process. The sintering phase
(coke that does not meet blast furnace require- to prepare iron ore is the responsibility of Duferco
Figure 5. The Carsid Joint Venture
1152
Sintering Plant while Usinor Coking Plant, dis- Part of the Joint Venture pattern specification
tills coal to turn it onto coke. The sinter charge is in the following:
and coke are used by Cokerill Blast Furnace to
produce cast iron by removing oxygen from sinter. Role JointManagerPrivateInterface Goal
Duferco Steel Making Plant transforms cast iron CoordinatePatterns
into steel to produce slabs and billets for Carlam Role JointManagerPublicInterface
Rolling Mill in charge of the hot rolling tasks. Goal TakeStrategicDecision
Carlam (Carolo-Laminoir). Sogepa, the public SoftGoal RepresentPartnershipInterests
partner, has the responsibility to develop regional Actor Partner
initiative to promote Carsid activities, particularly
in the Walloon Region and in Belgium.
Figure 6 abstracts the joint venture structures and the following structural (global) properties
explored in the case studies of Figures 4 and 5. must be satisfied:
The case studies suggest a number of constraints
to supplement the basic pattern: • ∀jmpri1, jmpri2: J
ointManagerPrivateInterface
• Partners depend on each other for provid- (jmpri1 = jmpri2)
ing and receiving resources. [Only one instance of the joint man-
• Operation coordination is ensured by the ager]
joint manager actor which depends on • ∀p1, p2: P artner, dep: Dependency (((dep.
partners for the accomplishment of these depender = p1 ∧ dep.dependee = p2) ∨
assigned tasks. (dep.depender =
• The joint manager actor must assume two p2 ∧ dep.dependee = p1)) → (dep.type
roles: a private interface role to coordinate = resource))
partners of the alliance and a public inter- [Only resource dependencies between
face role to take strategic decisions, define partners]
policy for the private interface and repre- • ∀jmpri: JointManagerPrivateInterface, p:
sents the interests of the whole partnership Partner, dep: Dependency((dep.dependee
with respect to external stakeholders. = p ∧ dep.depender = jmpri)
→dep.type = task)
Figure 6. The Joint Venture pattern
1153
[Only task dependencies between (Wooldridge & Jennings 1995) in the way that
partners and the joint manager, with they undertake action and communication in
the joint manager as depender] their domains. It can be then difficult to predict
• ∀jmpri: JointM anagerPrivateInterf ace, individual characteristics as part of determining
jmpui: JointM anagerPublicInterf ace, the behavior of the system at large. Generally,
dep: Dependency ((dep.depender = jmpri predictability is in contrast with the actors ca-
∧ dep.dependee = jmpui) pabilities to be adaptive and responsive: actors
→ (dep.type = goal ∨ dep.type = must be predictable enough to anticipate and plan
softgoal)) actions while being responsive and adaptive to
[Only goal or softgoal dependencies unexpected situations.
between the joint manager roles] Security. Actors are often able to identify
• ∀dep: Dependency, p1: Partner ((dep.de- their own data and knowledge sources and they
pender = p1 ∨ dep.dependee = p1) → may undertake additional actions based on these
((∃p2: P artner(p1≠ p2 ∧ (dep.de- sources (Woods & Barbacci 1999). Strategies
pender = p2 ∨ dep.dependee = p2)) ∨ for verifying authenticity for these data sources
(∃jmpi: J ointM anagerP rivateInterf by individual actors are an important concern in
ace ((dep.depender = jmpi ∨ dep. the evaluation of overall system quality since,
dependee = jmpi)))) in addition to possibly misleading information
[Partners only have relationships with acquired by actors, there is the danger of hostile
other partners or the joint manager external entities spoofing the system to acquire
private interface] information accorded to trusted domain actors.
• ∀dep: Dependency, jmpi: J ointM anagerP Adaptability. Actors may be required to adapt
rivateInterf ace ((dep.depender = jpmi ∨ to modifications in their environment. They may
dep.dependee = jpmi) → include changes to the component’s communica-
((∃p: P artner((dep.depender = p ∨ tion protocol or possibly the dynamic introduction
dep.dependee = p))) ∨ (∃jmpui: J of a new kind of component previously unknown
ointM anagerP ublicInterf ace ((dep. or the manipulations of existing actors.
depender = jmpui ∨ dep.dependee = Generally, adaptability depends on the capabili-
jmpui)))) ties of the single actors to learn and predict the
[The joint manager private interface changes of the environments in which they act
only has relationships with the joint (Weiss 1997), and also their capability to make
manager public interface or partners] diagnosis (Horling, Lesser, Vincent, Bazzan &
Xuan 1999), that is being able to detect and deter-
EVALUAtION mine the causes of a fault based on its symptoms.
However, successful organization environments
Patterns can be compared and evaluated with tend to balance the degree of reactivity and predict-
quality attributes (Shaw & Garlan 1996), also ability of the single actors with their capabilities
called non-functional requirements (Chung, to be adaptive.
Nixon, Yu & Mylopoulos 2000) For instance, Coordinability. Actors are not particularly
the requirements seem particularly relevant for useful unless they are able to coordinate with
organizational structures (Do, Faulkner & Kolp other agents. Coordination is generally (Jennings
2003, Kolp et al. 2006): 1996) used to distribute expertise, resources or
Predictability (Woods & Barbacci 1999). information among the actors (actors may have
Actors can have a high degree of autonomy different capabilities, specialized knowledge, dif-
1154
ferent sources of information, resources, respon- prevent failure, different actors can have similar
sibilities, limitations, charges for services, etc.), or replicated capabilities and refer to more than
solve interdependencies between actors’ actions one actor for a specific behavior.
(interdependence occur when goal undertaken Modularity (Shehory 1998) increases effi-
by individual actors are related), meet global ciency of service execution, reduces interaction
constraints (when the solution being developed overhead and usually enables high flexibility. On
by a group of actors must satisfy certain condi- the other hand, it implies constraints on inter-
tions if is to be deemed successful), and to make organization communication.
the system efficient (even when individuals can Aggregability. Some actors are parts of
function independently, thereby obviating the need other actors. They surrender to the control of the
for coordination, information discovered by one composite entity. This control results in efficient
actor can be of sufficient use to another actor that workflow execution and low interaction overhead,
both actors can solve the problem twice as fast). however prevents the organization to benefit from
Coordination can be realized in two ways: flexibility.
As an illustration, we evaluate the patterns
• Cooperativity. Actors must be able to with respect to coordinativity, predictability,
coordinate with other entities to achieve fallibility-tolerance and adaptability. The evalu-
a common purpose or simply their local ation can be done in a similar way for the other
goals. Cooperation can either be communi- non-functional requirements. Due to the lack of
cative in that the actors communicate (the space, we refer the author to the bibliography for
intentional sending and receiving of sig- the other attributes.
nals) with each other in order to cooperate
or it can be non-communicative (Doran, • The structure-in-5 improves coordinativ-
Franklin, Jennings & Norman 1997). In ity among actors by differentiating the data
the latter case, actors coordinate their co- hierarchy -the support actor -from the con-
operative activity by each observing and trol hierarchy -supported by the operational
reacting to the behaviour of the other. In core, technostructure, middle agency and
deliberative organizations, actors jointly strategic apex. The existence of three dif-
plan their actions so as to cooperate with ferent levels of abstraction (1 -Operational
each other. Core; 2 -Technostructure, Middle Line and
• Competitivity. Deliberative negotiating Support; 3 -Strategic Apex) addresses the
organization (Doran et al. 1997) are like need for managing predictability. Besides,
deliberative one, except that they have an higher levels are more abstract than lower
added dose of competition. The success of levels: lower levels only involve resources
one actors implies the failure of others. and task dependencies while higher ones
propose intentional (goals and soft-goals)
Availability. Actors that offer services to other relationships. Checks and control mecha-
actors must implicitly or explicitly guard against nisms can be integrated at different levels
the interruption of offered services. of abstraction assuming redundancy from
Fallibility-Tolerance. A failure of one actor different perspectives and increase con-
does not necessarily imply a failure of the whole siderably fallibility-tolerance. Since the
organization. The organization then needs to check structure-in-5 separates data and control
the completeness and the accuracy of information hierarchies, integrity of these two hierar-
and knowledge transactions and workflows. To chies can also be verified independently.
1155
The structure-in-5 separates independently precise and evaluates alternative organizational

the typical components of an organization, patterns against them, as shown in Figure 7. The
isolating them from each other and allow- analysis is intended to make explicit the space of
ing then dynamic adaptability. But since it alternatives for fulfilling the top-level require-
is restricted to no more than 5 major comments. The patterns are represented as operation-
ponents, more refinement has to take place alized requirements (saying, roughly, “model the
inside the components. organizational setting of the system with the
• The joint venture supports coordinativ- pyramid, structure-in-5, joint venture, arm’s-
ity in the sense that each partner actor in- length . . . pattern”).
teracts via the joint manager for strategic The evaluation results in contribution relation-
decisions. Partners indicate their interest, ships from the patterns to the non-functional re-
and the joint manager either returns them quirements, labeled “+”, “++”, “–”, “– –”. Design
the strategic information immediately or rationale is represented by claims drawn as dashed
mediates the request to some other part- clouds. They make it possible for domain char-
ners. However, since partners are usually acteristics (such as priorities) to be considered
heterogeneous, it could be a drawback to and properly reflected into the decision making
define a common interaction background. process, e.g., to provide reasons for selecting or
The central position and role of the joint rejecting possible solutions (+, –). Exclamation
manager is a means for resolving conflicts marks (! and !!) are used to mark priority require-
and preventing unpredictability. Through ments while a check-mark “√” indicates an ac-
its joint manager, the joint-venture pro- cepted requirements and a cross “ X ” labels a
poses a central communication controller. denied requirement.
It is less clear how the joint venture pat- Relationships types (AND, OR, ++, +, –,
tern addresses fallibility-tolerance, notably and – –) between NFRs are formalized to offer a
reliability. However, exceptions, supervi- tractable proof procedure. AND/OR relationships
sion, and monitoring can improve its over- corresponds to the classical AND/OR decompo-
all score with respect to these qualities. sition relationships: if requirement R0 is AND-
Manipulation of partners can be done eas- decomposed (respectively, OR-decomposed) into
ily to adapt the structure by registering new R1,R2,...,Rn then all (at least one) of the require-
ones to the joint manager. However, since ments must be satisfied for the requirement R0 to
partners can also exchange resources di- be satisfied. So for instance, in Figure 7, Coordi-
rectly with each other, existing dependen- nativity is AND decomposed into Distributivity,
cies should be updated as well. The joint Participability, and Commonality. Relationships
manager cannot be removed due to its cen- “+” and “– ” model respectively a situation where
tral position. a requirement contributes positively or negatively
Table 2 summarizes the strengths and weak-

nesses of the reviewed patterns. Table 2. Strengths and weaknesses of some patterns
To cope with non-functional requirements and
Structure-in-5 Joint-Venture
select the pattern for the organizational setting,
Coordinativity ++ +
we go through a means-ends analysis using the
Predictability + +
non functional requirements (NFRs) framework
Fallibility-Tolerance ++ +
(Chung et al. 2000). We refine the identified re-
Adaptability + +
quirements to sub-requirements that are more
1156
Figure 7. Partial evaluation for organizational patterns
towards the satisfaction of another one. For in- adopt from (Giorgini, Mylopoulos, Nicchiarelli &
stance, in Figure 7, Joint Venture contributes Sebastiani 2002) both qualitative and a numerical
positively to the satisfaction of Distributivity axiomatization for goal (requirements) modeling
and negatively to the Reliability. In addition, primitives and label propagation algorithms that
relationships “++” and “– –” model a situation are shown to be sound and complete with respect
where the satisfaction of a requirement implies to their respective axiomatization. In the following,
the satisfaction or denial of another goal. In Figure a brief description of the qualitative algorithm.
7, for instance, the satisfaction of Structure-in-5 To each requirement R, we associate two vari-
implies the satisfaction of requirements Reliability ables Sat(R), Den(R) ranging in {F, P, N} (full,
and Redundancy. partial, none) such that F > P > N, representing
The analysis for selecting an organizational the current evidence of satisfiability and deni-
setting that meets the requirements of the system ability of the requirement R. E.g., Sat(Ri) > = P
to build is based on propagation algorithms pre- states there is at least a partial evidence that Ai is
sented in (Giorgini, Mylopoulos, Nicchiarelli & satisfiable. Starting from assigning an initial set
Sebastiani 2002). Basically, the idea is to assign of input values for Sat(Ri), Den(Ri) to (a subset
a set of initial labels for some requirements of the of) the requirements in the graph, we propagate
graph, about their satisfiability and deniability, and the values through the propagation rules of Table
see how this assignment leads to the labels propa- 3. Propagation rules for AND (respectively OR)
gation for other requirements. In particular, we relationship are min-value function for satisfiabil-
Table 3. Propagation rules for satisfiability in the qualitative framework. A dual table is given for deni-
ability propagation.
+ - ++ --
G2→ G1 G2→ G1 G2→ G1 G2→ G1
Sat(G1) min{ Sat(G2), P } min{ Sat(G2), P } Sat(G2) N
Den(G1) N N Sat(G2)
1157
Figure 8. Schema of the label propagation al-

ity (max-value function) and max-value function
gorithm
(min-value function) for deniability. A dual table
is given for deniability propagation.
The schema of the algorithm is described in
Figure 8. Initial, Current and Old are arrays of
pairs Sat(Ri), Den(Ri), one for each Ri of the graph,
representing respectively the initial, current and
previous labeling status of the graph.
The array Current is first initialized to the
initial values Initial given in input by the user. At
each step, for every requirement Ri, Sat(Ri),
Den(Ri) is updated by propagating the values of
the previous step. This is done until a fixpoint is
reached, that is, no updating is mode possible
(Current == Old). The updating of Sat(Ri), Den(Ri)
works as follows. For each relation Reli incoming
in Gi, the satisfiability and deniability values satii sumption which distinguishes Tropos from other
and denii derived from the old values of the source methodologies is that actors and goals are used
requirements are computed by applying the rules as fundamental concepts for analysis and design
of Table 3. Then, it is returned the maximum during all phases of software development, not
value between those computed and the old values. just requirements analysis. That means that, in the
light of this paper, Tropos describes in terms of
the same concepts and patterns the organizational
A rEQUIrEMENts-DrIVEN environment within which a system will eventually
MEtHODOLOGy operate, as well as the system itself. Tropos spans
four phases of software development:
This research is conducted in the context of the
early requirements phase of Tropos (Giorgini et 1. Organizational modeling, concerned with the
al. 2004, Giorgini et al. 2005), a software devel- understanding of a problem by studying an
opment methodology for building multi-agent organizational setting; the output is an or-
systems which is founded on the concepts of ganizational model which includes relevant
actor and goal. actors, their goals and dependencies.
The Tropos methodology adopts ideas from 2. Requirements analysis, in which the system-
multi-agent systems technologies, mostly to to-be is described within its operational
define the detailed design and implementation environment, along with relevant functions
phase, and ideas from requirements engineering and qualities.
and organizational modeling, where agents/ac- 3. Architectural design, in which the system’s
tors and goals have been used heavily for early global architecture is defined in terms of
requirements analysis (Dardenne et al. 1993, Yu subsystems, interconnected through data,
1995). In particular, the Tropos project adopts Eric control and dependencies.
Yu’s i* model which offers actors (agents, roles, 4. Detailed design, in which behaviour of each
or positions), goals, and actor dependencies as architectural component is defined in further
primitive concepts for analyzing an application detail.
during organizational modeling. The key as-
1158
cONcLUsION rEFErENcEs
Modelers need to rely on patterns, styles, and idi- Anton, A. I. (1996). Goal-based requirements
oms, to build their models, whatever the purpose. analysis. In Proceedings of the 2nd Int. Conf.
We argue that, as with other phases of software On Requirements Analysis, ICRE’96’, Colorado
development, organizational modeling can be Spring, USA, pp. 136–144.
facilitated by the adoption of organizational pat- Bates, L. D. V. (2006), Advertising Agency. At
terns. This paper focuses on two such patterns and http://www.ldv.be.
studies them in detail, through examples, a formal-
ization using Formal Tropos, and an evaluation Chung, L. K., Nixon, B., Yu, E., & Mylopoulos, J.
with respect to desirable requirements. There have (2000), Non-Functional Requirements in Software
been many proposals for software patterns (e.g., Engineering. Kluwer Publishing.
(Kolp, Do & Faulkner 2005)) since the original
Dardenne, A., van Lamsweerde, A., & Fickas, S.
work on design patterns (Gamma, Helm, Johnson
(1993). Goal-directed requirements acquisition.
& Vlissides 1995). Some of this work focuses
Science of Computer Programming, 20(1–2),
on requirements patterns. For example, (Konrad
3–50. doi:10.1016/0167-6423(93)90021-G
& Cheng 2002) proposes a set of requirements
patterns for embedded software systems. These Do, T. T., Faulkner, S., & Kolp, M. (2003).
patterns are represented in UML and cover both Organizational multi-agent architectures for in-
structural and behavioral aspects of a requirements formation systems. In Proc. of the 5th Int. Conf.
specification. Along similar lines, (Fowler 1997) on Enterprise Information Systems, ICEIS’03’,
proposes some general patterns in UML. In both Angers, France, pp. 89–96.
cases, the focus is on requirements analysis, and
Doran, J. E., Franklin, S., Jennings, N. R.,
the modeling language used is UML. On a dif-
& Norman, T. J. (1997). On cooperation in
ferent path, (Gross & Yu 2002) proposes a sys-
multi-agent systems. The Knowledge Engi-
tematic approach for evaluating design patterns
neering Review, 12(3), 309–314. doi:10.1017/
with respect to non-functional requirements (e.g.,
S0269888997003111
security, performance, reliability). Our approach
differs from this work primarily in the fact that Dussauge, P., & Garrette, B. (1999). Coopera-
our proposal is founded on ideas from Organiza- tive Strategy: Competing Successfully Through
tion Theory and Strategic Alliances literature. We Strategic Alliances. Wiley and Sons.
have already described organizational patterns
Fowler, M. (1997), Analysis Patterns: Reusable
but to be used for designing multi-agent system
Object Models. Addison-Wesley.
architectures (Kolp et al. 2006) and e-business
systems (Kolp et al. 2004). Considering real world Fuxman, A., Liu, L., Mylopoulos, J., Roveri, M.,
organizations as a metaphor, systems involving & Traverso, P. (2004). Specifying and analyzing
many software actors, such as multi-agent sys- early requirements in tropos. Requirements En-
tems could benefit from the same organizational gineering, 9(2), 132–150. doi:10.1007/s00766-
models. In the present paper, we have focused 004-0191-7
on patterns for modeling organizational settings,
Gamma, E., Helm, R., Johnson, J., & Vlissides,
rather than software systems and emphasized the
J. (1995). Design Patterns: Elements of Reusable
need for organizational abstractions to better match
Object-Oriented Software. Addison-Wesley.
the operational environment of the system-to-be
during organizational modeling.
1159
Giorgini, P., Kolp, M., & Mylopoulos, J. (2002), Kolp, M., Do, T., & Faulkner, S. (2004). A social-
Multi-agent and software architecture: A com- driven design of e-business system. In Software
parative case study. In Proc. of the 3rd Interna- Engineering for Multi-Agent Systems III, Research
tional Workshop on Agent Software Engineering, Issues and Practical Applications. Edinburg, UK,
AOSE’02’, Bologna, Italy, pp. 101–112. pp. 70–84.
Giorgini, P., Kolp, M., Mylopoulos, J., & Castro, Kolp, M., Do, T., & Faulkner, S. (2005). Introspect-
J. (2005). A requirements-driven methodology for ing agent-oriented design patterns. In S. K. Chang,
agent-oriented software. In B. Henderson-Sellers (ed.), Handbook of Software Engineering and
& P. Giorgini, (Eds.), Agent Oriented Methodolo- Knowledge Engineering, 3, Recent Advances’,
gies. Idea Group Publishing, pp. 20–46. World Scientific, pp. 151–177.
Giorgini, P., Kolp, M., Mylopoulos, J., & Pistore, Kolp, M., Giorgini, P., & Mylopoulos, J. (2002a).
M. (2004). The tropos methodology. In M.-P. G. Information systems development through social
F. Bergenti & F. Zambonelli, (eds.), Methodolo- structures. In Proc. of the 14th Int. Conf. on Soft-
gies and Software Engineering for Agent Systems. ware Engineering and Knowledge Engineering,
Kluwer, pp. 89–105. SEKE’02’. Ishia, Italy, pp. 183–190.
Giorgini, P., Mylopoulos, J., Nicchiarelli, E., Kolp, M., Giorgini, P., & Mylopoulos, J. (2002b),
& Sebastiani, R. (2002). Reasoning with goal Organizational multi-agent architecture: A mobile
models. In Proceedings of the 21st International robot example. In Proc. of the 1st Int. Conf. on
Conference on Conceptual Modeling (ER 2002). Autonomous Agent and Multi Agent Systems,
Tampere, Finland, pp. 167–181. AAMAS’02’. Bologna, Italy, pp. 94–95.
GMT. (2006). Gmt consulting group. http://www. Kolp, M., Giorgini, P., & Mylopoulos, J. (2003).
gmtgroup.com/. Organizational patterns for early requirements
analysis. In Proc. of the 15th Int. Conf. on Ad-
Gomes-Casseres, B. (1996). The alliance revolu-
vanced Information Systems, CAiSE’03’. Velden,
tion: the new shape of business rivalry. Harvard
Austria, pp. 617–632.
University Press.
Kolp, M., Giorgini, P., & Mylopoulos, J. (2006).
Gross, D., & Yu, E. (2002). From non-functional
Multi-agent architectures as organizational
requirements to design through patterns. Require-
structures. Autonomous Agents and Multi-Agent
ments Engineering, 6(1), 18–36. doi:10.1007/
Systems, 13(1), 3–25. doi:10.1007/s10458-006-
s007660170013
5717-6
Horling, B., Lesser, V., Vincent, R., Bazzan, A.,
Konrad, S., & Cheng, B. (2002). Requirements
& Xuan, P. (1999). Diagnosis as an integral part
patterns for embedded systems. In Proc. of the
of multi-agent adaptability. Technical Report
10th IEEE Joint International Requirements En-
UM-CS-1999-003, University of Massachusetts.
gineering Conference, RE’02’. Essen, Germany,
Jennings, N. R. (1996). Coordination techniques pp. 127–136.
for distributed artificial intelligence. In G. M. P.
Mintzberg, H. (1992). Structure in fives: Designing
O’Hare & N. R. Jennings, (eds.), Foundations
effective organizations. Prentice-Hall.
of Distributed Artificial Intelligence. Wiley, pp.
187–210.
1160
Morabito, J., Sack, I., & Bhate, A. (1999). Orga- Weiss, G. (Ed.). (1997). Learning in DAI Systems.
nization modeling: Innovative architectures for Springer Verlag.
the 21st century. Prentice Hall.
Woods, S. G., & Barbacci, M. (1999). Architec-
Scott, W. R. (1998). Organizations: Rational, tural evaluation of collaborative agent-based
natural, and open systems. Prentice Hall. systems. Technical Report SEI-99-TR-025, SEI,
Carnegie Mellon University, Pittsburgh, USA.
Segil, L. (1996). Intelligent business alliances:
How to profit using today’s most important stra- Wooldridge, M., & Jennings, N. R. (1995). Intel-
tegic tool. Times Business. ligent agents: Theory and practice. The Knowledge
Engineering Review, 2(10).
Shaw, M., & Garlan, D. (1996). Software Archi-
tecture: Perspectives on an Emerging Discipline, Yoshino, M. Y., & Srinivasa Rangan, U. (1995).
Prentice Hall. Strategic alliances: An entrepreneurial approach
to globalization. Harvard Business School Press.
Shehory, O. (1998). Architectural properties of
multi-agent systems. Technical Report CMU-RI- Yu, E. (1995). Modelling Strategic Relationships
TR-98-28, Carnegie Mellon University. for Process Reengineering. PhD thesis, University
of Toronto, Department of Computer Science.
Wautelet, Y., Kolp, M., & Achbany, Y. (2006).
S-tropos: An iterative spem-centric software
project management process. Technical Report
IAG Working paper 06/01, IAGISYS Information
Systems Research Unit, Catholic University of
Louvain, Belgium. http://www.iag.ucl.ac.be/wp/.
1161
1162
Chapter 4.18
Mobile Strategy for
E-Business Solution
Anthony S. Atkins
Staffordshire University, UK
A. K. Hairul Nizam Pengiran Haji Ali

Staffordshire University, UK
AbstrAct utilize potential mobile applications. The use of

a mobile strategic framework will help provide
It is becoming evident that mobile technology can the insights to improving companies in their com-
enhance a current e-business system to provide mercial operations and examples of these mobile
competitive advantage in business activities. These solutions are outlined in relation to commercial
enhancements in mobile device applications such applications which have been implemented in
as in mobile hotel check-in system, m-payment hospitals, retail Supply Chain Management
system for parking tickets, and mobile donor trans- (SCM), and in Customer Relationship Manage-
plant system are evolving with usage of wireless ment (CRM). These types of systems are known
technology such as Wi-Fi, Bluetooth, and WiMax to improve quality of service and provide competi-
(Worldwide Interoperability for Microwave Ac- tive advantage. A mobile framework is presented
cess). Other examples include wearable mobile to introduce the application of user mobility to
technologies used in military observation tactics mobile usage as an extension of existing Intranet,
and civilian clothing accessories for entertainment Extranet, and Internet e-business application. This
purposes. The lack of current mobile strategies, Mobile Business Application Framework could
can cause some businesses to over spend or under assist practitioners in identifying the financial and
competitive aspects in relation to mobile technol-
DOI: 10.4018/978-1-60566-156-8.ch010 ogy applications into their business infrastructure.
Mobile Strategy for E-Business Solution
INtrODUctION decisions as they track available donors that match

with the patient’s profile. An indirect application
Mobility has become a key factor in Information of using these mobile phones in this situation can
Technology (IT) strategy (Savvas, 2007). The also be used to assist emergency crews in apply-
literature indicates that the use of mobile devices ing first aid. In the event of an accident, by taking
can assist the communication networks in busi- images of any unusual wounds and forwarding
ness activities, such as Supply Chain Management them to a specialists, a remote doctor can advise
(SCM), parcel tracking and Customer Relation the ambulance crew to apply appropriate first
Management (CRM) (UPS, 2005). Mobile Com- aid to the victim (British Red Cross, 2004). The
merce (m-Commerce) operates where mobile agility and mobility of using these mobile devices
devices facilitate business operations that enhance can prove advantageous to commercial operations
and improve commercial activities (Varshney and (Harrington, 2006), especially e-commerce appli-
Vetter, 2002). In commercial organisations, mobile cations. The increased demand of mobile devices
devices such as Personal Digital Assistant (PDA) is due to the improvements of wireless technology
phones can help business users to organise their that has allowed managers to select from a wide
daily activities such as taking notes, arranging variety of mobile technology to apply to their
appointments, storing contacts phones numbers, business. The use of m-commerce can provide
receiving emails and surfing the internet, from a both competitive advantage and modern image to
single mobile device (Jervanpaa, 2006). Stockbro- the business. However, there are risks associated
kers can receive critical information on their PDAs with it, such as the cost of development, security
such as changes in financial stocks and shares at issues relating to viruses and privacy policies. It
anytime and also access financial documents and is imperative that the business should analyse the
make amendments with the use of built in word type of mobile devices which can best enhance
processing applications. their business activity (Varshney and Vetter, 2002).
Short Messaging Services (SMS) is another The use of strategic IT frameworks could be used
mobile solution making mobile payments or for assessing mobile business application to aid
m-Payments (Serbedzija et al, 2005). In Croatia, business practitioners in decision making.
parking in the city can be difficult as parking ar- A combination in the use of mobile applications
eas can be on either side of the road, or one large and e-business can offer competitive advantage
central lot located far from the drivers destination, by creating new platforms to reach global markets
the hassle of walking to ‘top up’ a parking meter (Farhoomand, 2003). The e-business services
and then walking back to the destination can be allow business companies to reduce costs and in-
tedious. The driver receives an SMS acknowledge- crease revenue from distribution. Increased online
ment text to indicate the expiration time of the sales and the use of mobile devices can enhance
parking meter and can then choose to return to the mobile business activities. Mobile applications
vehicle or extend the parking meter by replying are currently being used in warehousing, Small
to the SMS text (Serbedzija, 2005). and Medium Enterprise (SME) and Customer
Addenbrooke’s Hospital in Cambridge, highly Relationship Management (CRM) systems that
regarded for the success of kidney operations in allow tracking of parcels etc. These types of mo-
the last few years is utilising Blackberry’s mobile bile devices suit the business needs and activities
phone application to wirelessly connect to the of the companies concerned. This principle of
hospital’s centralised database to keep track of distribution and transaction can be adapted for
available organ donors. The use of mobile devices e-government or e-society applications. In these
allows surgeons and doctors to make on the spot cases information can be distributed electronically
1163
similar to the e-society i2010, which the European and Value Chain (Porter, 2001), strategic tools for
Commission is developing for a European Infor- mobile applications appear to be limited (Varshney
mation Society that promotes growth and jobs in and Vetter, 2002; Chen and Nath, 2004), with little
the EU (Hines, 2007). John Hopkins Hospitals has financial appraisal. A framework to assess strategic
implemented mobile applications to assists in their applications of mobile technology and pertinent
medical activities, such as e-prescriptions. This infrastructure is outlined to assist business manag-
implementation saves the hospitals $1,000 per ers in making financial and technological decisions
day by providing the pharmaceutical information in mobile business environments.
and the medicine that the patient needs (Brian,
2006). The use of data retrieval through a mobile
device can be similarly used in warehousing; EXtENtION tO M-cOMMErcE
Wal-Mart stocks can be identified using Radio
Frequency Identification (RFID) tags read from a Improved computing infrastructure has provided
special mobile RFID reader (Sliwa, 2006). Nissan e-business with an advantage, because it allows
automobiles have a similar approach with CRM global accessibility from any access points either
activities that allow the salesperson to answer wired or wirelessly. The e-business serves its
customer queries on the spot (Greengard, 2006). purpose by providing the capability to achieve
There are two types of mobile data retrieval revenue from low costs distribution and increased
technologies, which are classified as web-enabled sales from the global market. Customers are able
and standalone. The Web-enabled architecture to access these websites not only from their homes,
involves mobile devices to send and retrieve but also in cyber cafes, public hotspots where the
information from a centralised database. With users can access the internet using wireless LANs
the standalone database, the data are distributed and even on a mobile phone. The development
and stored on the mobile devices. The database of 3G mobile phones and PDAs now have the
can share the data among mobile devices, but ability to capture pictures and send them through
the original data remain in the mobile device for MMS (Multimedia Messaging Service). The
example using PDAs for collecting questionnaire 3G phones also allow video calls, video clips,
data. The type of wireless technology used, such and surfing of the internet via GPRS (General
as Wireless Local Area Network (WLAN), Wifi, Packet Radio Service) or CDMA (Code-Division
Bluetooth and WiMax, will also affect the type of Multiple Access), rather than through the classic
mobile application that a company can apply in WAP (Wireless Access Protocol). Users are now
their business activities. The most popular ones are able to view web pages via their mobile devices
Bluetooth and Wi-fi connections which are simple depending whether their phones support Hyper-
and less expensive to use. Japanese Wagamama Text Mark-Up Language (HTML) or WAP capa-
restaurant chains in the UK (Terry, 2006) apply bilities. The increasing amount of mobile phone
this type of mobile devices. Wi-fi applications are users within the next decade (Roto, 2007), would
more technically diverse as they can support both enable businesses to sell products electronically
Personal Digital Assistants (PDA) and computer over the mobile platform, similar to earlier e-
laptops. The evolution of mobile technology has business concepts. One example of this is Ring
provided new opportunities by extending mobile tones, which has generated over $600 million
infrastructure into the business, which enables the in sales via consumers mobile phones (Mobile
business to gain competitive advantage. Although Youth, 2006). An example of a mobile applica-
there are many strategic tools and techniques tion that assists in improving business activities
available such as Porter’s competitive analysis is the PDA insurance quota service (Serbedzija,
1164
2005). This application has been implemented ment of 3G technologies and the introduction of
in Germany where young adults find the idea WiMax can involve around 47,000 base stations
of using a PDA device to electronically search in the UK of which two thirds of these installa-
insurance quotations for their automobiles very tions are on existing buildings. Another concern
useful, particularly as they are able to electroni- in the usage of mobile phones by the public is the
cally enter their details via the mobile device. As location awareness that traces the user’s mobile
a result, insurance companies in Germany have location anywhere in the world by satellite or cel-
used these PDA’s as their after sales activities to lular location awareness. This technology is similar
improve their Customer Relationship Manage- to the use of locating GPS (Global Positioning
ment (CRM) performance with their customers. System) which can be exploited by the military
Unfortunately not all mobile technology can be or police authorities with mobile users having no
applied to every business activities. Therefore, knowledge of being monitored.
strategic tools and techniques are needed to assist
companies in applying mobile applications into
their business activities (Atkins, et al., 2006). MObILE APPLIcAtIONs
The literature indicates that most mobile business FrAMEWOrK
applications are an extension of e-business rather
than a separate technology (Sliwa, 2004; UPS, Strategic planning is often used in starting a new
2007; Hadfield, 2006; www.e-health-insider.com, product or in improving business activities in a
2004; Cyber-Lab, 2005). competitive environment. Managers often use
strategic tools for forward planning, or to propose
a new solution to increase business profits. The
HEALtH, sAFEty, AND planning usually involves identifying weaknesses,
PrIVAcy IssUEs or opportunities in the business environment. The
use of strategic management tools can provide
This advancement of mobile devices has also helpful insights in competing against competi-
brought about health and safety concerns of the tors (Ghalayini and Noble, 1996). Some business
intense usage and associated radiation could lead strategic tools involve applying new technology or
to cancer problems in the future (www.who.int, merging with other business services to cut costs
2006). The radio frequency that mobile phones (Shaffer and Srinivasan, 2005). Other strategies
produce when they are on standby is very small, suggest concentrating on the source of income,
but during call times the type of wavelength it can such as ensuring customer satisfaction, or im-
emit is around 2 GHz during active transmission. proving relations with the suppliers and buyers.
The constant exposure to these mobile phones Some strategic ideas are ineffective because of the
can cause ‘heating’ or ‘thermal’ effect to the ear inability to predict environmental changes, coordi-
when the user makes the call. This thermal effect nation, and communication with the suppliers and
has made the general public perceive that mobile the customers. Identifying a strategic tool that can
phones could cause cancer (www.who.int, 2006). be used in specific business tends to be difficult
The effect on the brain when exposed to using this because of the array of tools available. Business
mobile technology is obviously a concern by the tools such as the SWOT (Strength, Weaknesses,
general public. The World Health Organisation Opportunities and Threats) and PEST (Political,
(WHO) has indicated general health awareness Economical, Socio-Cultural and Technological)
and the possible dangers of mobile devices and of are commonly used for marketing and in IT for
excessive use (www.who.int, 2006). The develop- business strategy. Strategic tools and techniques,
1165
such as the Strategic Grid (McFarlan et al, 1984), financial scorecard has four types of perspective
Porter’s Five Forces (Porter, 2001), Value Chain as shown in Section A: the Financial, Customer,
(Porter, 2001), IT Business Alignment tool (Hen- Learning, and Growth, and Internal Business
derson, 1993), can improve business activities Process respectively. Each of these perspectives
and assist in providing competitive advantage. evaluates the strength and weaknesses of the
The proposed Mobile Business Application business by assessing the previous records of its
Framework, in Figure 1, is an aid to business activities. An example of the use of the scorecard
management in decision making within the three is when a company is considering implementing
regions of user mobility. The framework consists a mobile technology to improve the inventory
of two sections, firstly the Financial Scorecard, system using Radio Frequency Identification
as shown in Section A of Figure 1. The Financial (RFID) technology. The company could evaluate
Scorecard is used to assess the business process; the expected benefits by using the process cycle
activities and financial standing, similar to a cost of the Financial Scorecard.
benefit analysis, and can identify the weaknesses
and opportunity of the company (Lorsh, 1993). • In Part 1 of Figure 1, the process cycle of
The Financial Scorecard helps the business to mea- the Financial Scorecard starts with the
sure performance and resolve customer relations Financial Perspective. Management assess
and strategy issues. Figure 1 illustrates that the Financial Perspective, looking at previous
Figure 1. Mobile business application framework
1166
records or accounts of the business activi- bands of mobile usage. These bands are colour
ties, to set objectives, measures and opera- coded to signify ‘critical data transfer,’ which
tional goals. At the end of the business pro- indicates the importance of data transmission
cess, the management can view the record to the business applications. The three different
and comment on the Management regions differentiate the types of mobile device
Initiatives. This feedback checks if the and determine how the software application can
business has reached their intended goals be implemented in relation to business operations.
or requires additional procedures to en- The type of mobile device, or application, used
hance their business. in the business is shown by the vertical axis of
• These Scorecard procedures are repeated the diagram. The Intranet and Extranet region of
for the Customer, Learning and Growth, the framework, illustrated in Figure 1, can apply
and the Internal Business Perspective. a fully integrated mobile device or customised
• The Internal Business Process uses the application using normal mobile phones, such as
Mobile RFID Technology to enhance Wal-Mart Wireless Inventory Checking (Sliwa,
stock location processes in the warehouse. 2004). Also users within the Internet region can
The Management Initiatives column indi- use normal mobile devices but they will link or
cated in Part 2 of Figure 1 will be used to visit a web-enabled e-commerce database, to
record the success factor at the end of the connect to an application, for example an SMS
implementation. vending machine (SAFECOM, 2007).
• As the cycle repeats, the Management
Initiatives for the Financial Perspective are
assessed and recorded, to monitor the ef- MObILE bUsINEss APPLIcAtION
fects of implementing a mobile technology FrAMEWOrK ON WAL-MArt
in the business. This iterative process is re-
peated for the Customer, and the Learning Wal-Mart stores provide a range of goods including
and Growth Perspective. grocery goods, clothing, to electronics and frozen
• After completing the cycle (anti-clockwise goods all in affordable prices (Sliwa, 2004). The
in Part 1 of Figure 1) the Internal Business current IT infrastructure of Wal-Mart is based on
Process for implementing the mobile tech- the provision of the efficient ‘upstream’ supply
nology is assessed against the target goals chain system. Their main intent is to support its
as shown in Part 3. strategy by providing support for the most effi-
cient stock control, warehousing and distribution
In Figure 1, Section B illustrates the strategic systems in the business (Cyber-Lab, 2005). The
framework and identifies the type of mobile de- application of RFID technology has significantly
vice that would be suitable to the business. The improved their upstream supply chain. They are
framework indicated in Part 1 illustrates a radial able to keep costs to a minimum by keeping track
wave, or a ripple, which depicts the mobility range of Point-Of-Sales (POS) system and delivery
for mobile applications involving the Internet, Ex- times of the suppliers. Radio Frequency Identi-
tranet, and Intranet (Mobile Application Region). fiers (RFID) or identity tags, are small devices
These three regions are shown on the horizontal that look like label tags. They function by emit-
axis of the Framework. The implementation of ting small radio signals, which can be read by
the mobile RFID technology is contained within an RFID reader that can display a considerable
the warehouse, consequently this is an Intranet amount of data tagged on them. Such devices are
region. In each region, there are also circular commonly used for locating and identifying stock
1167
crates and inventories in warehouses. Wal-Mart rELAtION WItH MObILE

is well known for applying such technology in bUsINEss FrAMEWOrK
their business activities, particularly since they
receive daily crates of perishable goods. Active In terms of comparing the proposed framework
and Passive tags are the two main types of RFID against the current inventory system in Wal-Mart,
technologies that both function different according its implementation follows correspondingly to the
to supply chain management in the warehouse. framework and fits into the Intranet Region. This
Active RFID tags have a built in battery which indicates that the use of the mobile devices in the
powers it continuously to emit constant Radio warehouse is to track and identify the crates that
Frequency (RF) signals for the reader to detect. arrive from deliveries and update accordingly
While Passive RFID relies on radio frequency into the inventory stock. This business strategy
energy from the reader in order to power that tag analyses the business operations, identifies the
temporarily and then responds to the reader (Savi, weaknesses and indicates access for improving
2007; Sensitech, 2007). In a warehouse where business activities. Figure 1 Section C outlines
crates are not moved constantly, then the use of the improvements needed to commence in the
Passive RFID will suffice. Applying an Active new procurement year. In the Internal Business
RFID in a low activity warehouse will under- Process Perspective, Figure 1 Section C highlights
utilise the RFID technology. Therefore they are the need for the use of mobile application systems.
best suited when there is high activity of moving In the next stage, the framework will illustrate to
crates and materials. The constant process is very the business managers what mobile devices can
dynamic and unconstrained, thus having the ability be applied in this situation. Figure 2 illustrates
to detect multi-tag Active RFID is the best option the type of mobile devices that can be used in the
for this situation (Savi, 2007). Integrated Mobile Device and Application for this
region of the business area. Mobile devices will
include the use of RFID tag, mobile RFID readers
and a database system that tracks the tagged items.
Figure 2. Integrated mobile/device and application
1168
• The use of these RFID tags is to identify ogy within their business for order taking. WiMax,
crates incoming and outgoing crates of the can be normally used to overcome geographical
Wal-mart warehouses. boundaries or avoid the problem of implementing
• The overall goal of the implementation is cables. The proposed mobile strategic framework
to improve the business activity by reduc- has been applied to several mobile commercial
ing costs and time taken to locate specific applicants is outlined (Pg Haji Ali, 2007; Pg Haji
supplier and inventories. Ali and Atkins, 2007). It demonstrates the use of a
Financial Scorecard to evaluate how the business
At the “Integrated Mobile/Device and Applica- cost effectiveness and departmental goals within
tion” stage shown in Figure 2 of the framework the company can be assessed, allowing managers
will determine the type of critical data transfer. The to identify and align strategic initiatives to the
data transferred between the centralised database business (Lorsch, 1993). In the case of mobile
and mobile devices will include type of stock, technology, the strategic framework assesses the
date, content and location of the crates. suitability of mobile devices for the business and
The following includes additional features considers the type of security and applications
into the system:- appropriate for a particular mobile device. The
mobile strategic framework is designed to assist
• Contingency arrangements include a bar- mangers to make informed decisions and increase
code or serial no. which the user can key competitive advantage in mobile infrastructure
in manually in the event that the RFID tag enhancements.
is damaged.
• In this instance, the original data transfer
is likely to be of the ‘High’ classification, rEFErENcEs
because a delayed delivery may result in a
severe chain reaction (i.e. compound ship- Alexander, L. (2007). Are Wireless Networks
ment delays) due to their use of a just in Safe?’ Retrieved 15 February 2007:-http://blogs.
time system. cio.com/node/656
Atkins, A. S., & Pengiran Haji Ali, A. K. Hairul
Nizam, & Shah, H. (October, 2006) Extending
cONcLUsION e-business applications using mobile technol-
ogy. In proceedings of International Confer-
The strategic framework tool presented here is ence on Mobile Technology, Applications and
designed to help managers identify where mobile Systems, IEE Mobility Conference. Bangkok
technology can enhance their strategic business Thailand. Retrieved from http://doi.acm.
operations. The business strategy should align org/10.1145/1292331.1292381
with IT strategy, organisational infrastructure and
information system infrastructure, in order to pro- Brian, K. (2006). Lowering Health Care Costs
vide competitive advantage. Wi-fi and Bluetooth Out-of-the-Box. Retrieved 27 September 2006:-
technologies are popular in businesses because of http://wireless.sys-con.com/read/40975.htm
the effective wireless range. The short wireless
British Red Cross. (2004). First aid at your finger-
range allows a more secure system, as the data
tips. Retrieved 6 July 2007:-http://www.redcross.
transmission signal may be contained within the
org.uk/news.asp?id=41091
business area. Japanese restaurant Wagamama
(Terry, 2006) has implemented Bluetooth technol-
1169
Chen, L., & Nath, R. (2004). A Framework for Jarvenpaa, S. L. (2006). Internet Goes Mobile:
Mobile Business Application. Int. J. Mobile How Will Wireless Computing Affect Your Firm’s
Communications, 2(4), 368–381. doi:10.1504/ Internet Strategy? Retrieved at May 2006:-http://
IJMC.2004.005857 web.hhs.se/cic/about/blkcoffee/Internet_Mo-
bile_6-19_sj.pdf
Cyber-Lab. (2005). A Supermarket uses RFID
to Control Stock Inventory. Retrieved 12 April Lorsch, J. W. (1993). Smelling Smoke: Why
2006:-http://www2.cpttm.org.mo/cyberlab/rfid/ Boards of Directors Need the Balanced Scorecard.
wal-mart.html.en Balanced Scorecard Report, September-October,
pp. 9.
E-Health Insider. (2005). Addebrooke’s installs
SMS Patient reminder system. Retrieved at 1 McFarlan, F. W., James, L. M., & Philip, P. (1983).
March 2006:-http://www.e-health-insider.com/ The Information Archipelago – Plotting a Course.
news/item.cfm?ID=1236 Harvard Business Review, 61, 145–155.
Farhoomand, A., & Ng, S. P. (2003). Creating Sus- Mobile Youth. (2005). Retrieved 3 April 2006:-
tainable Competitive Advantage Through Inter- http://www.mobileyouth.org/my_item/ringtone_
networked Communities. Communications of the sales_build_600_mn_2005_revenues
ACM, 46(9), 83–88. doi:10.1145/903893.903898
Pg Hj Ali, A. H. N. (2007). Mobile Business
Ghalayini, A. M., & Noble, J. S. (1996). The Application. Masters by Research, Computing
Changing Basis of Performance Measure- Science, Faculty of Computing, Engineering
ment. International Journal of Operations and Technology, Staffordshire University, United
& Production Management, 16(8), 63–80. Kingdom – unpublished thesis.
doi:10.1108/01443579610125787
Pg Hj Ali, A. H. N., & Atkins, A. S. (2007, June). A
Greengard, S. (2001). Customer Services Goes Strategic Business Tool For Mobile Infrastructure.
Wireless. Business Finance, 35. Wireless Ubiquitous Computing, International
Conference on Enterprise Information Systems,
Hadfield, W. (2006). Police get Streetwyse to
Funchal, Portugal. (pp.23-32).
mobile ID checks. Computer Weekly, 4, 6.
Porter, M. E. (2001, March). Strategy and the
Haines, L. Digital divide is self-repairing, says
Internet. Harvard Business Review, 63–78.
UK gov (2005). Retrieved 5 February 2007:-http://
www.theregister.co.uk/2005/09/06/digital-divide/ Roto, V. (2007). Browsing on Mobile Phones.
Retrieved at 6 June 2007:-http://www.research.
Harrington, (2006). Standard and Guidelines
att.com/~rjana/WF12_Paper1.pdf
WHO. Retrieved 15 February 2007:-http://http://
www.who.int/peh-emf/standards/en/ Savi Technologies. (2007). Savi’s RFID Solu-
tions. Retrieved 6 June 2007:-http://www.savi.
Henderson, J. C., & Venkatraman, N. (1993).
com/index.shtml
Strategic Alignment: Leveraging Information
Technology for Transforming Organisations. IBM Savvas, A. (2007, February). Mobility is increas-
Systems Journal, 32(1), 4–16. ingly key to strategy. Computer Weekly, 13, 12.
1170
Serbedzija, N., Fabiunke, M., Schön, F., & Beyer, KEy tErMs AND DEFINItIONs
G. (2006). Multi-Client Car Insurance. Proceed-
ings of the IADIS International Conference WWW/ Bluetooth Technology: Short-range wireless
Internet (ICWI) 2005, 2, 30-34. 19-22 October, radio technology for connecting small devices,
Lisbon, Portugal, ISBN: 972-8924-02-X such as wireless PDAs and laptops, to each other
and connect them in a wireless network.
Shaffer, D., & Srinivasan, M. (2005). Agile IT E-Business: Electronic Business is any busi-
through SOA Requires New Technologies: How ness process that relies on an automated informa-
Not to Fall Into the Trap of Applying Archaic tion system.
Integration Technologies! Retrieved at 6 June M-Payment: Mobile Payment is the process
2007:-http://www.alignjournal.com/index. for payment of goods or services with a mobile
cfm?section=article&aid=310 device such as a mobile phone, Personal Digital
Sliwa, C. (2004). Wal-Mart CEO prom- Assistant (PDA), or other wireless devices.
ises ‘tough love’ approach to RFID use. Re- Mobile Business Application: Commercial
trieved at 5 April 2006:-http://www.computer- usage of mobile electronic transaction in busi-
world.com/mobiletopics/mobile/technology/ ness operations.
story/0,10801,89011,00.html Mobile Technology: Use of mobile telephony,
mobile computing, and miscellaneous portable
Terry, L. (2006). Wireless Business and Technol- electronic devices, systems, and networks.
ogy. Serving up wireless. Retrieved 10 January Strategic Framework: A framework to pro-
2006 from:-http://wbt.sys-con.com/read/40916. vide a focused solution for assisting in business
htm decision-making.
UPS. (2005). Embracing Technology. Retrieved Wi-Fi: Wireless Fidelity commonly used to
19 September 2005 from:-http://www.ups.com/ broaden wireless interface of mobile computing
content/corp/about/history/1999.html devices, such as laptops in Local Area Network
(LAN).
Varshney, U., & Vetter, R. (2002). Mobile Com-
merce: Framework, Application and Networking
Support. Mobile Networks and Applications, 7,
185–198. doi:10.1023/A:1014570512129
1171
1172
Chapter 4.19
Application of Software
Metrics in ERP Projects
S. Parthasarathy
Thiagarajar College of Engineering, India
ABSTRACT collected data from an ERP project supports the

two hypotheses, leading to the conclusion that the
Business information system is an area of the great- software metrics are momentous in ERP projects.
est significance in any business enterprise today.
Enterprise Resource Planning (ERP) projects are
a growing segment of this vital area. Software en- INTRODUCTION
gineering metrics are units of measurement used to
characterize the software engineering products and Software process improvement often receives little
processes. The research about the software process orderly attention. If it is important enough to do,
has acquired great importance in the last few years however, someone must be assigned the responsi-
due to the growing interest of software companies bility and given the resources to make it happen.
in the improvement of their quality. Enterprise Re- Until this is done, it will remain a nice thing to do
source Planning (ERP) projects are very complex someday, but never today. Software engineering
products, and this fact is directly linked to their process is the total set of software engineering
development and maintenance. One of the major activities needed to transform a user’s requirement
reasons found in the literature for the failure of into software (Humphrey 2005). In other words,
ERP projects is the poor management of software software process is a set of software engineering
processes. In this chapter, the authors propose a activities necessary to develop and maintain soft-
Software Metrics Plan (SMP) containing different ware products. The reason for defining the software
software metrics to manage software processes dur- process is to improve the way the work is done. By
ing ERP implementation. Two hypotheses have been thinking about the process in an orderly way, it is
formulated and tested using statistical techniques possible to anticipate problems and to devise ways
to validate the SMP. The statistical analysis of the to either prevent or to resolve them.
The software processes that are of great concern
DOI: 10.4018/978-1-61520-625-4.ch004 during the ERP implementation are requirements
Application of Software Metrics in ERP Projects
instability, scheduling and software mainte- maintenance are not uncommon. A key issue
nance (Parthasarathy and Anbazhagan, 2006). in ERP implementation is how to find a match
Here, we use software metrics to manage and between the ERP system and an organization’s
improve these software processes during the business processes by appropriately customizing
ERP implementation. It explains specifically both the system and the organization (Arthur
how the software processes can be quantified, 1997). This is badly affected due to the instability
plotted, and analyzed so that the performance in the requirements proposed by the customer and
of ERP software development activities can be the poor capability of the ERP vendor.
predicted, controlled, and guided to achieve both The phase “Gap Analysis” in ERP implementa-
business and technical goals. As mentioned by tion is a step of negotiation between the company’s
Parthasarathy & Anbazhagan (2006), though requirements and the functions an ERP package
there are a handful of software processes for possesses (Carmel et al 1998). Poor requirements
ERP projects, the processes such as requirements specification and its instability badly affects the
stability, schedule slippage and monitoring the gap analysis phase of an ERP project which in
software maintenance tasks are considered more turn leads to schedule slippage and bubbles of
important and account for the performance en- problems during the maintenance phase of the
hancement of ERP projects. Hence, only these software project. The requirements instability
software processes are dealt with using software is probably the most important single software
metrics in the proposed Software Metrics Plan process issue for many organizations. The failure
(SMP) developed in this chapter. of many software projects can be directly linked
Many project managers acknowledge that to the requirements instability (Davenport 1998).
measurement helps them understand and guide Customization, the biggest technology headache
the development of their projects (Fenton et al is considered as the critical success factor for ERP
2003). They want to select particular metrics, implementation (Parthasarathy and Anbazhagan,
perhaps as part of an overall development plan, 2007).
but they do not know how to begin. The answers Even for those companies that have success-
to a manager’s metrics questions should appear fully implemented large-scale information systems
in the project’s metrics plan, so that managers projects in the past, ERP implementation still
and developers know what to collect, when to presents a challenge, because it is not simply a
collect it and how the data relate to manage- large-scale software deployment exercise. Also
ment decisions. The plan enables managers to as ERP implementation is often accompanied by
establish a flexible and comprehensive metrics large-scale organizational changes, agile software
program as part of a larger process or product processes could not create much impact on the
improvement program. The research about the ERP projects (Glass 1998). One of the major rea-
software process has acquired great importance sons found in the literature for the failure of ERP
in the last few years due to the growing interest projects is the poor software process management.
of software companies in the improvement of Hence a Software Metrics Plan (SMP) has been
their quality. ERP projects are very complex developed to deal with the software processes
products, and this fact is directly linked to their discussed in the literature review and found to be
development and maintenance. more important for successful ERP implementa-
The total quality management (TQM) notion tion. The software processes dealt with in the
of prevention rather than correction can be applied SMP for effective software process management
successfully in software engineering. The project during the ERP implementation are:
schedule slippage and tracking problems during
1173
i. )>> Requirements Stability Index (RSI) pieces, size and effort. So the plan will explain
ii. )>> Schedule Slippage (SS) how size and effort are defined and then how
iii. )>> Arrival Rate of Problems (ARP) they are combined to compute productivity. At
iv. )>> Closure Rate of Problems (CRP) the same time, the plan must spell out “where”
v. )>> Age of Open Problems (AOP) and “when” during the process the measurements
vi. )>> Age of Closed Problems (ACP) will be made. Some measurements are taken once,
while others are made repeatedly and tracked
The SMP contains a set of software metrics over time. The time and frequency of collection
to manage these software processes during ERP are related to the goals and needs of the project,
implementation. Two hypotheses have been for- and those relationships should be made explicit
mulated and tested using statistical techniques in the plan.
to validate the SMP. The statistical analysis of “How” and “Who” address the identification
the collected data from an ERP project supports of tools, techniques, and staff available for metrics
these two hypotheses, leading to the conclusion collection and analysis. It is important that the plan
that the software metrics are momentous in ERP discusses not only what measures are needed but
projects. The results give fruitful information to also what tools are to be used for data capture and
the ERP team to identify the influence of one storage, and who is responsible for those activities.
process on another. Often, metrics plans ignore these crucial issues,
and project members assume that someone else
is taking care of the metrics business; the result is
SOFTWARE METRICS PLAN (SMP) that everyone agrees that the metrics are important,
but no one is actually assigned to get the data.
A metrics plan is much like a newspaper article: it Likewise, responsibility for analysis often slips
must describe the who, what, where, when, how though the project cracks on and data pile up but
and why of the metrics. With answers to all of are never used in decision making.
these questions, whoever reads the plan knows The plan must state clearly what types of
exactly why metrics are being collected, how analysis will be created with the data, who will
they are used, and how metrics fit into the larger do the analysis, how the results will be conveyed
picture of software development of maintenance. to the decision makers, and how they will support
The plan usually begins with the “why”. It is in decisions. Thus, the metrics plan prints a com-
this introductory section that the plan lays down prehensive picture of the measurement process,
the goals or objectives of the project, describing from the initial definition of need to analysis and
what questions need to be answered by project application of the results. Software processes have
members and software process management. For an important influence on the quality of the final
example, if reliability is a major concern to the software product, and for this reason companies
developers, then the plan discusses how reliability are becoming more and more concerned with
will be defined and what reporting requirements software process improvement.
are imposed on the project; later sections of the Successful management of the software pro-
plan can then discuss how reliability will be cess is necessary in order to satisfy the final qual-
measured and tracked. ity, cost and time of the marketing of the software
Next, the plan addresses “what” will be mea- products (Kitchenham et al 2002). To improve the
sured. In many cases, the measures are grouped software processes (Pressman 2006), a great variety
or related in some way. For instance, productiv- of initiatives have arisen like the Capability Maturity
ity may be measured in terms of two component Model (CMM), etc. All these initiatives focus on
1174
Table 1. Metric-01 in the software metrics plan (SMP)
Metric Schedule Slippage

Formulae {(Actual number of days – Estimated number of days) / Estimated number of days} * 100
Actual number of days is the difference between the date when a particular activity (phase) or project got
Measure completed and the date when a particular activity/phase or project started. Estimated number of days is
(Indirect Metric) the difference between the planned start date of a particular activity/phase or project based on the latest
plan and the planned end date of a particular activity/phase or project based on the latest plan.
Actual End Date: Project Schedule
Source of Data Expected Start Date: Project Schedule
Expected End Date: Project Schedule
Collection Responsibility Project Manager
Frequency of Weekly Progress Review, Phase End (Milestone), Event Driven Analysis can be done using the Schedule
Analysis Tracking Sheet
Used to control the project activities and ensure on-time delivery. Can be used for better estimation for
Significance to ERP Project
future ERP projects leading to effective software process management.
software processes for developing, implementing or proving their software processes. The SMP will
improving a quality management system. An ERP have a set of well-defined metrics that will deal
project involves software processes that will exist with measuring the requirements stability, project
even after the implementation of the ERP system. In schedule slippage and tracking the problems arising
fact, the actual processes start only when the going during the maintenance phase. The SMP can also
alive is fixed during implementation. Traditional be further extended involving other aspects like
software projects have requirements collected from effort distribution, productivity, etc if desired. The
the customers for whom the software is developed. SMP will contain the following:
But in ERP projects, the software is made avail-
able to the customers readymade after fine tuning 1. )>> Source of data and how it will be captured
its functionality to match the exact requirements 2. )>> Periodicity of data capture
of the customers. The purpose of the SMP is to 3. )>> Formulae to measure the software process
enhance the performance of ERP projects by im- 4. )>> Team member responsible for data collection
Metric Requirements Stability Index (RSI)

Number of Requirements Changed
Formulae (Added/Deleted/Modified)/Total Number of Initial
Requirements
Number of requirements changed is the sum of the requirements that were either added/modified/de-
Measure
leted from the initial set of requirements approved by the client. Initial requirements are the number of
(Indirect Metric)
requirements initially approved by the client.
Initial requirement: Initial set of approved requirements from the client. Number of changes in the require-
Source of Data
ment can be found from Client Request Form (CRF) and Client Complaint Form (CCF)
Frequency of
At the end of every phase of ERP implementation using the RSI tracking sheet
Analysis
RSI is used to monitor the magnitude of change in requirements it gives a picture of clarity of the require-
Significance to ERP Project
ments in the customers mind. RSI is useful for fine software process management.
1175
Metric Arrival Rate of Problem (ARP)

Formulae Number of problems arrived (reported by the client) every month
Measure
Number of problems reported by the client in a particular month
(Direct Metric)
Source of Data Modification Request Form
Frequency of The problem arrival rate when compared with the problem closure rate gives an indication of the amount
Analysis of work still pending and helps to anticipate the work for the next month
Significance to ERP Project Used for effective software maintenance
5. )>> Nature of Measurement is known as correlation. Positive or negative

6. )>> Significance of the proposed metric to the values of coefficient of correlation ‘R’ between
ERP Project. two variables indicate positive or negative cor-
relation. We have used Karl Pearson’s Coefficient
The following tables (From Table 1 to Table of correlation which is defined below. Note that
6) collectively form the Software Metrics Plan R has no units and is a mere number. If there is
(SMP). In this SMP, six software metrics have some relationship between two variables, their
been defined, each one dealing with a specific scatter diagram will have points clustering near
software process leading to successful software about some curve.
process management.
The goal of this study is to establish the sig- 1
(∑ X i *Yi -X*Y)
nificance of software metrics in ERP projects. We N
R= 1/2 )>> (1)
use correlation and multiple correlations (Levine 1
1/2
 1 
 2  2
 N ∑ X i -X   N ∑ Yi -Y 
et al 1999) to test the two hypotheses formulated
2
 2

below to investigate the impact of one metric over
another. The relationship between two variables Equation (1) is used to compute the correlation
is such that a change in one variable result in coefficient R. Multiple correlations are used to
a positive or negative change in the other and find the degree of inter-relationship among three
also greater change in one variable results in a or more variables. The objective of using multiple
corresponding greater change in the other. This correlations is to find how far the dependent vari-
Metric Closure Rate of Problems (CRP)

Formulae Number of problems closed each month based on severity
Measure
Number of problems closed in a particular month
(Direct Metric)
Frequency of The problem closure rate when compared with the problem arrival rate gives an indication of the amount
Analysis of work still pending and helps to anticipate amount of work which can be completed next month
1176
Metric Age of Open Problems (AOP)

Formulae Sum of time (days per month) that problems have been open/No. of open problems per month
Measure
Number of problems open in a particular month of each severity
(Direct Metric)
Source of Data Modification request form
Frequency of The age of open problem gives the time for which a problem of a particular severity remains open and
Analysis helps in setting realistic schedule estimates.
able is influenced by the independent variables. Hypothesis 2: The metrics AOP, ACP de-
We denote the multiple correlations between x1, termine maintenance efforts (ME) (i.e)
the dependent variable and x2, x3,…xn, independent person-hours.
variables, by R1.234…n. The multiple correlation
coefficient of x1 on x2 and x3 is the simple correla- In the traditional software projects, from the
tion coefficient between the observed value of x1 view point of software engineering, there are five
and its estimated value b12.3 x2 + b13.2x3 denoted by phases in the software development life cycle
€1.23. We denote the multiple correlation of x1 on (SDLC). They are: Requirements Analysis, De-
x2 by R1(23). Equation (2) gives the value of R1(23). sign, Coding, Testing and Implementation. ERP
is a packaged software and its implementation in
R1(23) = Cov(x1, €1.23)/(√[Var(x1)] * √[Var(€1.23)] )>>(2) an enterprise involves six phases called the ERP
implementation lifecycle (Alexis Leon 2005).
The following two hypotheses are formulated They are: Requirements Analysis, Gap Analysis,
to focus our study on the usage of the software Reengineering, Configuration Management, Test-
metrics defined in the SMP for effective software ing and Maintenance. The RSI value and the SS
process management during the ERP implementa- value for these phases for the ERP project A are
tion and to facilitate statistical analysis. shown in the Table 7. The requirements of the
customers are collected and analyzed during the
Hypothesis 1: The metric RSI influences the requirements analysis phase. The gap analysis
metric schedule slippage (SS) phase is the step of negotiation between the cus-
Metric Age of Closed Problems (ACP)

Formulae Sum of time (days per month) that problems have been closed/No. of closed problems per month
Measure
Number of problems closed in a particular month of each severity
(Direct Metric)
Frequency of The age of closed problems gives the time taken to close a problem of a particular severity and helps
Analysis in setting realistic schedule estimates.
1177
Table 7. RSI and SS values for an ERP Project Á
Phases in ERP Requirements Gap Re- Configuration Main-

Testing
implementation Analysis Analysis engineering Management tenance
Requirements Stability Index
0.27 0.15 0.076 0.63 0.59 0.7
(RSI)
Schedule Slippage (SS) 0.11 0.059 0.072 0.29 0.36 0.51
tomer’s requirements and the functions the ERP This data analysis gives a great deal of support
package possesses (Alexis Leon 2005). to the two hypotheses formulated in this study.
Reengineering is defined as the fundamental The clearest result observed from the value of R
rethinking and radical redesign of business pro- and R1 (23) is: (i) The strong relationship between
cesses to achieve dramatic improvements in criti- RSI and SS; (ii) The influence of the Age of Open
cal, contemporary measures of performance, such Problems (AOC) and the Age of Closed Problems
as cost, quality, service and speed (Alexis Leon (ACP) over the Maintenance efforts (ME). The
2005). Table 7 provides the RSI and SS values most reasonable inference that can be drawn from
computed based on the data from ERP Project this study is the considerable influence of the
A. This Project A was done by a medium scale software processes in boosting the performance
software company involved in the development of of the ERP projects, because it helps the ERP
the ERP software for a manufacturing industry at team to effectively monitor the important software
a lower cost with standard configurations. It is not processes during the ERP implementation.
a large scale information system project like the
one being done by the leading ERP vendors SAP,
Oracle, etc., But from the research perspective, DISCUSSION
these data sets are found to be useful to validate
the hypotheses. The coefficient of correlation ‘R’ Measurement enables us to gain insight into the
between RSI and SS is calculated as 0.9406. process and the project by providing a mechanism
The value of R, being greater than zero, infor objective evaluation. Measurement can be
dicates a positive correlation between the two applied to the software process with the intent of
variables RSI and SS. This indicates that the improving it on a continuous basis. Measurement
requirements stability index has an impact over can be used throughout a software project to as-
the schedule slippage of the phases during the sist in estimation, quality control, productivity
ERP implementation. Hence, a Metric-02 (shown assessment and project control. Measurement
in Table 2) has been defined to monitor the RSI. is a management tool. If conducted properly, it
Data obtained from the same project A were used provides a project manager with insight. And as
to compute the simple correlations between the a result, it assists the project manager and the
metrics ME (x1), AOP (x2) and AOC (x3) as r12 = software team in making decisions that will lead
0.863, r13 = 0.648 and r23 = 0.709 respectively. to a successful project. The intent of the SMP is to
The multiple correlation R1 (23) computed using provide a set of process indicators that lead to long-
the Equation (1) gives 0.864. Since R1 (23) is very term software process improvement. The only
large, it follows that the variables x2 and x3 have rational way to improve any process is to measure
considerable influence on the variable x1. In other specific attributes of the process, develop a set of
words, regression equation of x1 on x2 and x3 will meaningful metrics based on these attributes, and
be excellent. then use the metrics to provide indicators that will
1178
lead to a strategy for improvement. SMP enables in the right path as and when there is a deviation.
the software engineering team to take a strategic The results obtained here underscore the impor-
view by providing insight into the effectiveness tance of effective software process management
of a software process. during the ERP implementation.
Many organizations jump into the implemen- A number of software processes are involved
tation without defining the project in “bite-sized in the software development and its maintenance.
chunks” that can be accomplished in a reasonable The six processes that require monitoring and ef-
period of time. As schedules drag on and require- fective management are considered in this study
ments are heaped on the initial phase, the customer and a SMP has been developed consisting of a
loses faith in the initiative and organizational set of software metrics to quantitatively measure
inertia can take hold. If requirements are managed the software processes. A software process is the
scrupulously and reflected in the form of clearly collection of processes for the various tasks in-
articulated scope elements, the entire project is volved in executing projects for building software
more likely to succeed, and chances are better for systems. As a result of changes in technology,
its ultimate adoption and survival. knowledge, and people’s skill, the processes for
Two kinds of observation are noted in this performing different tasks change. In other words,
research study on using the software metrics for processes evolve with time. With knowledge and
managing the software processes of the ERP proj- experience, processes can, and should, be “fine
ects. One observation is based on the statistical tuned” to give better performance.
analysis and another observation is related to the Software process management is concerned
usefulness of software metrics in ERP projects. with this tuning of the software process. There are
Statistical analysis shows that the SMP proposed many published reports showing the benefits the
in this study plays a significant role in the suc- software process management can bring to qual-
cessful ERP implementation. There is sufficient ity, productivity, and cycle time of ERP projects.
evidence from the literature review that the ERP Currently, most organizations that embark upon
projects are smashed heavily by poor software a software process management program tend
process management, especially the varying to use a framework like the quality models for
requirements due to organizational changes and their process improvement. From the application
customization. These varying requirements lead of the SMP to an ERP project A, it can be seen
to schedule slippage. that it is a straight forward approach to manage
It is apparent that large maintenance efforts the crucial software processes and help the ERP
are required for ERP projects as the real software implementation team to improve it.
processes get life only after ERP implementation. It A future research study could compare the
is evident that the proposed software metrics in the performance of various ERP projects using this
SMP is a pragmatic, useful method for improving SMP with those not using the SMP and the appli-
the software processes in ERP projects, because cation of other statistical techniques like multiple
it enables the ERP team to resolve the important linear regressions, etc. If the software metrics are
issues such as requirements instability, schedule linked with some software quality factors, then
slippage and efforts required for maintenance. the performance of the project can be further
Furthermore, the SMP determines decisions for improved (Mohammad Aishayeb 2003). Hence
the ERP team and provides against the possibilities generating suitable metrics to consider quality
of various defects that might arise during or after aspects of software processes will strengthen the
the implementation. In simple, the SMP helps the developed software product. It is also proposed
ERP team to keep track of their project and pilot it to develop a software tool to execute SMP. In the
1179
future, we plan to generate a project database to Fenton, N. E., & Pfleeger, S. L. (2003). Software
make the results of SMP from each ERP project Metrics-A rigorous and Practical approach. New
publicly available for the ERP team. York: Thomson Publisher.
Glass, R. L. (1998). Enterprise Resource Plan-
ning—Breakthrough and/or term problems. The
REFERENCES
Data Base for Advances in Information Systems,
Aishayeb, M., & Li, W. (2003). An empirical 29(2), 14–16.
validation of object-oriented metrics in two dif- Holland, C. P., & Light, B. (1999). A critical suc-
ferent iterative processes. IEEE Transactions on cess factors model for ERP implementation. IEEE
Software Engineering, 29(11). Software, 16(3), 30–36. doi:10.1109/52.765784
Alexis, L. (2005). Enterprise resource planning. Humphrey, W. (2005). Managing the software
New Delhi, India: Tata McGraw-Hill. process. Reading, MA: Addison-Wesley.
Ambler, S. W. (2002). Agile Modeling: Effective Kitchenham, B. A., Huges, R. T., & Linkman, S.
Practices for Extreme Programming and the Uni- G. (2002). Modeling software measurement data.
fied Process. New York: Wiley. IEEE Transactions on Software Engineering,
Arthur, L. J. (1997). Quantum improvement in 27(9), 788–804. doi:10.1109/32.950316
software system quality. Communications of the Levine, D. M., Berenson, M., & Stephan, D.
ACM, 40(6), 47–52. (1999). Statistics for managers. Upper Saddle
Beck, K. (1999). Embracing changes with River, NJ: Prentice Hall.
extreme programming. Computer,Embracing Lucus, H. C. (n.d.). Implementation: The Key
change with extreme programming, 70–77. to Successful Information Systems. Columbia
doi:10.1109/2.796139 Management, 11(4), 191–198.
Carmel, E., & Sawyer, S. (1998). Packaged soft- Luo, W., & Strong, D. M. (2004, August). A frame-
ware development teams: what makes them dif- work for evaluating ERP implementation choices.
ferent? Information Technology & People, 11(1), IEEE Transactions on Engineering Management,
7–19. doi:10.1108/09593849810204503 51(3). doi:10.1109/TEM.2004.830862
Chidamber, S. R., & Kemerer, C. F. (1994). A Parthasarathy, S., & Anbazhagan, N. (2007).
metrics suite for object-oriented design. IEEE Evaluating ERP Implementation Choices using
Transactions on Software Engineering, 20(6). AHP. International Journal of Enterprise Infor-
doi:10.1109/32.295895 mation Systems, 3(3), 52–65.
Davenport, T. H. (1998). Putting the enterprise into Pressman, S. (2001). Software engineering-A
the enterprise system. Harvard Business Review, practioner’s approach. New Dehli, India: Tata
76(4), 121–131. McGraw-Hill.
Dunsmore, H. E. (1984). Software metrics: an Robey, D., Ross, J. W., & Boudreau, M. C. (2002).
overview of an evolving methodology. Informa- Learning to implement enterprise systems: An
tion Processing & Management, 20, 183–192. exploratory study of the dialectics of change. Jour-
doi:10.1016/0306-4573(84)90048-7 nal of Management Information Systems, 19(1).
1180
Sambamurthy, V., & Kirsch, L. J. (2000). An Stensurd, E., & Myrtveit, I. (2003, May). Iden-
integrative framework of the information sys- tifying high performance ERP projects. IEEE
tems development process. Decision Sciences, Transactions on Software Engineering, 29(5).
31(2), 391–411. doi:10.1111/j.1540-5915.2000.
tb01628.x
Sharma & Goyal. (1994). Mathematical statistics.
Meerut, India: Krishna Prakashan Mandir.
Issues, edited by S. Parthasarathy, pp. 51-60 , copyright 2010 by Information Science Reference (an imprint of IGI Global).
1181
Section V
Organizational and
Social Implications
This section includes a wide range of research pertaining to the social and organizational impact of enterprise
information systems. Chapters included in this section analyze the impact of power relationships in system imple-
mentation, discusses how enterprise systems can be used to support internal marketing efforts, and demonstrate
that perceived shared benefits, system characteristic, and the degree of knowledge of the system are significant
influences on an individual’s willingness to use enterprise resource planning systems. The inquiries and methods
presented in this section offer insight into the implications of enterprise information systems at both a personal
and organizational level, while also emphasizing potential areas of study within the discipline.
1183
Chapter 5.1
Optimization of Enterprise
Information System through a
‘User Involvement Framework
in Learning Organizations’
Sumita Dave
Shri Shankaracharya Institute of Management & Technology, India
Monica Shrivastava
Shri Shankaracharya Institute of Management & Technology, India
AbstrAct steel manufacturing public companies in India,

where ERP is to be adopted. In spite of the fact
Enterprise resource planning (ERP) today is be- that it has a strong backbone of resources in terms
ing adopted by business organizations worldwide of information technology (IT) infrastructure, the
with a view to maximize their capabilities. But implementation process is virtually on a standstill.
more often than not the expected outcomes are In this chapter, an evaluation of the psychological
not delivered due to inaccurate calculations with capabilities of the organization is done. This can
respect to the organization’s ability to adapt to the be evaluated through the mindset of the workforce
change. Although the benefits of enterprise infor- and the willingness with which they are ready to
mation systems in streamlining the functions of the adopt change.
organization cannot be questioned, preparing the
organization to adopt the new system needs more
focused efforts. In order to ensure that the existing INtrODUctION
capabilities of the organizations are an enabler and
not an inhibitor in the adoption process, they need Information technology
to be learning organizations. A study was conducted
in Bhilai Steel Plant (BSP), one of the leading Information Technology is the key driver for change
and is instrumental in the creation of lean organiza-
DOI: 10.4018/978-1-60566-723-2.ch005 tions where technology fully supports the implemen-
Optimization of Enterprise Information System
tation of quality enhancement techniques to meet Hence, in order to ensure that the IT enabled
the growing demands of competition. Moreover change acts as an enabler of growth, it becomes
the competitive pressures and escalating mainte- necessary to evaluate the firm’s learning capa-
nance costs is pressuring organizations to replace bilities. Organizational learning takes place when
the legacy system of operations. The envisioned successful organization learning is transferred to
benefits of IT enabled change is the enhancement an organization’s shared beliefs. Learning is the
of competitive ability through the networking of key competency required by any organization that
geographically distant work groups and a more wants to survive and thrive in the new knowledge
effective utilization of man, material and machine. economy. As organizations grow old though they
While evaluating the benefits of enterprise accumulate competencies, resources and knowl-
information systems, the explicit outcome is edge, there is a possibility that their structures
change in the organization’s system as a whole to become a hindrance to their ability to respond
implement the new practices and processes and to the challenges posed by the competition. A
ideas. With the introduction of a knowledge base, constructivist-learning environment is a place
the challenge for the organization gets magnified where people can draw upon resources to make
as the perceived flexibility when evaluated in sense out of things and construct meaningful solu-
physical terms may be accurate but may fall short tions to problems. It emphasizes the importance
in meeting the much needed psychological flex- of meaningful, authentic activities that help the
ibility. Hence, ERP and other forms of IT enabled learner to construct understandings and develop
solutions, which are being widely adopted with skills relevant for solving problems.
a view to maximize capabilities, are not able to “Make learning part of every day office
deliver the expected outcomes due to such inac- environment” is the mantra to survive in this
curate calculations. competitive world. The Learning Organization
The implementation of any IT enabled op- is one that learns continuously and transforms
erations systems requires a systematic approach itself. Learning takes place in individuals, teams,
which includes the evaluation of the organization’s the organizations, and even the communities with
learning capabilities. Hammer and Champy (1993) which the organizations interact. Learning results
focused on IT based organizational reengineer- in changes in knowledge, beliefs, and behaviors.
ing. Their vision can be summarized along the Learning also enhances organizational capacity
following points. for innovation and growth. The Learning Organi-
zation has embedded systems or mechanisms to
1. Radical transformation: It is time consuming capture and share learning. Thus organizational
and does not happen overnight. learning is an important part of Organizational
2. Changes come from a clean slate through the Transformation process.
conceptualization of gradual work arrange-
ments unlike total quality management. Enterprise Information system
3. The focus of change should be process based.
4. The change needs to be initiated at the top An Enterprise Information System (EIS) is a type
and then directed downwards throughout of management information system made to fa-
the organization. and cilitate and support the information and decision
5. Seamless access to information to one and making needs of senior executives by providing
all. easy access to both internal and external informa-
tion relevant to meeting the strategic goals, of the
1184
organization. It is commonly considered as a spe- indicators of company performance, based on the

cialized form of a Decision support system. EISs information in ERP database and external factors.
are defined as computerized information systems
designed to be operated directly by executive
managers without the need of any intermediar- LEGAcy systEMs
ies. Their aim is to provide fast and easy access
to information from a variety of sources (both The phrase “legacy replacement” has crept its way
internal and external to the organisation). They into almost everyone’s vocabulary throughout
are easily customizable and can be tailored to the both the business and technical ranks. Generally
needs and preferences of the individual executive legacy systems replacement term is used to imply
using it. They deliver information of both soft improvement, or elimination of the negative or
and hard nature. This information is presented obsolete. According to the dictionary, replace
in a format that can be easily accessed and most means “provide an equivalent for” but the new
readily interpreted. This is usually achieved by ‘IT’ based solutions provide much more than the
the utilization of multiple modes of accessing data older systems. Enterprise wide system replace-
and the use of Graphical User Interfaces (GUIs). ments can be extremely disruptive to the entire
Choosing the appropriate software is vital to organization, with catastrophic results if not
designing an effective EIS. Therefore, the software well implemented. Flexibility, speed, immediate
components and how they integrate the data into access and anywhere-anytime computing are no
one system are very important. According to the longer luxuries. They are now the necessities of
Wikipedia, the basic software needed for a typical interacting with customers in an on-line envi-
EIS includes four components: ronment. Use of internet and the availability of
information online have left no stones unturned
1. Text base software. The most common form to increase customer expectations for immediate
of text is probably documents; service and this relatively new expectation has
2. Database. Heterogeneous databases resid- left many businesses scrambling to adopt modern
ing on a range of vendor-specific and open technology like ERP in order to meet the demand.
computer platforms help executives access During the past few years, business houses have
both internal and external data; gone through a technology and process learning
3. Graphic base. Graphics can turn volumes of curve to respond to these new service demands.
text and statistics into visual information for For a multitude of technical reasons, these legacy
executives. Typical graphic types are: time systems simply cannot be adequately retrofitted to
series charts, scatter diagrams, maps, motion address current and future business requirements
graphics, sequence charts, and comparison- in a cost- or time-effective manner.
oriented graphs (i.e., bar charts);
4. Model base. The EIS models contain routine success story Due to
and special statistical, financial, and other Legacy systems
quantitative analysis.
Bhilai Steel Plant, a unit of Steel Authority of
Enterprise Resource Planning (ERP) helps to India Limited (SAIL), a Navratna, is one of the
cater to the changing information requirements most profits making steel industry in India. It was
of the Enterprise Information System very effi- established in the year 1958. Many legacy systems
ciently and effectively. An EIS can provide key are working in the organization presently, some of
1185
Figure 1. (Adapted from C&IT Dept, Bhilai Steel Plant)
which are going to be replaced by the ERP system help of the Digital Company, got transformed on
soon. (see Figure 1) introduction of SUN Servers, Oracle database
Studies have revealed that the rate of success and Developer 2000 and finally went on the web
of IT enabled change is rather low. Although, with Oracle 10g, Apache Web Servers and JSP,
IT enabled change has significantly altered the all developed in-house.
managerial paradigms within the organization with The Materials Management System was de-
a view to raise the organization to a higher plane veloped to take care of procurement of Stores and
of performance through a greater involvement of Spares, quality management, logistics manage-
the employees at large, by functionally integrating ment, bill processing, bill payment and inventory
the various departments, it was found in a study management of these procured items. To facilitate
conducted by us that success begets success may the system, masters like material master, vendor
not always hold true. master, etc were extensively developed and sec-
Some of the areas where legacy systems are tions identified to maintain them.
currently working and running successfully in With the introduction of ATM network across
BSP are shown in Table 1. the Plant, the systems were extended beyond the
Over the years substantial data has been col- boundaries of Materials Management, to the end
lected which is helping the decision making with users. Planning sections across the Plants were able
respect to different business processes. One of the to prepare Purchase Requisitions and send them
most extensive System of Bhilai Steel Plant is the online through the approval process to the Purchase
Materials Management System. This system is Department for procurement action. Customized
as old as computerization in BSP. It started with forms were developed in Legacy system to cater
IBM mainframes and tape drives and graduated to this. They were also able place reservations for
to VAX VMS based system, developed with the items lying in the central stores. Various queries
1186
Table 1.
Groups Areas Benefits

Financial Management Finance, VMS, Oprn. A/c, RMBA, EFBS, Cost Timely closing of accounts, online details of financial
Control, Assets, Braman, ORAMS, E-Payment Transactions, Online CPF Loan, Faster & accurate pay-
etc. ment to parties through e-payment, Advance processing
of Employee Tours etc.
Sales and Order progressing OASIS, CISF Manpower Deployment System, Online Invoice Preparation Vehicle monitoring, Daily
PRAHARI, VATACS etc. Duty Chart for CISF, etc
Employee and Estate Ser- Payroll for regular & DPR employees, Estate Qtr allotment, Salary payment on time, e-TDS, HB
vices Services, CPF & Allied Jobs, Time Offices, Advances and other Loans & recoveries, CPF Ledger,
Leave Accounting, VRS Module, Final Settle- Estate Third Party Billing, Incentives Calculation, VRS
ment module etc. details etc
Personnel Administration and HRIS, HRD, HMS, Contract Mgmt, Contract Personal details, LTC/LLTC details, Training Details,
Hospital Management and Web Labour System, UDAN, DATES, PADAM, DBA Patients & Pharmacy accounting, All details & moni-
Services Activities, Assignment Monitoring System, Web toring of Contacts in non works areas, Registration &
Services etc. online attendance system, Various departmental home
pages etc
Networking, Hardware and H/w Maintenance, Planning & Procurement, Hardware Complaints & monitoring, Procurement of
Procurement Networking and Computer Operation. Computer Assets and consumables including stationeries,
24 x 7 computer operation and monitoring of servers,
Various reports for Users and Top Management etc.
were provided to end users thereby improved the way in reducing the Purchasing Lead time thereby
visibility of stock and reduced over indenting of saving a lot of revenue for the company. Other
stores and spares. This had a tremendous impact units of SAIL have outsourced this application
on the inventory turnover ratio which has consid- through a company Metal Junction. The in-house
erably gone down. development and maintenance of the E – procure-
With the introduction of the concept of Door ment solution has saved a lot of outflow of revenue
Delivery through the Store Issue Note Generation of the company. About 65% of the total volume
System (SINGS), a number of material chasers of procurement is being carried out through the
were redeployed for other jobs in the shops as E – procurement route.
the responsibility of delivery of stores and spares Computerization of Store Bills and Store Ac-
was shifted to the stores department which car- counting section and linking it to the Materials
ried out this function centrally. The vehicles used Management System lead to a huge reduction of
by various shops were redeployed. Thus due to manpower who were required before to process
introduction of this system there was substantial bills manually. The payment lead time also got con-
reduction of manpower and vehicle requirement siderably reduced thereby winning the goodwill
in the shops. of the suppliers of the company. The introduction
With the complete Purchasing activity on the of E – Payment has further reduced the lead time
system, the Purchasing lead time got considerably and the manpower required.
reduced. The introduction of in-house E – Pro- With the built-up of database of transactions and
curement system was a major step forward for historical data, a number of analyses are helping the
the Materials Management System as with this company take correct decisions. Inventory analysis,
the legacy system was treading into the challeng- ABC analysis, XYZ analysis, lead time of procure-
ing Internet world where encryption, security & ment of items, analysis of non moving items and slow
transparency were the new challenges. This again moving items has helped in reducing inventory and
was successfully implemented and has gone a long also take informed decisions during procurement.
1187
Lately the control of Procurement Budget and will cater to all their requirements and also be
Consumption Budget has been strengthened and completely up-to-date. Realizing the requirement
regular monitoring is being done by the higher of such organizations, software like Enterprise
management. Another area of concern of the Resource Planning solution incorporating best
management – Shop Inventory has been addressed business practices have been designed, which
by a legacy system which monitors the shop floor will offer an integrated software solution to all
inventory of BSP. All these are possible only due the business functions of an organization.
to the strong legacy system in Bhilai Steel Plant. In the ever growing business environment the
With the introduction of various legacy systems following demands often plague an organization:
in the company, executives and non executives
were forced to take up learning computers. This • Cost control initiatives
has improved the skill set of employees and also • Need to analyze costs or revenues on a
of the developers of the legacy systems who have product or customer basis
to continuously upgrade themselves to match the • Flexibility to respond to ever changing
pace with the industry. With the introduction of business requirements
E – procurement, even the small suppliers too • More informed management decision
have taken up computers and are hugely benefit- making by seamless information flow
ting out of this. • Changes in ways of doing business
• Simulate the complexity of variables ef-
Limitations of the Present fecting businesses
system in bsP
Difficulty in getting accurate data, timely in-
The legacy systems being used in the plant are formation and improper interface of the complex
able to help meet the demands of the plant in the natured business functions have been identified as
present scenario, but in days to come with the the hurdles in the growth of any business. Time
advent of newer technologies and new competi- and again depending upon the speed of the grow-
tion the present system will not be that effective. ing business needs, many solutions have come up
Hence, the management has decided to implement and ERP is one of such solution.
Enterprise Resource Planning in order to meet the Some of the features of ERP which are com-
changing demands of the market. pelling organizations to make it part and parcel
The present systems have been made to cater of their business activities are:
to demands of a particular area and hence are
not holistic in nature. Because of this, there is • It facilitates company-wide Integrated
no centralized database and the systems are not Information System covering all func-
integrated, resulting in less user involvement and tional areas like Manufacturing, Selling
interaction with the system. and distribution, Payables, Receivables,
Inventory, Accounts, Human resources,
Need for change Purchases etc.,
• It performs core activities and increases
Most organizations across the world have real- customer service.
ized that in a rapidly changing environment, it • It bridges the information gap across the
is impossible to create and maintain a custom organisation i.e. results in seamless flow of
designed or tailor made software package which information.
1188
• It supports multiple platforms, multiple Organization Learning

currency, multimode manufacturing and is
multi-linguistic in nature. “Organizations where people continually expand
• It provides for complete integration of their capacity to create the results they truly desire,
Systems not only across the departments in where new and expansive patterns of thinking
a company but also across the companies are nurtured, where collective aspiration is set
under the same management. free, and where people are continually learning
• It allows automatic introduction of latest to learn together” (Peter Senge, 1990)
technologies like Electronic Fund Transfer,
Electronic Data Interchange (EDI),
Internet, Intranet, E-Commerce etc. Learning organizations are those that have in place
• It not only addresses the current require- systems, mechanisms and processes, that are used
ments of the company but also provides the to continually enhance their capabilities and those
opportunity of continually improving and who work with it or for it, to achieve sustainable
refining business processes. objectives - for themselves and the communities
• It integrates all the levels of informa- in which they participate. (see Table 2)
tion system and provides business intelli- According to Mr. Udai Parekh, “The concept
gence tools like Decision Support Systems of learning organization is a natural extension of
(DSS), Executive Information System organizational learning”. Organizations today are
(EIS), management dashboards, and stra- changing in terms of values, structures, processes
tegic enterprise management, Reporting, and expectations. We need to help the employees
Data Mining and Early Warning Systems prepare for the living in the new organizations.
(Robots) for enabling people to make bet- In the new organizations, it is expected that em-
ter decisions and thus improve their busi- ployees have substantial content knowledge in
ness processes. their work specialization and be well prepared
• It also enables collaborative processes with in the process and behavioral dimensions of their
business partners and links suppliers, orga- experiences in the changing organizations. The
nization and customers more effectively. process and behavioral dimensions such as effec-
tive communication skills and negotiation skills
In order to reap the benefits and implement are attributes that play a vital role in individual
change effectively the companies need to be and group learning. Individuals at all levels in the
learning organizations. organization must combine the mastery of some
technical expertise with the ability to interact
effectively with customers and clients, work pro-
Table 2. (Adapted from FORUM Journal Volume ductively in teams and critically reflect upon and
93.1) then change their own original practices.
Learning organizations creates a language that
Learning Organizations
is ideal, an approach, a vision to move towards
Respond to environment change
a type of organization that one wants to work in
Tolerate stress
and which can succeed in the world of increasing
Compete
change and interdependency. A learning organi-
Exploit new niches
zation requires a basic shift in how we think and
Take risks/ mutate
interact and these changes go to bedrock assump-
Develop symbiotic relationships
tions and habits of our culture.
1189
Key Variables Information Flow (X5): Transparency in

working, seamless flow of information, sharing
According to Mr. Udai Parekh, the key variables of critical information at all levels, minimum
which play a very important role during organiza- grapevine, encouraging development by sharing
tion learning are:- is a very important aspect.
Emotional maturity (X6):. An environment
1. Holistic frame (X1) where integrity, discipline, devotion, teamwork,
2. Strategic thinking (X2) transparency, mutual respect and moral responsi-
3. Shared vision (X3) bility should be created which will help to evolve a
4. Empowerment (X4) culture of trust thus, enabling the employees to be
5. Information Flow (X5) more loyal, committed, able to control their emo-
6. Emotional maturity (X6) tions and keep the organizations achievement a
7. Learning (X7) step higher than their own personal achievements.
8. Synergy (X8) Learning (X7): A learning environment en-
courages self-development allowing space for
Holistic framework (X1): Learning should be discussions, enquiries, freedom of speech and
done within holistic framework, i.e, by taking into reward schemes. The management should felici-
account the environment in which the organization tate people who are ready to accept the change.
functions, putting emphasis on the causatives of Synergy (X8): The dynamic energetic atmo-
the problem rather than the symptoms, learning sphere is created when participants interact and
not just for short term gains but for a vision, by productively communicate with each other and in
understanding the relationships and interrelations groups. The cooperative efforts of the participants
between the various facets of an organization. create an enhanced combined effect compared to
Strategic thinking (X2): careful strategies the sum of their individual effects. An effort for
should be design in terms of which areas to be coordinated activities, enhanced teamwork, cross-
targeted first, what would be the implications of functional teams, is a must for the organizations
each step, clearly defining the roles and policies of to be called learning organizations.
the organization, creating an environment which Bhilai Steel Plant, a unit of SAIL with strength
will support learning. The strategy should be well of 34,800 employees, is an integrated steel plant
communicated at all levels to ensure its success. with end-to-end processes of generation of Raw
Shared vision (X3): Vision should be devel- Materials to selling the finished goods. Different
oped by employee’s participation. Communicating departments have different systems with less of
vision and freezing it is also an important aspect, integration resulting into duplicity and limited
which needs to be taken into consideration. The information flow. With the competitive business
top management must lay emphasis on the creating scenario and the need of advanced functionalities
an environment full of transparency, motivation like Strategic Enterprise Management, Business
and help nurture creativity and leadership traits. Information Warehousing, Business Process
Empowerment (X4): creating an environment Simulation, etc. the need of a standard world-
where there are decentralized structures which class product is being felt which could enhance
enable more participation, more power, a culture effectiveness of the processes and functions within
of trust, faster decision making, rewards and the system. The existing bureaucratic culture is
incentives is very important aspect and cannot impairing holistic framework, empowerment,
be ignored. organization climate, shared vision, which is
1190
making employees feel less motivated and hence Level of Consideration: 5% level of signifi-
the overall learning process has taken a backseat. cance
The major players of learning process, namely
structure, processes and information flow in the
current scenario is becoming a deterrent in making Hypothesis
the company, outsmart its competitors and face
the change in economy. H0: All parameters have the same potential ac-
Hence an LOP survey was done on BSP to cording to the employees
identify the key areas where improvement is pos-
sible to enhance the learning capabilities of the (X1=X2=X3=X4=X5=X6=X7=X8)
organization so as to facilitate a smooth adoption
of technologies.
H1: All the parameters do not have the same po-
research Methodology tential according to the employees.
Sampling: Simple Random sampling was un- (Xi ≠ Xj), (where i, j =1,2,3,4,5,6,7,8)
dertaken on Bhilai Steel Plant and five (5) major
departments namely; materials management,
plant maintenance, purchase, finance and quality Assumption
management were covered.
30 points are assumed to be the standard value of
Sample Size: 50 Executives. parameters (where x >= 30) (see Table 3)
Data Collection Period: Prior to the imple- Findings

mentation of ERP.
The X1, X3 and X6 (Holistic framework, shared
Statistical Instruments Used: Single factor vision, emotional maturity) parameters are below
ANOVA to test whether the 8 parameters are 30 (all respondents hold the same opinion.)
significantly different or not.
Table 3. ANOVA: Single factor
For Elements X1, X3 and X6

SUMMARY
Groups Count Sum Average Variance
X1 50 1463 29.26 100.32
X3 50 1411 28.22 125.28
X6 50 1409 28.18 99.46
ANOVA
Source of Variation SS df MS F P-value F crit
Between Groups 37.49 2 18.75
0.17 0.84 3.06
Within Groups 15927.58 147 108.35
Total 15965.07 149
H0: Accepted Conclusion: X1, X3 and X6are not significantly different at 5% level of significance
1191
Inference 1. A holistic framework of operations

It is concluded that parameters X1, X3 and X6 are 2. Shared vision – free flow of communication,
really below the assumed value indicating that they hence transparency and
are weak parameters. All the employees support 3. Emotional maturity – Poor teamwork and
this opinion (refer to table 1). lack of commitment and mutual respect
The rest five parameters (strategic thinking,
empowerment, information flow, emotional suggestions
maturity, learning and synergy) are above 30 (all
respondents hold the same opinion.) (see Table 4) In order to overcome the identified deficiencies
and to induce organization learning in a better way,
Inference it is essential that BSP adopts practices in order
It is further concluded that parameters X2, X4, to make ground more fertile to adopt ERP thus,
X5, X7, X8, are the strong parameters and all the initiate the process of transformation. It needs to
employees support this.
It can be supported by table –3 where all pa- 1. Search for novel solutions
rameters are taken together and all the employees 2. Give and seek information
have different opinions which suggests that there 3. Honour the contributions of others
is bipolarization of opinion. (see Table 5) 4. Seek and give evaluation
5. Question basic assumptions and practices
Inference
From 1-8 parameters, if ANOVA is conducted at a In order to encourage experimental learning
time all employees hold different opinions about and collaborative teamwork which is both problem
different parameters. finding as well as problem solving, suggested
It can be observed that BSP lacks methods to stimulate creative learning include
Table 4.
For Elements X2, X4, X5, X7 and X8

SUMMARY
X2 50 1657 33.14 89.43
X4 50 1614 32.28 135.96
X5 50 1561 31.22 165.73
X7 50 1858 37.16 217.16
X8 50 1588 31.76 181.66
ANOVA
Between Groups 1124.344 4 281.09
1.78 0.13 2.41
Within Groups 38706.520 245 157.99
Total 39830.864 249
H0: Accepted
Conclusion: X2, X4, X5, X7 and X8 are not significantly different at 5% level of significance
1192
Table 5.
For All Elements from X1 to X8

SUMMARY
X1 50 1463 29.26 100.32
X2 50 1657 33.14 89.43
X3 50 1411 28.22 125.28
X4 50 1614 32.28 135.96
X5 50 1561 31.22 165.73
X6 50 1409 28.18 99.46
X7 50 1858 37.16 217.16
X8 50 1588 31.76 181.66
ANOVA
Between Groups 3110.0975 7 444.30
3.19 0.003 2.03
Within Groups 54634.1000 392 139.37
Total 57744.1975 399
H0: Rejected Conclusion: Elements from X1 to X8 are significantly different at 5% level of significance
1. RAT (Role Analysis Transaction) but also enhances commitment thus stimulating
2. Diagnostic Window critical thinking about the organization needs
and priorities.
RAT is a tool for improving the effectiveness RAT and Diagnostic window can be most
of work groups. It helps to clarify the role expec- helpful in conditions of structural change such as
tations; i.e. the expectations the members of the that of BSP as they help to address personal and
work group have for their own performance and behavioral adjustments to the creation of change.
for the performance of other group members. The
role requirements are determined by consensus,
which ultimately results in more effective and cONcLUsION
mutually satisfactory performance. It enhances
participation and collaboration, hence enhanced To be able to maintain the market share there is
holistic framework, shared vision and emotional a definite need for change and the organizations
maturity. with is capable of learning are able to implement
Diagnostic Window – It helps the groups the same in more effective manner. Change is ac-
to identify important issues and problems. The companied by resistance which can be managed
issues are discussed according to the following
quadrants shown in Table 6.
The idea is to reach a consensus on issues Table 6.
and the likelihood of change where it is required
Amenable to change Potential Operational
and help the group to define some action plans
Not Amenable to change Disaster Temporary
to address the needed change. This activity not
only encourages participation and communication
1193
by proper user training and communication. Thus, Dayal, I., & Thomas, J. M. (1968). Role
although it is very necessary for organizations to analysis technique, operation KPE: Develop-
adopt IT enabled operations; a proper analysis of ing a new organization. The Journal of Ap-
the existing legacy system needs to be done in the plied Behavioral Science, 4(4), 473–506.
light of the existing system so as to ensure full doi:10.1177/002188636800400405
and timely acceptability of the new technology.
Garwin, D. A. (1993). Building a learning orga-
nization. Harvard Management Review, 88-91.
rEFErENcEs Gupta, V. (2004). Transformative organizations:
A global perspective. Response Books.
Amburgey, T., Kelly, D., & Barnett, W. P. (1993).
Resetting the clock: The dynamics of organiza- Hammer, M., & Champy, J. (1993). Reengineering
tion change and failure. Administrative Science the corporation: A manifesto for business revolu-
Quarterly, 38, 51–73. doi:10.2307/2393254 tion. New York: Harper Collins.
Argyris, C. (1982). Reasoning learning and ac- Hammer, M., & Stanton, S. A. (1995). The reen-
tion: Individual and organizational. Jossey Bass. gineering revolution handbook. London: Harper
Collins.
Argyris, C. (1991). Teaching smart people how
to learn. Harvard Business Review. Leedy, P. D., & Ormrod, J. E. (2004). Practical
research: Planning and design. PHI.
Baron, J. N., Burton, D. M., & Hannan, M. T.
(1996). The road taken: Origins and evolution Linden, R. (1984). Seamless government: A
of employment. practical guide to reengineering in public sector.
Booth, W. C., Williams, J. M., & Colomb, G. Lyons, P. (1999). Assessment techniques to en-
G. (2003). The craft of research. University of hance organization learning. Opinion papers.
Chicago. Pareek, U. (2002). Training instruments in HRD
Bordens, K. S. (2005). Research design methods, & OD, 2nd edition. TMH.
6th edition. Bruch & Abbot, TMH. Ruma, S. (1974). A diagnostic model for organi-
Coulsin-Thomas, C. (1996). Business process zational change, social change (pp. 3-5).
reengineering: Myth and reality. London: Kogna Seinge, P. (1991). An interview with Peter Seinge:
Page. Learning organizations made plain. Training and
Courtney, N. (1996). BPR sources and uses. In Development.
C. Coulsin-Thomas (Ed.), Business process engi- Senge, P. (n.d.). The fifth discipline: The art and
neering: Myth and reality (pp. 226-250). London: practice of learning organizations. Doubleday/
Kogna Page. Currency.
Davenport, T. H. (1996). Process innovation: Re- Shajahan, S. (2005). Research methods for man-
engineering work through information technology. agement, 3rd edition. Jaico Publishing House.
1194
Vaughan, D. (1997). The trickle down effect:

Policy decisions, risky work, and the challenge
tragedy. California Management Review, 39(2),
80–102.
for Reliable and Scalable Operations, edited by Nijaz Bajgoric, pp. 78-90, copyright 2010 by Information Science Reference
1195
1196
Chapter 5.2
Authority and Its
Implementation in Enterprise
Information Systems
Alexei Sharpanskykh
Vrije Universiteit Amsterdam, The Netherlands
AbstrAct organizations of different types. Moreover, the

article demonstrates how the formalized authority
The concept of power is inherent in human or- relations can be integrated into an EIS.
ganizations of any type. As power relations have
important consequences for organizational vi-
ability and productivity, they should be explicitly INtrODUctION
represented in enterprise information systems
(EISs). Although organization theory provides a The concept of power is inherent in human orga-
rich and very diverse theoretical basis on organizations of any type. Power relations that exist
nizational power, still most of the definitions for in an organization have a significant impact on
power-related concepts are too abstract, often its viability and productivity. Although the notion
vague and ambiguous to be directly implemented of power is often discussed in the literature in
in EISs. To create a bridge between informal social studies (Gulick &Urwick, 1937; Parsons,
organization theories and automated EISs, this 1947; Friedrich, 1958; Blau & Scott, 1962; Pea-
article proposes a formal logic-based specification body, 1964; Hickson et al., 1971; Bacharach &
language for representing power (in particular Aiken, 1977; Clegg, 1989), it is only rarely de-
authority) relations. The use of the language is fined precisely. In particular, power-related terms
illustrated by considering authority structures of (e.g., control, authority, influence) are often used
Authority and Its Implementation in Enterprise Information Systems
interchangeably in this literature. Furthermore, mechanisms. In particular, the architecture ARIS

the treatment of power in different streams of (Scheer & Nuettgens, 2000) used for development
sociology differs significantly. One of the first of EISs identifies responsibility and managerial
definitions for power in the modern sociology was authority relations on organizational roles, how-
given by Max Weber (1958): Power is the prob- ever, does not provide general mechanisms for
ability that a person can carry out his or her own representing such relations and does not address
will despite resistance. Weber and his followers change of these relations over time. The enterprise
(Dahl, Polsby) considered power as an inherently architecture CIMOSA (1993) distinguishes re-
coercive force that implied involuntary submission sponsibilities and authorities on enterprise objects,
and ignored the relational aspect of power. Other agents, and processes/activities. However, no
sociologists (Bierstedt, Blau) considered power precise meaning (semantics) is attached to these
as a force or the ability to apply sanctions (Blau concepts, which may be interpreted differently
& Scott, 1962). Such view was also criticized as in different applications. Also, different aspects
restrictive, as it did not pay attention to indirect of authorities are not distinguished both in ARIS
sources and implications of power (e.g., informal and in CIMOSA (e.g., authority for execution, au-
influence in decision making) and subordinate’s thority for supervision, authority for monitoring).
acceptance of power. Parsons (1947) considered Often EISs realize extensive access schemata
power as “a specific mechanism to bring about that determine allowed actions for roles and modes
changes in the action of organizational actors in of access of roles to information (Bernus, Nemes,
the process of social interaction.” & Schmidt, 2003). Normally, such schemata are
Most contemporary organization theories based on power relations established in orga-
explore both formal (normative, prescribed) and nizations. Thus, to ensure consistency, unam-
informal (subjective, human-oriented) aspects of biguousness and completeness of EISs’ access
power (Peabody, 1964; Clegg, 1989; Scott, 2001). schemata, organizational power relations should
Formal power relations are documented in many be precisely identified and specified using some
modern organizations and, therefore, can be ex- (formal) language. To this end, theoretical findings
plicitly represented in models on which enterprise on organization power from social science are
information systems (EISs) are based. The repre- useful to consider. However, there is an obstacle
sentation of formal power in EISs has a number to the direct implementation of this knowledge in
of advantages. First, it allows a clear definition of EISs—the absence of operational definitions of
rights and responsibilities for organizational roles power-related concepts in social theories.
(actors) and a power structure. Second, based on The first step to make the concept of power
the role specifications, corresponding permissions operational is to provide a clear and unambigu-
for information, resources and actions can be ous meaning for it (or for its specific aspects). In
specified for each role. Third, explicitly defined this article, this is done by identifying the most
rules on power enable the identification of viola- essential characteristics and mechanisms of power
tions of organizational policies and regulations. described in different approaches and by integrat-
Fourth, data about power-related actions (e.g., ing them into two broad categories: formal power
empowerment, authorization) can be stored in (or authority) and informal power (or influence),
an EIS for the subsequent analysis. which are described in the Power, Authority and
For modeling of power relations, the rich Influence section. Further, this article focuses on
theoretical basis from social science can be used. the formal representation of authority, for which a
Notably, many modern EISs implement no or very formal language is described in the Authority: A
simplified representations of power relations and Formal Approach section. Moreover, this section
1197
illustrates how the introduced formal language can Although authority and influence often stem
be used to model authority systems of different from different sources, they are often interrelated
types of organizations. The next section discusses in organizations. For example, the probability of
the integration of formal authority relations into the successful satisfaction of organizational goals
an automated EIS. Finally, the article concludes increases, when a strong leader (meaning a leader
with a Discussion section. that has a great value of influence) occupies a supe-
rior position of authority. Furthermore, sometimes
patterns of influence that frequently occur in an
POWEr, AUtHOrIty AND organization may become institutionalized (i.e.,
INFLUENcE may become authority relations).
Modeling methods for authority and influence
As in many contemporary social theories (Pea- are essentially different. While authority relations
body, 1964; Clegg, 1989), we assume that power are often prescriptive and explicitly defined,
can be practiced in an organization either through influence relations are not strictly specified and
(formal) authority or through (informal) influence may vary to a great extent. Therefore, whereas
relations. Authority represents formal, legitimate authority relations can be generally represented in
organizational power by means of which a regu- EISs, the specification of influence relations is de-
lated normative relationship between a superior pendant on particular (cognitive) models of agents
and a subordinate is established. Usually, author- that represent organizational actors. Relations
ity is attached to positions in organizations. For between authority and influence can be studied
example, authority of some managerial positions by performing simulation with different types of
provides power to hire or to fire; to promote or to agents situated in different organizational environ-
demote; to grant incentive rewards or to impose ments. The focus of this article is on modeling of
sanctions. In many approaches, it is assumed formal authority relations. Influence relations and
that authority implies involuntary obedience relations between authority and influence will be
from subordinates. Indeed, as authority has a considered elsewhere.
normative basis that comprises formal, explicitly
documented rules, it is expected that subordinates,
hired by the organization, should be aware of and AUtHOrIty: A FOrMAL
respect these rules, which implies the voluntary APPrOAcH
acceptance of authority.
All manifestations of power that cannot be First, a formal language for specifying authority-
explained from the position of authority fall into related concepts and relations is introduced.
the category of influence. In contrast to author- The next section discusses how the introduced
ity, influence does not have a formal basis. It is language can be used for representing authority
often persuasive and implies voluntary submis- structures of organizations of different types.
sion. Some of the bases of influence are techni-
cal knowledge, skills, competences and other A Formal Language
characteristics of particular individuals. Influence
is often exercised through mechanisms of leader- Simon (1957) describes three contributions of
ship; however, possession of certain knowledge authority for an organization: (1) the enforcement
or access to some resources, as well as different of responsibility, (2) the specialization of decision
types of manipulation may also create influence. making, and (3) the coordination of activity. Based
Influence may be realized in efforts to affect on this and other theoretical findings that describe
organizational decisions indirectly.
1198
power, duties and responsibilities of organizational All types of decisions with respect to a particu-
positions (Mintzberg, 1979), a number of relations lar process can be divided into two broad groups:
for the specification of formal authority can be technological and managerial decisions (inspired
identified. These relations are defined on positions by Bacharach and Aiken [1977]). Technological
(or roles), without considering particular agents decisions concern technical questions related to
(individuals). The relations are formalized using the process content and are usually made by tech-
the order sorted-predicate language (Manzano, nical professionals. Managerial decisions concern
1996) and are presented graphically in Figure 1. general organizational issues related to the process
We represent all activities of an organization (e.g., the allocation of employees, process schedul-
(including decision making and personnel-related ing, the establishment of performance standards,
activities) by processes. Each organizational role provision of resources, presenting incentives and
is associated with one or more process. Roles sanctions). Managers of different levels (i.e., from
may have different rights and responsibilities the lowest level line managers to strategic apex
with respect to different aspects of the process [top] managers) may be authorized for making
execution. Furthermore, often several roles may different types of managerial decisions varying
potentially execute or manage certain processes. from in scope, significance and detail. A particu-
This is represented by the relation lar decision type is specified as an aspect in the
is_authorized_for relation. The same holds for
is_authorized_ for: r: ROLE x aspect: ASPECT technological decisions. Whereas consulting has
x a: PROCESS, a form of recommendation and implies voluntary
acceptance of advices, decisions imposed on a
where aspect has one of the values {execution, role(s) that execute(s) the process are considered
monitoring, consulting, tech_des (making techno- as imperatives with corresponding implications.
logical decisions), manage_des (making manage- Authorization for execution implies that a
rial decisions), user_defined_aspect}. role is allowed to execute the process according
Figure 1. Graphical representation of the concepts and relations of the language used for specifying
formal authority relations
is_ re p o n s ib le _ fo r
is_ a u th o rize d _ fo r a u th o rize s _ fo r/
d isa llo w s b in a ry re la tio n
n -a ry re la tio n
... ...
in te ra cts_ w ith
0 :n c o n c e pt
0 :n
0 :n
subro le _ o f p ro c e ss
aspect 0 :n role
0 :n
1 :n
1 :n 1 :n
1 :n has_contro l_ o v e r
agent is_ a llo c a te d _ to
re source
assigns_ re sp o n s ib ility_ is_ in s ta n c e _ o f 0 :n
0 :n to _ fo r 1 :1
uses
is_ c om m itte d _ to 0 :n
is_ s u b ord in a te _ o f_ fo r ta sk
w ishes 1 :n
1 :n 1 :n
1 :n goal is_ re a lizable _ b y 0 :n

1 :n
1 :n m e a s u re s
is_exp re ssed_ove r PI
1 :n 1 :n
1199
to existing standards and guidelines. Whenever authorizes_ for: r1: ROLE x r2: ROLE x aspect:
a problem, a question or a deviation from the ASPECT x a:PROCESS: role r1 gives the
standard procedures occurs, the role must report authority for aspect of process a to role r2
about it to the role(s) authorized for making tech-
nological/managerial (depending on the problem disallows: r1: ROLE x r2:ROLE x aspect: ASPECT
type) decisions and must execute the decision(s) x a: PROCESS: role r1 denies the authority
that will follow. for aspect of process a for role r2.
Monitoring implies passive observation of
(certain aspects of) process execution, without However, to make a role actually responsible
intervention. for a certain aspect of the process, another role
Notice that other aspects of process execu- besides the authority to make managerial decisions
tion described in the managerial literature (e.g., should also be the superior of the role with respect
control, supervision) can be represented as a to the process. Superior-subordinate relations with
combination of already introduced aspects. In respect to organizational processes are specified
particular, control can be seen as the conjunction by: is_subordinate_of_for: r1: ROLE x r2: ROLE
of monitoring and making technological and/or x a: PROCESS. Then, responsibility is assigned/
managerial decisions aspects; supervision can retracted using the following relations:
be defined as the combination of consulting and
control. Furthermore, the designer is given the assigns_responsibility_to_ for: r1: ROLE x r2:
possibility to define his/her own aspects and to ROLE x aspect: ASPECT x a: PROCESS:
provide an interpretation to them. role r1 assigns the responsibility for aspect
Although several roles in an organization of process a to role r2.
may be authorized for a certain aspect related to retracts_responsibility_from_for: r1: ROLE x r2:
some process, only one (or some) of them will be ROLE x aspect: ASPECT x a: PROCESS:
eventually (or are) responsible for this aspect. For role r1 retracts responsibility from role r2
example, the responsibility of a certain role with for aspect of process a.
respect to the process execution means that the
role is actually the one who will be performing Using these relations, superiors may delegate/
the process and who holds accountability of the retract (their) responsibilities for certain aspects
process execution. Furthermore, responsibility for of processes execution to/from their subordinates,
the process execution implies allowance to use and may restrict themselves only to control and
resources required for the process performance. making decisions in exceptional situations.
The responsibility relation is specified as: In Hickson et al. (1971), control over resources
is identified as an important source of power.
is_responsible_ for: r:ROLE x aspect:ASPECT x Therefore, it is useful to identify explicitly which
a: PROCESS: process a is under responsibil- roles control resources by means of the relation
ity of role r with respect to aspect (defined has_control_over: r1: ROLE x res: RESOURCE.
as for authorized_for). In the proposed modeling framework, the notion
of resource includes both tangible (e.g., materi-
Some roles are authorized to make managerial als, tools, products) and abstract (information,
decisions for authorizing/disallowing other roles data) entities.
for certain aspects with respect to process execu- Our treatment of authority is different from
tion. The authorization/disallowance actions are both formal approaches that consider authority as
specified by the following relations: an attribute or a property inherent in an organiza-
1200
tion (Gulick & Urwick, 1937; Weber, 1958) and may be provided for some time interval). To ex-
from the human-relation view that recognizes press temporal aspects of authority relations, the
authority as an informal, non-rational and sub- temporal trace language (TTL) (Jonker & Treur,
jective relation (e.g., Follett, Mayo, cf. [Clegg, 2003) is used.
1989]). As many representatives of converging TTL allows specifying a temporal development
approaches (e.g., C.I. Barnard, Simon [1957]), of an organization by a trace. A trace is defined
we distinguish between the formal authority as a temporally ordered sequence of states. Each
prescribed by organizational policies and actual state corresponds to a particular time point and is
authority established between a superior and his/ characterized by a set of state properties that hold
her subordinate in the course of social interac- in this state. State properties are formalized in a
tions. In the latter case, a special accent lies on standard predicate logic way (Manzano, 1996)
the acceptance of authority by a subordinate. In using state ontologies. A state ontology defines
Clegg (1989), different cases of the authority ac- a set of sorts or types (e.g., ROLE, RESOURCE),
ceptance are discussed: orders anticipated and sorted constants, functions and predicates.
carried out (anticipation); acceptance of orders States are related to state properties via the
without critical review; conscious questioning but formally defined satisfaction relation |=: state(γ,
compliance (acceptance of authority); discusses t) |= p, which denotes that state property p holds
but works for changes; ignores, evades, modifies in trace γ at time t. For example, state(γ1, t1)
orders (modification and evasion); rejection of |= is_responsible_for(employee_A, execution,
authority (appeals to co-workers or higher rank p1) denotes that in trace γ1 at time point t1 the
for support). Depending on the organizational employee_A is responsible for the execution of
type, varying administrative sanction may be process p1.
applied in case an employee does not accept an Dynamic properties are specified in TTL by
authoritative communication, when he/she: (a) relations between state properties. For example,
correctly understands/interprets this communica- the following property expresses the rule of
tion; (b) realizes that this communication complies a company’s policy that an employee is made
with formal organizational documents and/or is responsible for making technological decisions
in line with organizational goals; (c) is mentally with respect to process p1 after s/he have been
and physically able to perform the required ac- executing this process for two years (730 days):
tions. In many modern organizations, rewards
and sanctions form a part of authority relation, ∀γ: TRACE ∀t1: TIME ∀empl: EMPLOYEE
thus, explicitly defined: state(γ, t1) |= is_responsible_for(empl, ex-
ecution, p1) & ∃t2: TIME state(γ, t2) |= as-
grants_reward_to_ for: r1: ROLE x r: REWARD signs_responsibility_to_for(management,
x r2: ROLE x reason: STRING: role r1 grants empl, execution, p1) & t1-t2 = 730
reward r to role r2 for reason ⇒ state(γ, t1) |= assigns_responsibility_to_
for(management, empl, tech_des, p1).
imposes_saction_on_for: r1: ROLE x s: SANC-
TION x r2: ROLE x reason: STRING: role Other specific conditions (e.g., temporal, situ-
r1 imposes sanction s to role r2 for reason. ational) under which authority relations may be cre-
ated/maintained/dissolved are defined by executable
Sometimes authority relations may be defined rules expressed by logical formulae. The specifica-
with respect to particular time points or intervals tion of these rules will be discussed in the Integration
(e.g., responsibility for some aspect of a process of Autority Relations Into an EIS section.
1201
Modeling Authority relations in The owners controlled all resources (∀r:

Different types of Organizations RESOURCE ∀t: TIME ∀γ: TRACE state(γ, t)
|= has_control_over(firm_owner, r)). Currently,
Authority is enforced through the organizational similar types of organizations can be found in
structure and norms (or rules) that govern the orga- family business and small firms.
nizational behavior. In general, no single authority With the growth of industry, which caused
system can be equally effective for all types of joining of small firms into larger enterprises,
organizations in all times. An organizational au- owners were forced to hire subcontractors, who
thority system is contingent upon many organiza- took over some of their managerial functions. This
tional factors, among which organizational goals; can be modeled using the introduced language
the level of cohesiveness between different parts as assigning responsibility to subcontractors by
of an organization, the levels of complexity and of the owner for some managerial and technological
specialization of jobs, the level of formalization decisions, as well as monitoring and consulting
of organizational behavior, management style (a of workers with respect to some processes execu-
reward system, decision making and coordination tion. For example, the responsibility assignment
mechanisms), the size of an organization and its to role subcontractor_A for making managerial
units. Furthermore, the environment type (its and technological decisions related to the process
uncertainty and dynamism; the amount of com- p1 is expressed as:
petitors), as well as the frequency and the type
of interactions between an organization and the ∀γ: TRACE ∃t: TIME state(γ, t) |=
environment exert a significant influence upon [ assigns_responsibility_to_for(firm_owner,
an organizational authority structure. subcontractor_A, tech_des, p1) ∧ as-
In the following it will be discussed how author- signs_responsibility_to_for(firm_owner,
ity is realized in some types of (mostly industrial) subcontractor_A, manage_des, p1) ].
organizations and how it can be modeled using
relations introduced in the previous section. The owner reserved often the right to control
Authority in small firms of the early industrial for himself, which included granting rewards
era was completely exercised by their owners and imposing sanctions to/on subcontractors and
through mechanisms of direct personal control. workers, realized through superior-subordinate
Firm owners were managers and technical pro- relations. For example, the following rule describes
fessionals at the same time, and, therefore, had the superior-subordinate relations between the
authority and responsibility for all aspects related firm owner and subcontractor_A, responsible for
to processes, except for their execution, respon- making technological decisions related to process
sibility for which was assigned to hired workers. p1 and employee_A responsible for execution of
This can be expressed using the introduced formal process p1:
language as follows:
∀γ: TRACE ∀t: TIME state(γ, t) |= is_subordi-
∀p: PROCESS ∀t: TIME ∀γ: TRACE ∃empl: nate_of_for(subcontractor_A, firm_owner,
HIRED_EMPLOYEE state(γ, t) |= [ is_re- p1) & is_subordinate_of_for(employee_B,
sponsible_for(firm_owner, control, p) & is_ firm_owner, p1).
responsible_for(firm_owner, supervision, p)
& is_responsible_for(empl, execution, p) ]. Organizational resources were usually con-
trolled by the owner.
1202
Large industrial enterprises of the 20th century the following superior-subordinate relations can
are characterized by further increase in number of be identified: is_subordinate_of_for(manager2,
managerial positions structured hierarchically by manager1, p) & is_subordinate_of_for(manager3,
superior-subordinate relations. Such organizations manager2, p).
are often defined as mechanistic (Scott, 2001) Initially, such enterprises operated in rela-
and have the following typical characteristics: tively stable (however, sometimes complex)
strong functional specialization, a high level of environmental conditions that reinforced their
processes formalization, a hierarchical structure structure. However, later in the second half of
reinforced by a flow of information to the top of the 20th century, to survive and to achieve goals
the hierarchy and by a flow of decisions/orders in the changed environmental conditions (e.g.,
from the top. Responsibilities were clearly de- a decreased amount of external resources; in-
fined for every position in a hierarchy. In most creased competition; diversification of markets),
organizations of this type, responsibility for enterprises and firms were forced to change their
execution was separated from responsibilities to organizational structure and behavior. In response
make decisions. Managerial positions differed in to the increased diversity of markets, within some
power to make decisions depending on the level enterprises specialized, market-oriented depart-
in the hierarchy. Often, technological decisions ments were formed. Such departments had much
were made by managers of lower levels (or even of autonomy within organizations. It was achieved
by dedicated positions to which also execution by assigning to them the responsibility for most
responsibilities were assigned), whereas manage- aspects related to processes, which created prod-
rial decisions were made by managers at the apex. ucts/services demanded by the market. Although
For example, the following formal expression department heads still were subordinates of (apex)
identifies one of the upper managers responsible manager(s) of the organization, in most cases
for making strategic decisions related to process the latter one(s) were restricted only to general
p, one of the middle level managers responsible performance control over departments. Often
for making tactical decisions related to p and one departments controlled organizational resources
of the first level managers responsible to making necessary for the production and had the structure
technological decisions related to p: of hierarchical mechanistic type.
Although a hierarchical structure proved to be
∃manager1: UPPER_MANAGER ∃manager2: useful for coordination of activities of organiza-
MIDDLE_LEVEL_MANAGER ∃man- tions situated in stable environments, it could
ager3: FIRST_LEVEL_MANAGER ∀γ: cause significant inefficiencies and delays in
TRACE ∀t: TIME state(γ, t) |= [ is_respon- organizations situated in dynamic, unpredictable
sible_for(manager1, making_strategic_de- environmental conditions. Furthermore, the for-
cisions, p) ∧ is_responsible_for(manager2, malization and excessive control over some (e.g.,
making_tactical_decisions, p) ∧ is_respon- creative and innovative) organizational activities
sible_for(manager3, tech_des, p) ]. often can have negative effects on productivity.
Nowadays, large enterprises often create project
In many of such organizations, managers at teams or task forces that are given complex,
the apex shared responsibility for making (some) usually innovative and creative tasks without
decisions with lower-level managers. Therefore, detailed descriptions/prescriptions. As in the case
decisions that were usually proposed by lower with departments, teams are often assigned the
level managers had to be approved by the apex responsibility to make technological and (some)
managers. In connection to the previous example, managerial decisions and are given necessary
1203
resources to perform their tasks. For example, other allow much flexibility in defining authority
the following formal expression represents the relations between roles. In the former case formal
responsibility assignment to the team_A for authority is replaced by socially created informal
making technological and strategic managerial rules. In the latter case, authority may be tempo-
decisions related to the process of development rally provided to the role that has the most relevant
of a design for a new product: knowledge and experience for current organiza-
tional tasks. In many organic organizations formal
∀γ: TRACE ∃t: TIME state(γ, t) |= control and monitoring are replaced by informal
[ assigns_responsibility_to_for(management, mutual control and audit. For the investigation
team_A, tech_des, develop_design_new_ of dynamics of organic organization, informal
product_A) ∧ assigns_responsibility_to_for aspects such as influence, leaderships, mental
(management, team_A, strategic_manageri- models of employees are highly relevant, which
al_des, develop_design_new_product_A) ]. will be discussed elsewhere. Often interactions
between organic organizations (e.g., of network
Usually teams have highly cohesive plain type) are regulated by contracts. Usually, contracts
structures with participants selected from different specify legal relationships between parties that
organizational departments based on knowledge, explicitly define their rights and responsibilities
skills and experience required for the processes with respect to some processes (e.g., production,
assigned to these teams. Although many teams supply services). Several organizations may be
implement informal communication and par- involved in the process execution (e.g., supply
ticipative decision making principles (Lansley, chains for product delivery); therefore, it is needed
Sadler, & Webb, 1975), also formal authority to identify particular aspects of responsibility
relations can be found in teams. In particular, in in contracts for such processes. The introduced
some project teams superior-subordinate rela- language may be used for specifying such respon-
tions exist between the team manager and team sibilities and their legal consequences through
members. In this case, whereas responsibility for reward/sanctions mechanisms.
making technological decisions is given to team
members, the responsibility for most managerial
decisions is assigned to the team manager. Then, INtEGrAtION OF AUtHOrIty
the members of such teams, being also members rELAtIONs INtO AN EIs
of some functional departments or groups, have
at least two superiors. In other teams the team In our previous work, a general framework for
manager plays the integrator role and does not formal organizational modeling and analysis is
have formal authority over team members. In this introduced (Popova & Sharpanskykh, 2007c).
case, the responsibility for decisions made by a It comprises several perspectives (or views) on
team lies on all members of the team. Sometimes organizations, similar to the ones defined in the
to strengthen the position of a team manager, s/he Generalized Enterprise Reference Architecture
is given control over some resources (e.g., budgets) and Methodology (GERAM) (Bernus, Nemes,
that can be used, for example, to provide material & Schmidt, 2003), which forms a basis for com-
incentives to the team members. parison of the existing architectures and serves
The principles on which teams are built come as a template for the development of new archi-
close to the characteristics of the organic organiza- tectures. In particular, the performance-oriented
tional form (Scott, 2001). Some of such organiza- view (Popova & Sharpanskykh, 2007b) describes
tions do not have any formal authority structure, organizational goal structures, performance indi-
1204
cators structures, and relations between them. The The first axiom on the authority relations
process-oriented view (Popova & Sharpanskykh, expresses that roles that are responsible for a
2007a) describes task and resource structures, and certain aspect related to some process should be
dynamic flows of control. In the agent-oriented necessarily authorized for this:
view different types of agents with their capabilities
are identified and principles for allocating agents Ax1: ∀r ROLE ∀a: PROCESS ∀aspect: ASPECT
to roles are formulated. Concepts and relations ∀γ: TRACE ∀t: TIME state(γ, t) |= [ respon-
within every view are formally described using sible_for(r, aspect, a) ⇒ authorized_for(r,
dedicated formal predicate-based languages. The aspect, a) ].
views are related to each other by means of sets
of common concepts. The developed framework Another axiom expresses the transitivity of
constitutes a formal basis for an automated EIS. the is_subordinate_of_for relation: r1: ROLE x
To incorporate the authority relations intro- r2: ROLE x a: PROCESS:
duced in this article into this framework, both
syntactic and semantic integration should be Ax2: ∀r1, r2, r3: ROLE ∀a: PROCESS ∀γ, t
performed. The syntactic integration is straight- state(γ, t) |= [ is_subordinate_of_for(r2, r1,
forward as the authority relations are expressed a) ∧ is_subordinate_of_for(r3, r2, a)] ⇒
using the same formal basis (sorted predicate logic) is_subordinate_of_for(r3, r1, a)]
as the framework. Furthermore, the authority
relations are specified on the concepts defined in One more axiom (Ax3) that relates the interac-
the framework (e.g., tasks, processes, resources, tion (communication) structure of an organization
performance indicators). For the semantic inte- with its authority structure based on superior-
gration rules (or axioms) that attach meaning, subordinate relations expresses that there should
define integrity and other types of organization be specified a communication path between each
constraints on the authority relations should be superior role and his/her subordinate(s). Such a
specified. A language for these rules is required path may include intermediate roles from the
to be (1) based on the sorted predicate logic; (2) authority hierarchy and may consist of both in-
expressive enough to represent all aspects of teraction and inter-level links.
the authority relations; (3) executable, to make The following axiom expresses that only roles
constraints (axioms) operational. Furthermore, as that have the responsibility to make managerial
authority relations are closely related to dynamic decision with respect to some process are allowed
flows of control that describe a temporal ordering to authorize other roles for some aspect of this
of processes, a temporal allocation of resources, process:
and so forth, a language should be temporally
expressive. A language that satisfies all these re- Ax4: ∀r1,r2:ROLE ∀a: PROCESS ∀asp: ASPECT
quirements is the temporal trace language (TTL). ∀γ, t state(γ, t) |= [ authorizes_for(r1, r2, asp,
In Sharpanskykh and Treur (2006), it is shown that a) ⇒ is_responsible_for(r1, manage_des,
any TTL formula can be automatically translated a) ].
into executable format that can be implemented
in most commonly used programming languages. In general, rules that describe processes of
In the following, the semantic integration rules authorization, assigning/retracting of responsibili-
and several examples of constraints defined for ties may have many specific conditions. However,
particular organizations are considered. to assign responsibility for some aspect of a pro-
cess a role should necessarily have at least the
1205
responsibility to make managerial decisions and ⇒ ∃r1,r2,r3: ROLE state(γ, t) |= [ is_responsible_

be the superior (with respect to this process) of a for(r1, manage_des, a) ∧ is_responsible_
role, to which the responsibility is assigned. All for(r2, tech_des, a) ∧ is_responsible_for(r3,
other conditions may be optionally specified by execution, a) ].
the designer. Responsibility may be assigned on
a temporal basis. To specify that a responsibility Another example is related to rewards/sanc-
relation holds in all states that correspond to time tions imposed on a role depending on the process
points in the time interval limit, a responsibility execution results. As shown in Popova and Shar-
persistency rule should be defined: panskykh (2007b), performance indicators (PIs)
may be associated with organizational processes
C1: ∀asp: ASPECT ∀r1,r2:ROLE ∀a: PROCESS that represent performance measures of some
∀γ, ∀t1, t2:TIME state(γ, t1) |= is_respon- aspects of the tasks execution. Depending on the
sible_for(r1, asp, a) & state(γ, t2) |= as- PIs values, a company may have regulations to
signs_responsibility_to_for(r1, r2, asp, a) provide/impose some rewards/sanctions for roles
& (t1-t2) < limit (agents) responsible for the corresponding pro-
⇒ state(γ, t1+1) |= is_responsible_for(r1, asp, a). cesses. Although such rules are rarely completely
automated, still an EIS may signal to managers
Using concepts and relations from other or- about situations, in which some rewards/sanctions
ganizational views, more complex constraints can be applied. For example, the system may de-
related to formal authority can be described. For tect and propose a reward granting action to the
example, “the total amount of working hours for manager, when a role has been keeping the values
role r1 should be less than a certain limit”: of some PI(s) related to its process above a certain
threshold for some time period [period_start,
C2: sum([a: PROCESS], case(∃t1 state(γ, t1) |= period_end]. In TTL:
is_responsible_for(r1, execution, a), a.max_dura-
tion, 0)) < limit. C4: ∀γ, t1 t1 ≥ perod_start & t1 ≤ perod_end &
state(γ, t1) |= [ is_responsible_for(r2,
This property can be automatically verified execution, a1) ∧ measures(PI1, a1) ∧ is_sub-
every time when roles are assigned additional ordinate_of_for(r2, r1, a1) ∧ PI1.value >
responsibilities for some processes. This is par- limit ]
ticularly useful in matrix organizations (Scott, ⇒ state(γ, period_end+1) |= grants_reward_to_
2001), in which roles often combine functions for(r1, bonus_5_procent, r2, excellent_per-
related to different organizational formations formance_of_a1).
(departments, teams), and, as a result, their actual
workload may not be directly visible. The axioms Ax1-Ax4 can be checked on a
Another constraint expresses that when the specification of organizational formal authority re-
execution of a process begins, for each of the lations. To this end, simple verification algorithms
basic aspects for this process (execution, tech_des, have been implemented. Whereas the constraints
and manage_des) a responsible role should be C1-C4 and similar to them need to be checked
assigned: on actual executions of organizational scenarios
(e.g., traces obtained from an EIS). An automated
C3: ∀a: PROCESS ∀γ, t state(γ, t) |= process_ method that enables such types of analysis is
started(a) described in (Popova & Sharpanskykh, 2007a).
1206
Furthermore, the identified rules can be used enterprise architectures that comply with the
to determine for each user of an EIS relevant to requirements of the GERAM (e.g., CIMOSA)
him/her information and a set of allowed actions based on which modern EISs are built. However,
that are in line with his/her (current) responsibili- to enable semantic integration of the author-
ties defined in the system. Moreover, (possible) ity concepts, an EIS is required to have formal
outcomes of each action of the user can be evalu- foundations, which are missing in many existing
ated on a set of (interdependent) authority-related enterprise architectures and systems.
and other organizational constraints, and based In the future it will be investigated how the
on this evaluation the action is either allowed or proposed authority modeling framework can be
prohibited. applied for the development of automated sup-
port for a separation task (i.e., maintaining a safe
distance between aircrafts in flight) in the area of
DIscUssION air traffic control. Originally this task was man-
aged by land controllers, who provided separation
This article makes the first step towards defining instructions for pilots. With the increase of air
the formal operational semantics for power-related traffic, the workload of controllers rose also. To
concepts (such as authority, influence, control), facilitate the controllers’ work, it was proposed to
which are usually vaguely described in organiza- (partially) delegate the separation task to pilots.
tion theory. In particular, this article addresses This proposal found supporters and opponents
formal authority, different aspects of which are both among controllers and pilots. The resistance
made operational by defining a dedicated predicate to a large extent was (is) caused by ambiguity and
logic-based language. It is illustrated how the vagueness of issues related to power mechanisms.
introduced relations can be used for representing Such questions as “whom to blame when an in-
authority structures of organizations of different cident/accident occurs?”, “which part of the task
types. may be delegated?”, “under which environmental
Modern enterprises can be described along dif- conditions the task can be delegated?” still remain
ferent dimensions/views, that is, human-oriented, open. By applying the framework proposed in
process-oriented and technology-oriented. How- this article, one can precisely define responsibili-
ever, most of the existing EISs focus particularly ties of both controllers and pilots and conditions
on the process-oriented view. An extension of the under which the responsibility can be assigned/
models on which EISs are built with concepts and retracted. Notice that these conditions may include
relations defined within the human-oriented view relations from different views on organizations
allows conceptualizing more static and dynamic (e.g., “current workload is less than x,” “has abil-
aspects of organizational reality, thus, resulting in ity a”), which allows a great expressive power in
more feasible enterprise models. Among the rela- defining constraints.
tions between human actors, authority deserves
special attention, as it is formally regulated and
may exert a (significant) influence on the execu- rEFErENcEs
tion of enterprise processes. This article illustrates
how the concepts and relations of authority can Bacharach, S., & Aiken, M. (1977). Communica-
be formally related to other organizational views, tion in administrative bureaucracies. Academy of
thus resulting into an expressive and versatile Management Journal, 18, 365-377.
enterprise model. The introduced authority rela-
tions may be also incorporated into other existing
1207
Bernus, P., Nemes, L., & Schmidt, G. (2003). Peabody, R. (1964). Organizational authority:
Handbook on enterprise architecture. Berlin: Superior-subordinate relationships in three public
Springer-Verlag. service organizations. New York, NY: Atherton
Press.
Blau, P., & Scott, W. (1962). Formal organizations.
Chandler Publishing. Popova, V., & Sharpanskykh, A. (2007a). Process-
oriented organization modeling and analysis. In:
CIMOSA. (1993). CIMOSA—open system archi-
J. Augusto, J. Barjis, U. Ultes-Nitsche (Eds.),
tecture for CIM. ESPRIT consortium AMICE.
Proceedings of the 5th International Workshop on
Berlin: Springer-Verlag.
Modelling, Simulation, Verification and Valida-
Clegg, S. (1989). Frameworks of power. London: tion of Enterprise Information Systems (MSVVEIS
Sage. 2007)(pp. 114-126). INSTICC Press.
Friedrich, C. (Ed.). (1958). Authority. Cambridge, Popova, V., & Sharpanskykh, A. (2007b). Model-
MA: Harvard University Press. ling organizational performance indicators. In: F.
Barros, et al. (Eds.), Proceedings of the Interna-
Gulick, L., & Urwick, L. (Eds.). (1937). Papers
tional Modeling and Simulation Multi-conference
on the science of administration. New York, NY:
IMSM’07 (pp. 165-170). SCS Press.
Institute of Public Administration.
Popova, V., & Sharpanskykh, A. (2007c). A
Hickson, D., Hinings, C., Lee, C., Schneck, R.,
formal framework for modeling and analysis of
& Pennings, J. (1971). A strategic contingency
organizations. In: J. Ralyte, S. Brinkkemper, B.
theory of intra-organizational power. Administra-
Henderson-Sellers (Eds.), Proceedings of the Situ-
tive Science Quarterly, 16, 216-229.
ational Method Engineering Conference, ME’07
Jonker, C., & Treur, J. (2003). A temporal-inter- (pp. 343-359). Berlin: Springer-Verlag.
activist perspective on the dynamics of mental
Scheer, A.-W., & Nuettgens. M. (2000). ARIS
states. Cognitive Systems Research Journal, 4,
architecture and reference models for business
137-155.
process management. In: W. van der Aalst, et al.
Lansley, P., Sadler, P., & Webb, T. (1975). Organi- (Eds.), LNCS 1806, Berlin, 366-389
zation structure, management style and company
Scott, W. (2001). Institutions and organizations.
performance. London: Omega.
Thousand Oaks, CA: Sage Publications.
Manzano, M. (1996). Extensions of first order
Sharpanskykh, A., & Treur, J. (2006) Verifying
logic. Cambridge, UK: Cambridge University
inter-level relations within multi-agent systems.
Press.
Proceedings of the 17th European Conference on
Mintzberg, H. (1979). The structuring of orga- AI, ECAI’06 (pp. 290-294). IOS Press.
nizations. Englewood Cliffs. NJ: Prentice Hall.
Simon, H. (1957). Administrative behavior (2nd
Parsons, T. (1947). The institutionalization of ed.). New York, NY: Macmillan Co.
authority. In: M. Weber, The theory of social and
Weber, M. (1958). From Max Weber: Essays in
economic organization. New York, NY: Oxford
sociology. In: H. Gerth, & C. Mills (Eds.). New
University Press.
York, NY: Oxford University Press.
for Reliable and Scalable Operations, edited by Nijaz Bajgoric, pp. 252-264, copyright 2010 by Information Science Refer-
ence (an imprint of IGI Global).
1208
1209
Chapter 5.3
Enterprise Systems,
Control and Drift
Ioannis Ignatiadis
University of Bath, UK
Joe Nandhakumar
University of Warwick, UK
AbstrAct that control and drift arising from the use of an

Enterprise System are outcomes of the processes
Enterprise Systems are widespread in current of embedding and disembedding human actions,
organizations and seen as integrating organiza- which are afforded (enabled or constrained) by
tional procedures across functional divisions. An the Enterprise System.
Enterprise System, once installed, seems to enable
or constrain certain actions by users, which have
an impact on organizational operations. Those INtrODUctION
actions may result in increased organizational
control, or may lead to organizational drift. The Implementation of an Enterprise System (also
processes that give rise to such outcomes are known as Enterprise Resource Planning-ERP
investigated in this chapter, which is based on a System) in an organization may have profound
field study of five companies. By drawing on the impact on organizational processes (Boudreau
theoretical concepts of human and machine agen- & Robey, 1999; Koch, 2001; Martin & Cheung,
cies, as well as the embedding and disembedding 2000; Schrnederjans & Kim, 2003; Siriginidi,
of information in the system, this chapter argues 2000), as well as on information flow and trans-
parency (Bernroider & Koch, 1999; Besson &
DOI: 10.4018/978-1-60566-146-9.ch016 Rowe, 2001; Gattiker & Goodhue, 2004; Legare,
Enterprise Systems, Control and Drift
2002; Markus & Tanis, 2000; Newell et al., 2003; looked at internal operations, and the use of such
Shang & Seddon, 2000). Much of the research in systems referred only to internal actors, without
Enterprise Systems however, is concerned with examining external linkages, which was beyond
the implementation process and providing insights the purposes of this research.
into success factors of Enterprise Systems imple- The rest of the chapter is structured as follows:
mentation (e.g. Akkermans & van Helden, 2002; in the following section, we review the relevant
Al-Mashari & Al-Mudimigh, 2003; Bingi et al., literature on Information Systems, control and
1999; Holland & Light, 1999; Hong & Kim, 2002; drift, as well as human agency, which are top-
Nah et al., 2001; Shanks et al., 2000; Somers & ics central to our research. We then present our
Nelson, 2001). Only a few studies investigate theoretical foundations, in which we frame our
issues relating to the post implementation of ES analysis and discussion. Our research approach
(e.g. Elmes et al., 2005). Hence we have limited is then outlined, followed by a description of the
understanding of issues affecting the use of Enter- companies that participated in this research. We
prise Systems in organizations and their potential follow this with an analysis of the data gathered
for organizational impact. from the companies, across the dimensions of
This chapter therefore concentrates on the ac- control and drift. We then discuss our findings
tual use of an Enterprise System, post-implemen- and present our conceptualisation of Enterprise
tation. It examines the impact of actions performed System use, and conclude with some theoretical
by humans (users), or a machine (the Enterprise and practical implications of our research.
System), on control and drift within an organiza-
tion. We propose a theoretical conceptualisation
to describe the impact of those actions by drawing LItErAtUrE rEVIEW
on a field study of five companies that have an
Enterprise Resource Planning System installed. Enterprise systems,
The significance of this research is twofold. First, control and Drift
our conceptualisation developed in this chapter
enhances the understanding of the processes that The link of Information Systems with organiza-
result in organizational control (or drift) through tional control has been investigated by a variety
the use of an Enterprise System. Second, our re- of scholars in the field (e.g. Coombs et al., 1992;
sults also pinpoint issues of practical interest to Duane & Finnegan, 2003; Malone, 1997; Tang et
companies that are using (or thinking of installing) al., 2000). Many point to the paradox that while
an Enterprise System. Information Systems can empower employees
Although ERP systems were originally with increased decision-making capabilities, at the
designed to be used within an organization, in same time they can serve to increase control within
the last years they have evolved considerably the organization (e.g. Bloomfield & Coombs,
to include or link with external functionalities 1992; Bloomfield et al., 1994; Orlikowski, 1991).
such as Customer Relationship Management Although control in a general information
(CRM), Supply Chain Management (SCM) and systems setting has been examined to a large
e-business (B2B and B2C). The current trend is extent, the number of studies in an Enterprise
also to repackage ERP systems as a collection Systems setting in particular is still quite limited.
of interoperable modules with standards-based What distinguishes Enterprise Systems from other
interfaces, in accordance with the mandates of Information Systems is their scale, complexity,
Service-Oriented Architectures. The examina- and potential for organizational impact. Because
tion of ERP systems in this chapter however only of this, they deserve special attention with regards
1210
to the issue of control. From the limited number more visible to others, who can then more easily
of studies in this area, the characteristic ones are exercise control over them.
those by Hanseth et al. (2001), Sia et al. (2002), Regarding the issue of drift, Ciborra (2002)
and Elmes et al. (2005). The main findings of defines drift as the processes of matching between
these three studies are outlined below. situated human interventions of use and open
Hanseth et al. (2001) claim that Enterprise Sys- technology. Technology drifting can be the result
tems (such as ERP systems), with their emphasis of passive resistance, learning-by-doing, sabotage,
on standardization, streamlining, and integrating radical shifts in conditions, or plain serendipity In
business processes, are an ideal control technol- addition, Ciborra (2000) mentions the case when
ogy. However, they point to a surprising result: control has to decrease, when it is associated with
That implementing an ERP system over a global the power to bring to life sophisticated and evolv-
organization in order to enhance control may as ing infrastructures. Although Ciborra concentrates
well have the opposite effect, i.e. reduce control. mainly on technology drift, this chapter is con-
This can be explained with the ubiquitous nature cerned with drift at the organizational level, which
of side effects. In that sense, the more integrated is implied by decrease of organizational control.
the system is, the faster and farther side effects In particular regarding the issue of drift, van
have an impact, and the bigger their consequences. Fenema and van Baalen (2005) have looked
Sia et al. (2002) have examined the issues of into strategies for dealing with drift during the
empowerment and panoptic control of ERP sys- implementation of ERP systems. They distinguish
tems. They summarize the panoptic control aspect between three such strategies, from which they
of ERPs in three dimensions: comprehensive argue the third strategy (drift containment) is the
system tracking capability, enhanced visibility most realistic in ERP implementation projects:
to management, and enhanced visibility to peers
(through workflow dependency and data inter- • Control strategy aims at eliminating drift-
dependency). The findings by Sia et al. (2002) ing and risk.
indicate that although an ERP implementation has • Incremental strategy considers drifting to
the potential for both employee empowerment and be a normal part of technology implemen-
managerial control, managerial power seems to tations. In this case “bricolage” is used to
be perpetuated through an ERP implementation. adapt technology to its context.
Elmes et al. (2005) have identified two seem- • Drift containment recognizes the inevita-
ingly contradictory theoretical concepts in an ble drifting in technology implementations
Enterprise System: reflective conformity and and the fact that drifting may even contrib-
panoptic empowerment. Reflective conformity ute to the stabilisation of technology. The
refers to the way that the integrated nature of question is then how to balance control and
the Enterprise System leads to greater employee drift and use drift as a source of stabilisa-
discipline, while at the same time requiring them tion of technology projects.
to be reflective in order to achieve organizational
benefits from the system. Panoptic empowerment Although the above studies looked (amongst
describes the greater visibility of information, others) at either the issue of control or drift (but
which is provided by the shared database of the not both at the same time), Nandhakumar et al.
Enterprise System. This empowers employees to (2005) have looked into the contextual forces of
do their work more effectively and efficiently, but an ERP implementation, and how those influence
at the same time makes their work in the system control and drift during the implementation pro-
cess. These contextual forces were interrelated,
1211
and referred to the affordance of the technology, (2004) for example argues that ERP systems help
as well as the social structure, practices and norms, shape human agency and institute patterns of ac-
either within the organization or external. The tion and communication in organizations. They
analysis by Nandhakumar et al. (2005) was based accomplish this by delineating the paths along
on examining managers’ intentions, the power and which human agency should take place. This
cultural context within the organization, as well is afforded by the dissection of organizational
as the affordances of the technology. activities in discreet terms, and the provision of
Control was seen by Nandhakumar et al. to the procedural sequences for the execution of
be an outcome of managerial intentions regard- particular tasks. In this sense, ERP systems are
ing the trajectory of implementation of the ERP mostly concerned with streamlining, control and
system. This depended on both system affordances standardisation of organizational operations. In
and social structure. (Technological) drift was so doing, ERP systems enable the construction of
then seen by Nandhakumar et al. (2005) to occur accountable and governable patterns of behaviour
from the “organizational members’ planned and in organizations.
unplanned actions in response to both previous Similarly, Boudreau & Robey (2005) have
technology and organizational properties they pointed to the fact that when looking at organiza-
have enacted in the past” (p. 239). Unintended tional change arising from the use of IT, an agency
consequences of the implementation would perspective may mean limited possibilities for
then mean that the technology would drift from radical IT-induced change. An agency perspec-
the planned implementation outcomes. Control tive of IT in this case takes the position that IT
and drift during the implementation of the ERP is socially constructed and open to a variety of
system in this case were interrelated, and would social meanings and potential uses. Boudreau &
be operating in continuous cycles, in response to Robey argue that certain technologies allow for
contextual forces shaping the actual implementa- a greater degree of human agency and others to
tion of the ERP system. a lesser degree. Their views agree with those of
Although Nandhakumar et al. directly ac- Orlikowski (2000), who acknowledges that while
knowledge the influence of users in accepting or users can and do use technologies as they were
rejecting the ERP system, their study mainly ex- designed, they also can and do circumvent the
amined managerial as opposed to user intentions, intended uses of technologies, either by ignor-
and in the implementation, as opposed to the use ing certain properties, working around them, or
stage of an ERP system. This chapter therefore inventing new ones.
complements the viewpoint by Nandhakumar et The research by Boudreau & Robey looked
al. by arguing that the way users use the system at ERP systems, which are seen as inflexible
according to the affordances of the technology software packages constraining user-inspired
also has a large part to play on impacting control action (human agency). Their results, however,
and drift within an organization. indicate that although ERP systems are seen as
rigid control mechanisms, there is still scope for
Enterprise system Use human agency to take place within such systems,
and Human Agency contradicting Kallinikos (2004). The research by
Boudreau & Robey indicated that technical system
One important strand of Enterprise System constraints on human agency can be overcome
research which is also relevant to the current through a process of initial inertia (rejection of
chapter is the linkage of Enterprise System (e.g. the system), improvised learning as a result of
ERP) use with the agency of humans. Kallinikos pressure to use the system, and finally reinven-
1212
tion of usages of the system, to match their own that are inscribed in the Enterprise System. The
previous experiences and background. symbolic token is then the information that is held
and represented within the Enterprise System. The
social relations that are being disembedded and
tHEOrEtIcAL FOUNDAtIONs reembedded derive from the actions of users in
the system. These actions are disembedded from
This section presents our theoretical foundations the local context where they are carried out, to
for the development of our conceptualisation in time-space stretches of the company’s operations.
the chapter. These theoretical foundations are the Those actions are consequently reembedded, by
concepts of embedding-disembedding and agen- impacting on, and being appropriated by, other
cies (human and machine). These concepts were users of the system.
chosen as they were deemed important in explain-
ing the processes that lead to organizational control Human and Machine Agency
and drift. The theoretical foundations presented
below together with the concepts of control and Giddens (1984) defines agency is the “capabil-
drift that were described in the previous section, ity to make a difference, that is to exercise some
will inform our analysis and discussion. sort of power” (p. 139). In other words, agency is
synonymous to the carrying out (or intentionally
Embedding-Disembedding not carrying out) of an action. With regards to
actions in an Information Systems setting, Rose
Giddens (1990) defines disembedding as “the lift- et al. (2003) have questioned the relationship
ing out of social relations from local contexts of between the social and technical aspects of IS,
interaction and their restructuring across indefinite in other words, how do social systems act upon
spans of time-space” (p. 21). Conversely, embed- technology, and vice versa. Rose & Jones (2004)
ding (or reembedding) is according to Giddens, have drawn on both Actor-Network Theory and
“the reappropriation or recasting of disembedded Structuration Theory, as well as Pickering (1995),
social relations so as to pin them down (however to develop a model called the “Double Dance of
partially or transitorily) to local conditions of time Agency”. In this model, the distinction is made
and place” (pp. 79-80). between human agency and machine agency, but
For Giddens (1990) there are two types of dis- the two are interwoven and affect each other.
embedding mechanisms: symbolic tokens and ex- Rose & Truex (2000) have proposed to give
pert systems (not expert systems in an information an understanding of machine agency as perceived
systems sense). Although Giddens concentrates autonomy. Machine agency is then the emergent
mainly on money, symbolic tokens in general are property of the development process and becomes
media of exchange that can be circulated without embedded in the completed machine. However,
regard to specific characteristics of the people Nandhakumar et al. (2005) in their study of an
or groups that handle them. Expert systems are ERP implementation, draw from Gibson (1979)
then organizations of technical accomplishment and Norman (1988) to view machine agency as
or professional expertise that make a significant affordance. They also draw from Giddens (1984),
contribution to the material and social environ- who attributes intentionality to human agency.
ment in which we live. In the context of this chapter, we will similarly
In the current context of an Enterprise System, assume that machine agency is characterised by
it can be argued that the expert system as defined affordance, whereas human agency is character-
by Giddens consists of the rules and procedures
1213
ised by intentionality. The next section presents pret the segment of text. Thirdly, pattern codes
the research approach employed in our research. are even more inferential and explanatory.
In the current research, coding was done
with the aid of the qualitative analysis software
rEsEArcH APPrOAcH NVivo. Within NVivo, descriptive coding was
done through a first pass analysis of the interview
Our research approach is interpretive case study transcripts, where the relevant text was exam-
(Walsham, 1993). In interpretivism, the reality is ined, and portions of it (sentences, paragraphs,
socially constructed by human agents (Walsham, or sections) were assigned a code according
1995). Interpretive studies reject the notion of to the phenomenon that they were describing.
an objective or factual account of events and Some pattern coding also occurred at this stage,
situations, seeking instead a relativistic, albeit as codes were grouped into categories at various
shared, understanding of phenomena (Orlikowski levels that linked those codes together. This cod-
& Baroudi, 1991). ing was more akin to a grounded theory approach
In the current research, five companies (Glaser & Strauss, 1967), where codes evolved
(NGlobe, TechniCom, SecSys, FCom and Trans- directly from the data. Memos were also kept in
Com – all pseudonyms) have been examined with NVivo about issues of interest that emerged, and
regards to the use of their Enterprise System. The reflections from the part of the researcher on the
research took place between October 2004 and data examined. This prompted a second pass of
August 2005, and involved formal semi-structured coding, where the coding was more theory-driven,
(face-to-face mostly with a few telephone) inter- according to identified themes in the first pass,
views with company staff (office employees and and literature that could support the concepts from
managers). It also involved informal conversations the first pass of coding (e.g. relevant literature on
with them, as well as observing user interactions control, drift, human action, machine action, inten-
with the system. Interviews were mostly with one tionality, affordance, embedding, disembedding).
person, although a couple of them involved two The second pass of coding was more interpretive,
persons, and lasted anything between 40 minutes in that segments of text were interpreted accord-
and 2 hours, with an average of 1 hour per inter- ing to the chosen literature. However, data were
view. All of the interviews (with the exception of not forced into predefined categories, as codes
one due to personal sensitivities) were recorded emerged not only from the literature, but also as a
and transcribed verbatim. The table below shows result of the first pass coding, which was grounded
the interviews with members of the 5 companies on the interview data. In addition, pattern coding
researched.(Table 1) also occurred at this stage, as several codes were
Analysis was carried out using a qualitative grouped into higher-order patterns.
data analysis approach proposed by Miles & The analysis below is structured in terms of
Huberman (1994). They distinguish between three the two concepts of control and drift, also taking
types of codes: Descriptive codes entail little into account human and machine actions, and
interpretation, but attribute a class of the phenom- the respective intentionalities and affordances.
ena to a segment of text (such as interview tran- Indicative quotes are also presented. Discussion
script). The same segment could also be coded of the results then incorporates the concepts of
with an interpretive code, where the researcher embedding and disembedding. The next section
uses some of his or her knowledge about the briefly describes the background of the compa-
background of the phenomenon in order to inter- nies participating in the research, followed by the
analysis and discussion of our results.
1214
Table 1. Illustration of the interviews carried out
Area Positions Interviewed No. of Interviews

Company: NGlobe
- Business Centre IS Manager
IS Department 2
- General IS Manager
- Accounts Payable Employee
Finance Department - Cash Management and Treasury Process Director 3
- Record to Report Process Director
Company: TechniCom
- Supply Chain Manager 1
Procurement 2
- Supply Chain Manager 2
Company: SecSys
Procurement - Purchasing Manager 2
Company: FCom
IT Department - IT Manager 2
Company: TransCom
- Sales Facilitator
Sales 3
- Commercial Assistant
- Billing Clerk
- Accounting Reports Manager
Finance - Accounts Payable Clerk 6
- Assistant Accountant
- Assistant Finance Manager
- Maintenance Policy Leader
- Head of Production
- SAP Facilitator
- Production Planner
Service Management 12
- Shift Planning Coordinator
- Flow Repairable Controller
- Reliability Group Leader
- Abnormal Work Manager
- Business Improvement Coordinator
Warehouse and Distribution - Logistics Director 5
- Inventory Planner
- Materials Controller 1
Materials Management - Materials Controller 2 5
- Materials Planner
Purchasing - Purchasing Manager 1
- IT Manager
IT Management 4
- Global Information Systems Director
Total Interviews 47
cOMPANy DEscrIPtIONs was a criterion when selecting those companies,

in order to be able to observe the impacts of the
The companies that participated in the research Enterprise System after it had been used for a
are described below (with pseudonyms). All of number of years, and not during or immediately
the companies have had an Enterprise System after implementation. The companies come from
(ERP) installed for a minimum of 2 years. This different sectors, have different customer markets,
1215
and use different ERP systems, which adds to the niCom is using the PeopleSoft ERP system. For
generalizability of the results. billing they have had a bespoke system built for
them, which harmonizes all the different billing
transcom systems from all the companies that TechniCom
bought.
TransCom operates in the transport sector, and
has 5 billion Euros in annual sales worldwide. Fcom
TransCom is using the SAP R/3 ERP system,
currently on version 4.5 in most places where it FCom is a vertical retail manufacturer, which
is installed, although there are some later versions means that the bulk of the products that they sell
(4.6) or earlier versions (e.g. 3.1, or even R/2) in in the stores, they also manufacture themselves.
some countries. The system was fully installed in It operates more like a family-run business, rather
January 2002 in the UK. In addition to the UK, SAP than a large corporate public company. FCom is
is currently fully installed in Spain, France, Swe- a £55 million turnover company operating in the
den, Romania, Chile, and the USA. The modules UK, currently having 61 stores and employing
of SAP used are Materials Management, Service around 500 people. FCom has the SAP R/3 system
Management, Finance, Sales & Distribution. installed. The modules used are Payroll, Finance,
Manufacturing, and Sales and Distribution.
NGlobe
secsys
NGlobe has a presence in almost 200 countries,
and generates most of its revenue by selling fi- SecSys is owned by an American company, and
nancial information. NGlobe has the Oracle 11i operates in the UK under the service support
ERP system installed, used for the modules of sector. There are around 25 offices of SecSys in
Finance (accounts receivable, payable, general the UK. SecSys has the SAP (R/3, version 3.1h)
ledger, fixed assets, and purchasing) and HR. ERP system, which was installed in November
In addition, the Oracle’s learning module is also 1998 in order to overcome the millennium is-
installed. The design of the ERP system took place sue. The modules of SAP that are currently used
between 2000-2001, and NGlobe first went live are Finance, Logistics, Materials Management,
in 2001, with the whole implementation finishing Inventory Management, Purchasing, Sales and
by the end of 2003. NGlobe was one of the first Distribution. The company’s intention within the
implementations of one global instance of Oracle. next 2 years is to update their ERP system.
This was enabled by the fact that NGlobe is a very
homogenous company.
cAsE EVIDENcE AND ANALysIs
technicom
The field study data from the companies were
TechniCom is a company in the technological analyzed based on the categories of control and
sector. It has recently been restructured, and drift, which are central to our research. The analy-
over a 4-year period TechniCom has bought 32 sis also drew on the notions of the intentionality
companies, all with different bespoke systems. of human agency (users), and the affordances of
TechniCom is using Oracle version 11 for the the machine agency of the Enterprise System.
financials part, mainly for accounts payable and Human agency in the analysis below is implied
accounts receivable. For their HR function Tech- by words such as ”intention” and ”intentionality”.
1216
Machine agency is implied by words such as ”af- the ITUK team for being unresponsive to their
ford”, ”enable”, ”constrain”. Only data from the needs, slow and inefficient. Some times there were
companies that are relevant to the main theme of tensions between users and the ITUK team, as
this research are presented here. well as between the ITUK team and ERPGlobal.
Before examining the actual use of the En- In addition, the ITUK team claimed that they
terprise System, our research data indicated that were also understaffed, with the result that they
there are also factors which influence its use, and could not respond very quickly to user requests.
which are discussed next. The examination of This impacted training as well, and one common
these factors comes mostly from TransCom, due complaint from most users of the system was that
to the larger number of interviews carried out in they had not received enough training, and that
that company. either they did not understand how the system
worked, or could not use it to its full potential.
Factors Impacting As a result, the ITUK team who was responsible
Enterprise system Use for training was seen as quite irresponsible and
unresponsive with regards to the training needs
Cultural and Organizational Context of the users.
There were also cultural reasons which could
In TransCom there was a global group responsible influence the use of the system. For example, the
for the implementation and maintenance of the implementation of SAP gave TransCom the oppor-
ERP system worldwide, named ERPGlobal. They tunity to record and control time, and how long it
interfaced with the UK via a local country group actually took to do a job on a train. Nevertheless,
named ITUK. Within the sites examined in this the employees at the Manchester depot refused
research, there were some negative views about to do this when SAP was initially implemented
the ITUK team, as well as about ERPGlobal. As at their site. On closer examination however, it
ERPGlobal wanted to keep the configuration and was revealed that they would refuse to do this
use of the ERP system as standard worldwide, it anyway, whether there was the SAP system or
was very hesitant in carrying out updates to it, not. The intention of not recording time was a
unless those would affect the majority of coun- strong cultural aspect, but it was also local to the
tries where the system was installed. Even in specific area. This refusal to record working times
cases where updates were agreed by ERPGlobal at the Manchester depot was in fact mentioned to
however, those took a very long time to imple- be tied to contracts of employment and the influ-
ment according to the users, and this was viewed ence of trade unions.
negatively.
However, as most users in the UK interfaced Resistance
with the ITUK team directly, and not with ER-
PGlobal, most of the negative criticisms were In addition, when the SAP system was installed in
directed towards the ITUK team. This tended to TransCom, there was resistance to the system at
put the ITUK team in an awkward position, as various levels. Part of middle managers and users
they needed sanctioning from ERPGlobal to carry did not use the system fully. In order to overcome
out user requests, but ERPGlobal was reluctant to this, initial training regarding the system was car-
give its consent in many matters in order to avoid ried out, in order for the employees to understand
deviation from standards regarding the configura- what the system could offer. However, there was
tion and use of SAP. As a result, many users who still resistance from middle managers to fully
did not have interaction with ERPGlobal blamed use SAP as a management tool. SAP was seen
1217
by many managers in TransCom as a financial As a result, the business processes were written
overhead, and consequently there was not a lot around the system, in order to make the system
of enthusiasm in supporting it. work. The system was conceived and implemented
This also had on impact on the users of the by a consultancy company, who looked at what
system. Resistance was evident there, also because the business strategy was, and the processes that
users were used to the old systems installed in TransCom required. They then tailored SAP to
TransCom, some of which were still operational. suit the business needs of the company at the time
In that sense, there was resistance of SAP, also SAP was implemented, in terms of for example
fuelled by the fact that many of the users were the order processes, purchasing, inventory, ware-
considered as technophobes, and were not used house, and logistics.
to computing systems. The main complaints As a parts business did not exist before, people
about the system were then that the system was within TransCom did not know in detail how the
difficult to use, was not user friendly, and users business should be supported by the ERP system.
did not understand what all the fields and text in The consultancy company therefore implemented
SAP were used for. To this lack of knowledge and their own ideas to a large degree, resulting in a
apprehension of the system also contributed the system that was according to the views of many
general lack of proper training. However, this type interviewees very inadequate. The consultancy
of resistance was considered normal, and seen to company was thought to be responsible for this,
be overcome once the users felt more confident having implemented the system:
with the system:
They [the consultancy company] thought they
That’s just the normal resistance to a new system, knew how to implement a spare parts management
because a lot of the people had other systems, system in SAP, we did not have the expertise here
which they knew and they were confident, and in-house, which meant that they were always able
they knew what they could get out of them. And to convince the people they were dealing with what
so it’s just this transition to another system that the right answers were. And it was a very poor
they’ve got to learn and understand. But I’m sure implementation. (Logistics Director, TransCom)
once they can understand it, once they can use it,
they’ll be quite happy with it. (Reliability Group
Leader, TransCom) Power Differentials
During the implementation of SAP some em-

Implementation History ployees of TransCom were released in order to
participate in the implementation and contribute
The implementation of SAP was also the impetus their business experience towards the develop-
for creating a new business unit within TransCom, ment of the system. Many of these people did
whose aim was to stock and supply the depots not have any proper training at all in SAP, but
in the UK (and especially the West Coast area) got to know it from working together with the
with spare parts for trains. This was important, as consultants. Although those TransCom employ-
TransCom did not have a proper parts business ees were released back to the business when the
to deal with spare parts for trains, and therefore implementation of SAP finished, their experience
it also did not have the systems to manage the of the system gained during its implementation
spare parts. made them become perceived as SAP experts by
other users of the system:
1218
And whilst the consultants were here for 3 months Organizational control from
helping other people load their data, I sat with Enterprise system Use
the consultants, learning the system, saying, right,
if I do this, how does that happen, how do I do The case analyses indicated that amongst the
this, and things like that. And just general ques- managerial intentions for installing an Enterprise
tions, saying, right, there’s more information to System, and the purpose of its configuration,
get. One thing led to another, and I became one was to increase control of the company’s opera-
of the, sort of, SAP experts within the business. tions. Control was implemented with a variety
(Business Improvement Coordinator, TransCom) of mechanisms in the Enterprise System. Those
were the setting of access controls, the monitoring
As a result of gaining SAP technical knowl- capabilities of the Enterprise System, the rules and
edge, the perceived SAP experts at TransCom procedures (process flow) that were embedded in
became the first point of contact when users had the Enterprise System, and various checks carried
problems with the system, as opposed to logging out by the system.
their problems with the ITUK team, and the ITUK
team then trying to identify what the solution Control from Access Profiles
would be. The help of the perceived SAP experts
was asked as they could go to the users’ desk quite With regards to the access profiles that specified
quickly to see what the problem was, and explain which types of users had access to different types
to them what they were doing wrong. However, the of information in the system, it could happen that
business rules indicated that in case of problems those controls were either too strict, or too lax. In
users should be contacting the ITUK team in the this sense, there were two ways by which those
first place. By bypassing those rules and seeking access controls could be changed: The first was to
the help of the perceived SAP experts instead, the redefine what a particular access level allowed a
power of the latter was increased in the company, person to do, and the second was that an individual
as they could influence the way that the ERP sys- could be given a less or more open type of user
tem was used by other users. This meant that the access. In general however, the system controls
perceived SAP experts could direct the agency of implemented through access profiles were seen
other users in terms of the way they were using to be quite strict at NGlobe, making it difficult
the system. If the directions given to the users to bypass them:
were right, then the intended control from using
the system would be re-enacted, if the directions In terms of bypassing controls, the only ways
given to them were flawed, then it would be pos- to bypass a control are either to be given some-
sible that drift would be propagated, by the end body else’s password, or to do it in cahoots with
users using the system incorrectly. somebody else. (Cash Management and Treasury
Having described some factors that could influ- Process Director, NGlobe)
ence the use of the ERP system, the next section
discusses the various ways that organizational At the time of the interviews there were many
control from the use of the Enterprise System complaints from the users with regards to the set-
could be enacted. Results from all the companies ting of those profiles. Most of those complaints had
examined are presented, although TransCom to do with the limited access to screens and trans-
provided the biggest case study of this research. actions in the system, when access was required,
but was not given due to the incorrect setting of
access levels. As one interviewee mentioned:
1219
That’s quite annoying for me, because I have the In this case the system tagged user actions in the
information that I need to put into SAP, I’ve got system with information about who carried out a
the information of how to do it, but I’m not given transaction. This could be examined if the need
the access to it, because someone has made the arose, and that was deemed beneficial by some
decision that only one person in the company is of the interviewees:
allowed to do it. (Shift Planner, TransCom)
You can see who’s carried out the transaction, and
On the other hand, users also pointed out that you can take corrective action to understand why
although needed access was unnecessarily limited it happened. So, the system allows that visibility,
in many cases, there were also many other cases to see where things have gone wrong, who’s made
where users were allowed to carry out tasks in the what transactions, who’s done what purchasing.
system, which they did not necessarily need to. (Purchasing Manager, SecSys)
This increased access to the system was mainly
due to the inattention that was paid to the correct The monitoring capabilities of the Enterprise
setting of the access profiles. The result was that System also resulted in power differentials, as
the intended controls in the system were seen to people who were allowed to see the actions of other
be very lax in some cases, because people had users in the system, had authority over the latter. In
authorizations to do many things outside their some cases, this could lead to uncertainties about
immediate area. how the information obtained by monitoring user
In some cases users “abused” their increased actions in the system was used. As some users of
access, because it was easier for them to do a the system mentioned:
transaction in the system themselves that they
should not be doing, rather than asking the person I think people can use [monitoring] to their own
that should be carrying out that transaction. As advantage. Sometimes there’s too much informa-
one interviewee mentioned: tion there, if you make a mistake or whatever, it
can be used to the wrong advantage, which I am
For example, if there is something to be posted not too happy about… It’s there; it can be done.
in the material master, because they [users out- That’s the worrying thing (Shift Planning Coor-
side the Materials Management area] have got dinator, TransCom).
authorisation and they’ve got some knowledge of
the material master, they think, right, OK, I’ll do The thing with SAP though, it’s the traceability.
it myself, rather than going to somebody who’s If we were trying to bypass the system, it’s all
got better knowledge, and say, right, OK, can you traceable. You can look into who did that, what
add this so I can carry on with my processing. changes you’ve made, they will come and hit you
(Business Improvement Coordinator, TransCom) if you do try and bypass anything, because that’s
one thing with SAP, it doesn’t lie, does it? It’s got
to put your name against everything, every time.
Control from Monitoring Capabilities (Inventory Planner 1, TransCom)
of the Enterprise System
This apprehension with regards to monitoring
Our interviews indicated that the managerial of user actions in the system led some users to
intention of increased control through the use of believe that anyone with the right access could go
an Enterprise System was also afforded by the and monitor what other users were doing in the
monitoring capabilities of the Enterprise System. system, and this could be used to highlight areas
1220
of inefficiencies with respect to the work carried with their own work. This mechanism of forcing
out by them. This was seen quite negatively by users to work in a certain way was mentioned by
those users, who considered the system to be the many managers, characteristically:
means of giving people the opportunity to spy on
other people’s work, and then used for blaming ERPs force you to undertake business in a certain
them if something went wrong. way. And when I say force you, I mean the fact that
This apprehension of the system was also you have to put a workflow in, to determine that
confirmed in the case of recording hours for a par- these are the activities that need to be placed...
ticular job done on a train in TransCom. Although Now you have a workflow, you put a control in.
this required the inputting of individual hours for (Supply Chain Manager 2, TechniCom)
each user that worked on a train, the company was
only interested in the total amount of time it took This mechanism [of ERP process flow] enables
to do a job, rather than how much each individual us to ensure that nothing goes on to the lorries
spent in doing that job. However, users seemed to be delivered, unless it’s also been flagged as
hesitant to follow the recommended procedure having been physically finished within the factory.
of inputting hours worked in the system, as they (IT Manager, FCom)
could not see the business benefit of it, and were
worried about how their data would be used:
Control from System Checks
You have to record the individual’s clock number.
And that in itself is problematic, because you SAP in TransCom was configured to carry out
are identifying a particular individual, and they various checks that would ensure the quality and
may think, oh well, you know, that’s a bit too Big integrity of data in the system. For example, us-
Brother, you’re watching my every movement. ers were constrained from inputting an invoice
But at the end of the day, we’re not, as a busi- twice. This could occur for example when users
ness interested that it takes one man 20 minutes, tried to create an invoice as a copy of the original
but it takes another man 30 minutes to do a job. one. In this case the system would check whether
So, as a business we need to know that it took X an invoice for the particular work and customer
man-hours for a particular job altogether. (SAP already existed, and come up with a warning
Facilitator 2, TransCom) about this. This would stop the user from enter-
ing a duplicate invoice, and was seen as a good
control mechanism in SAP, in terms of minimizing
Control from Enterprise duplication of data.
System’s Process Flow Another example of checks carried out by
the system was if VAT on invoices was entered
Our interviews also indicated that another way (manually) incorrectly, which would result in an
control was exercised through an Enterprise Sys- incorrect total balance in the end. This could hap-
tem was by the rules and procedures afforded by pen more at month ends, when users at finance
such a system. This meant that certain workflows would be busy trying to input everything in the
were carried out in a certain order, and the output system. In this case mistakes would be more
of one stage in the workflow was used as the input possible than other periods, and so the relative
in the next stage. Therefore, the Enterprise System financial journals would not balance in SAP. The
forced users in one department to complete their checking of the balance across various ledgers was
work before users in another department could start done automatically by the clicking of a button,
1221
and if there were any mistakes SAP would come Organizational Drift from
up with a warning saying that the changes could Enterprise system Use
not be posted in the system because the relative
journals would not balance. Other smaller-scale System Configuration
examples of checks carried out by the system also
included the format of the fields entered (e.g. for Based on our analysis, it was evident that the actual
numbers, dates, etc), the range of the values the configuration of the system was quite important in
field could accept, etc. ensuring that the required controls were properly
Although the system provided those checks implemented. If the controls described in the pre-
to ensure a level of quality of the data that was vious section were not correctly configured, this
input into the system, in other cases the system could result in users taking actions in the system
was not capable of carrying out some checks that that they shouldn’t, and this could have a “ripple”
users deemed necessary. For example, when an effect to other departments. For example, as one
item needed to be returned as broken to the sup- interviewee mentioned, lax system controls al-
plier, it needed to be recorded into the system as lowed individuals to carry out transactions in the
such. However, there were no checks that the item system that they were not allowed to, and this had
was not accidentally booked into the good stock, an impact on various departments of the company:
although it was declared as broken in the system.
The users in the workshops in this case deemed it Somebody from the other sites went and built
necessary to have a warning message coming up some transaction on my site the other day, which
that would inform them that they were trying to messed up my stock basically… [The impact of
book broken parts into good stock. However, the this transaction would be] financial, because
system as it was did not carry out those checks, the figures wouldn’t have matched. (Materials
and would let the users carry on. This was actu- Controller 1, TransCom)
ally an example of drift that could occur instead
of control due to required checks not being sup- However, the system could be intentionally
ported by the enterprise system. configured to allow more lax controls, in order to
In general however the four mechanisms de- cater for business necessities, for example:
scribed above (access profiles, monitoring capa-
bilities, process flow, and system checks) aided the There are benefits to being able to do it, like for
managerial intention to enhance control over the example, if say, the person who places the order
company’s operations. The control mechanisms in another site is off sick, and there’s only say,
then applied on the users of the system and the somebody here who can place the order, then
way they carried out their work. However, while that means that he can carry out the transaction.
users worked on the system, it was possible that (Materials Controller 1, TransCom)
some degree of drift could occur. Our analysis
indicated that this might be due to the configuration In other cases, the affordance of the system
of the system, which could allow the omission of could force to choose a configuration of the system
important information, as well as possibly allowing (e.g. setting of access profiles) amongst a range
the bypassing or workaround of controls by users; of limited possibilities, which did not accurately
the use of systems outside the Enterprise System reflect the needs of the company. In this case, drift
was also identified as conducive to drift. Each of could occur if the intention to use the system in
these factors will be examined in turn below, with other than the prescribed ways was there. As one
relevant evidence from the companies quoted. interviewee mentioned:
1222
Say for instance with the guys out the back who could be imposed with the use of usernames and
change bin locations, that means they’ve got ac- passwords, if those were shared by the employees
cess to the material master. So they can change it then the intended controls lost their meaning:
if they want to. But unfortunately you can’t sepa-
rately set it so that they can only view it. They’ve Everybody has their own login. But if say there
got access to it or not. So if they wanted to, not were 4 or 5 of us on duty today, somebody wouldn’t
that they would, but they have access to change come to it and log on, we’d use whatever was in,
anything. (Materials Controller 2, TransCom) just for speed. I suppose really, if I walk away
from it, I should log off, and the next person who
The way that the system was configured, ac- comes would log in. But it just takes time to keep
cording to the business necessities and what was logging off and logging in when you’re busy.
afforded by it, could result in users using it in (Materials Planner, TransCom)
other than the prescribed ways, by working around
intended controls, or not inputting important in- Using one log-in for everybody in the workshop
formation in the system. Those two reasons for essentially meant that the intended controls in the
drift are now presented below. system were bypassed by the users. If a generic
logon approach was followed, then it would be
Bypassing or Workaround of Controls difficult to tell who was or was not using the
system. It would also be impossible if the need
Our analysis indicated that organizational drift arose to identify in the system which user actually
could arise from users bypassing the system carried out a transaction.
controls, or working around them. The degree In addition, in the implementation of SAP at
to which they could do this was enabled or con- TransCom there existed a “source list”, which
strained by the properties of the Enterprise Sys- contained the suppliers from which the company
tem. It also depended on how the access profiles could buy items required for the maintenance and
in the system were configured. For example, as repairing of trains. The source list existing in the
one interviewee mentioned, although access to system was created and maintained by the system
information on one screen was not allowed, the administrators at ITUK, and a block was put in
same information could be obtained by accessing the system to disallow other users from amend-
an alternative screen: ing it, so that only authorised suppliers identified
in the source list could be used to buy materials
Now, if I go into Oracle inquiry for Accounts from. However, users identified a way to bypass
Receivable for example, I can see invoice informa- this, by creating a new source list in the system,
tion, but I can’t see credits in there... If I go into rather than using the one created for them by the
Business Objects with my inquiry access, it runs ITUK team. This meant that they could effectively
invoices and credits and it does so pretty happily. include any supplier in their own source list,
So I can get them that way. (Cash Management and buy from any of them, without reference to
and Treasury Process Director, NGlobe) the approved suppliers in the ITUK-maintained
source list.
Although there might not have been intention
in doing so, controls could still be bypassed un- Missing Information
consciously, because it was easier to bypass them
than follow the prescribed ways of working with From the interviews carried out we realised that
the system. For example, although access controls users could choose not to enter important infor-
1223
mation in the system. This might be due to lack information that gets put in is completely rubbish.
of this information, or users resisting the system And that can only be changed by the people using
by not inputting this information. It could also be it, there’s nothing SAP could do to make that any
because users didn’t understand the workings of better, that’s entirely human. (Reliability Group
the system, and the impact that their inaction of Leader, TransCom)
inputting that information had on other depart-
ments and the company as a whole. If important In addition to the issue of the quality of infor-
information was missing in the system, negative mation that was input into the system, the timing
organizational consequences were certain to of inputting that information was also important.
occur. In this case, money might be lost, as one For example, due to the vast number of jobs done
interviewee indicated: on trains in TransCom and the associated number
of service orders, work on a train could start before
There are vehicles coming to our depot, and we the relevant service order was created in SAP. If
don’t know who’s sending them. There’s no contract this mistake was realised later on the same day
set up for them, so the commercial department or the next day, this could be too late, as the train
can’t allocate working hours… We did quite a lot would be repaired by then, and left the depot.
of work for these customers, but we don’t have any This would then mean that it would be difficult
idea who to bill, or how to get the money back, to identify without any records what work was
or how much. (Production Planner, TransCom) done on a train. Even worst, if this mistake was
not found at all and the service order was not cre-
One way to overcome the effects of missing ated, then the train would have been repaired for
information in the system was by introducing free, without charging the customer for it.
mandatory fields, or using the affordances of the
system to carry out checks that this information was Use of External Systems
there. However, although appealing this approach
might seem, there were practical considerations In all of the companies examined, most of their
that would make it unworkable: reporting requirements were carried out by extract-
ing data from the system and inputting them for
If it was for me, I would say, make all the fields manipulation into Excel. In SecSys for example, a
mandatory, and you have to fill them all in. In the key process within the supply chain was customer
real world, you could never do that. There are time order satisfaction, examining where the company
constraints for one, availability of information, was able to satisfy customer orders within the de-
two. So you would have the end user who wouldn’t livery time that the customer wanted. The relevant
use the system. (SAP Facilitator, TransCom) data was extracted from SAP into Excel in this
case, in order to produce a management report
The quality of information that was input into that was presented on a monthly basis to senior
the system was generally recognised as very im- management. Similarly in TransCom, some of the
portant for the correct functioning of the system: reports that were running included for example
the number of open and closed service orders in
I think the big problem we’ve got with SAP at the a given period, the number of purchase invoices
moment is the quality of data being entered. I think processed per site, number of materials issued,
there is a perception that nobody is using SAP, received or transferred in a certain period, etc.
for once the information is entered, it’s forgotten, Although some of the reports taken directly
it’s never used again. And therefore a lot of the out of the system would give all the information
1224
that was needed, the layout would generally be DIscUssION AND tHEOry
not right for showing those reports to manage- DEVELOPMENt
ment; therefore the need to manipulate these data
in Excel. Excel in this case was considered more In this section we draw on the analysis presented
advanced, in terms of being able to summarize and above on the ways control and drift can occur, as
carry out calculations on the data, produce graphs, well as our theoretical concepts of disembedding /
manage the layout, etc. However, by taking data reembedding and human and machine agencies, in
out of the ERP system and manipulating them order to develop our theoretical conceptualisation
in another piece of software (Excel) that did not that explains the processes that lead to control and
impose access restrictions, data could be (inten- drift within the context of an Enterprise System.
tionally or unintentionally) falsified, producing
erroneous results. If this occurred, then control Human and Machine Agencies
over the accuracy and truthfulness of those data
would be lost, and drift would occur: The actual use of the system in the analysis of
the five companies can be conceptualised by the
You have that functionality, it’s very good, you can interplay between human and machine agencies. A
export it [data from the system]. The problem is, human agency position in the use of the Enterprise
that you can then manipulate the data [in Excel] System in this case assumes that humans (Enter-
into any way you want, you know, anyhow you prise System users in this case) could choose to
want. And for me, that’s potential loss of control, use the system minimally, invoke it according to
because, OK, if you imagine, 2 groups of people needs, or improvise in ways that could produce
are producing the same data in theory, manipu- unanticipated consequences (Boudreau & Robey,
lating it slightly differently, and potentially you 2005; Orlikowski, 2000). On the other hand, ma-
turn up with 2 individuals at the same meeting, chine agency is assumed to embody rules guiding
with 2 different sets of data. (SAP Facilitator 2, human action, limiting choice alternatives and
TransCom) monitoring human action (Boudreau & Robey,
2005; Winner, 1977; Zuboff, 1988). Although ma-
The use of Excel was also not always a panacea chines themselves are products of human agency
for the production of reports. Users had to know (when the development stage is considered), when
Excel quite well, in order to be able to manipulate they are installed and left to operate they then
the data coming out of the Enterprise System. In become constraints or enablers of human agency
the Materials Management area in TransCom for (Boudreau & Robey, 2005; Nandhakumar et al.,
example, the output from more than one areas 2005; Rose et al., 2003).
of SAP had to be extracted at a time, and each In other words, the assumption in the han-
of those outputs had to be entered into an Excel dling of agency in this chapter is that human
spreadsheet and then combined using specialist (user) agency is characterised by intentionality
functions such as vlookup. This was seen to be too to perform certain actions in the system, while
complicated and time-consuming to do, distracting machine (system) agency is characterised by af-
the users from the main job that they should be fording (enabling or constraining) the intended
doing, rather than having to learn complex func- human actions in the system. This position follows
tions in Excel to manipulate the data that were amongst others the works of Jones (1999), Rose
taken out of the system. and Jones (2004) and Rose et al. (2003), who argue
that by defining agency as the “capacity to make
a difference” (Giddens, 1984), both machines
1225
and humans can be viewed as possessing agency. operate in cycles, i.e. the output of one phase of
However, humans have intentionality and self and the workflow being used as the input in the next
social awareness, which result in capabilities for phase of the workflow. This would also mean that
interpretation of particular situations and actions. an action of a user in the system would impact
Machines however lack those capabilities (at least other users of the system, in a cause and effect,
to a large degree relative to humans and based on or action/consequence manner.
current technologies). However, we argue that the employment of
the disembedding/reembedding concepts goes
Embedding-Disembedding a step further from simply describing cause and
of Human Agency effect. Due to the global and integrative nature of
the Enterprise System, the action of a user in the
In this chapter we argue that human agency can system is not simply the cause for another action
also be further elaborated into the disembedding by another user, but also entails visibility across
and reembedding of user actions in the system. geographical, functional and time dimensions.
This refers to the ways that actions of one user This means that the disembedded action in the
are disembedded to other users, and these actions system is recorded, and could be examined by
then being reembedded by them as input for their authorised users in other locations of the company,
own work in the system. The disembedding of in other functional areas, or in the future if need
actions in this case can be linked to data input arose. This action therefore then ceases to be lo-
by users in the system, and the reembedding of cal, but is “disembedded” across time and space.
actions can be linked to data use by users of the As discussed, the disembedding and reembed-
system. Disembedded actions refer to users in ding of human actions in the system can be seen
one department using the system to carry out a to be facilitated by the machine agency of the
transaction (such as finance giving authorisation Enterprise System. For example, the fact that the
to do the stock count). This action then gets disem- system consists of a single global database enables
bedded for example from the finance area to other the user actions to be disembedded in terms of be-
departments or users concerned (for example the ing visible to other users. The single database also
Materials Management area). This disembedded enables other users to view disembedded actions
action then gets reembedded locally (in this case in the system and reembed them, in terms of using
by the Materials Management users) in terms of them as the input for their own work. The disem-
being able to use it as input for their own work bedding and reembedding of human actions can
(in this case carrying out the stock count). also be seen to be facilitated by the workflow and
The disembedding and reembedding of hu- access controls in the ERP system, which specify
man actions in the system can be seen to be fa- which actions users could take in the system in
cilitated by the workflow that is inscribed in the order for them to be disembedded, and to which
Enterprise System, and the single database that other users those actions would impact upon, in
enables monitoring of user actions. Authorisa- terms of those actions being reembedded by them.
tions for the disembedding and reembedding of When the user actions in the system are
information are then made possible with the use disembedded however, any errors, omissions
of access profiles that specify which users could or inconsistencies by the user are disembedded
disembed and reembed various pieces of infor- as well. This could have a “ripple” effect to the
mation in the system. Because of the workflow company’s operations as a whole, and therefore
inscribed in the ERP system, the disembedding drift might occur. As has been discussed in the
and reembedding of information could be seen to analysis section, this drift might occur because of
1226
the system configuration, which could allow for Although the factors from Enterprise System
missing information in the system, the bypassing use in Figure 1 are shown to lead to organizational
of controls by users, or the use of external systems control and/or drift, this result is in essence the
such as Excel. interpretation of the researcher according to the
unique context where these factors are identified.
conceptualisation of Enterprise It can be argued for example however, that the
system Use and Organizational same factors could be interpreted in another way
consequences (control and Drift) in a different context.
For example, while missing information was
The figure below presents our conceptualisation interpreted as resulting in organizational drift, it
of the organizational impact (classified as control could be argued that such missing information
or drift) of the use of Enterprise Systems, accord- was the outcome of users not inputting the required
ing to the data we gathered from the case study information in order to signal problems with the
companies and the theoretical frameworks that system or data incompatibilities, and this could
we employed. be viewed as a control factor (e.g. Kavanagh,
2004; Marakas & Hornik, 1996; Markus, 1983).
Figure 1. The impact of enterprise systems on control and drift: A human and machine agency perspec-
tive through the embedding and disembedding of human actions
1227
Similarly, with the case of the monitoring of user both possess and exhibit agency. We have based
actions afforded by the Enterprise System which our conceptualisation on the assumption that the
was interpreted to lead to organizational control, Enterprise System acts by affording (enabling or
in other contexts it could potentially lead to users constraining) states of control and drift. We have
resisting the system. In this sense, some users in also based our conceptualisation on the assumption
TransCom have expressed their perception of the that human actions are characterised by intention-
visibility offered by the ERP system as a kind of ality, and have shown that those actions can be
“Big Brother”. Although these users were subse- embedded and disembedded, consequently having
quently educated and trained to use the system, an impact on organizational control and drift.
the visibility functionality of the ERP system Our findings draw upon and complement the
could otherwise potentially result in user resistance findings by Hanseth et al. (2001), where side
which would be detrimental to the company effects can lead to loss of control (drift) due to
(Kossek et al., 1994; Martinko et al., 1996), and the integrative nature of an ERP, through which
which would therefore cause organizational drift these side effects are “rippled”. We have enhanced
as far as the use of the system is concerned. As this understanding by arguing that drift can oc-
such, the interpreted nature of the factors from cur through the disembedding and reembedding
ERP use leading to either organizational control of user actions in the system (see Figure 1). Our
or drift must be mentioned, which depends on the results also refine Boudreau & Robey (2005),
context where these factors are encountered. who argue that although ERPs can constrain hu-
The importance of this conceptualisation then man agency, this can still take place within such
does not lies in enumerating the various factors systems. We argue that this exercise of human
affecting ERP use, as well as the factors from agency from the part of users can lead to drift,
ERP use leading to either control or drift. The as shown in Figure 1. Finally, we argue that the
importance of the conceptualisation lies more in concept of panoptic empowerment as identified
sensitizing the reader with regards to the need to by Elmes et al. (2005), occurs through the dis-
examine the context where an Enterprise System embedding and consequent reembedding of user
is used, as well as the actual use of the system actions in the system.
and its outcome in terms of organizational control For practitioners, our findings indicate that
and drift. This means that when examining the attention needs to be paid to the setting of ac-
use of an ERP system, the factors impacting its cess profiles, according to what is afforded by
use must be examined, as well as the affordances the system. Attention should also be paid to the
of the system, the actions of users in the system, actual configuration of the system, to minimize
and the interdependencies between actions of the bypassing of controls and the unwanted or
different users (through the disembedding and unauthorised use of the system, which can lead to
reembedding of those actions). drift. The importance of inputting required infor-
mation in the system should also be emphasized
to employees, although the system may not force
cONcLUsION AND IMPLIcAtIONs users to enter it. This also links to user training in
order to increase the understanding of the system.
The aim of this chapter has been to present and If users don’t understand how the system works
theorize the impact of actions related to an Enter- and how their actions in the system impact other
prise System on organizational control and drift. users, then errors, omissions or inconsistencies
We have taken an agency perspective, where hu- will keep occurring and the effects of those will
mans (users) and machine (the Enterprise System) be “rippled” across the company, resulting in drift.
1228
On the other hand, too tight controls in the Bloomfield, B. P., & Coombs, R. (1992). Informa-
system can reduce the resiliency of the company tion Technology, Control and Power: The Central-
to respond to future challenges and adapt to new ization and Decentralization Debate Revisited.
realities (Ignatiadis & Nandhakumar, 2005). In Journal of Management Studies, 29(4), 459–484.
that sense, some degree of calculated drift may be doi:10.1111/j.1467-6486.1992.tb00674.x
beneficial, as long as it is not caused by inadvertent
Bloomfield, B. P., Coombs, R., & Owen, J. (1994).
or uninformed user actions. The thin line between
The social construction of information systems -
control and drift is different for each company,
the implications for management control. In R.
and depends on factors internal and external to
Mansell (Ed.), Management of Information and
it. Our model presents an understanding of the
Communication Technologies - Emerging Patterns
processes that lead to control and drift, so that
of Control (pp. 143-157). London: Aslib.
managers can tailor these processes according to
the idiosyncrasies of their companies. Boudreau, M.-C., & Robey, D. (1999). Organiza-
tional transition to enterprise resource planning
systems: theoretical choices for process research.
rEFErENcEs Paper presented at the 20th International Confer-
ence on Information Systems, Charlotte, North
Akkermans, H., & van Helden, K. (2002). Vicious Carolina, United States.
and virtuous cycles in ERP implementation: a case
study of interrelations between critical success Boudreau, M.-C., & Robey, D. (2005). Enacting
factors. European Journal of Information Systems, Integrated Information Technology: A Human
11(1), 35–46. doi:10.1057/palgrave/ejis/3000418 Agency Perspective. Organization Science, 16(1),
3–18. doi:10.1287/orsc.1040.0103
Al-Mashari, M., & Al-Mudimigh, A. (2003).
Enterprise Resource Planning: A taxonomy of Ciborra, C. U. (2000). A Critical Review of the
critical factors. European Journal of Operational Literature on the Management of Corporate Infor-
Research, 146(2), 352–364. doi:10.1016/S0377- mation Infrastructure. In C. U. Ciborra, K. Braa,
2217(02)00554-4 A. Cordella, B. Dahlbom, A. Failla, O. Hanseth, V.
Hepso, J. Ljungberg, E. Monteiro & K. A. Simon
Bernroider, E., & Koch, S. (1999). Decision mak- (Eds.), From Control to Drift - The Dynamics of
ing for ERP investments from the perspective of Corporate Information Infrastructures (pp. 15-
organizational impact - Preliminary results from 40). Oxford: Oxford University Press.
an empirical study. Paper presented at the 5th
Americas conference on information systems, Ciborra, C. U. (2002). The Labyrinths of Informa-
Milwaukee, WI, USA. tion: Challenging the Wisdom of Systems. Oxford:
Oxford University Press.
Besson, P., & Rowe, F. (2001). ERP project dynam-
ics and enacted dialogue: perceived understanding, Coombs, R., Knights, D., & Willmott, H. C.
perceived leeway, and the nature of task-related (1992). Culture, Control and Competition;
conflicts. The Data Base for Advances in Informa- Towards a Conceptual Framework for the
tion Systems, 32(4), 47–65. Study of Information Technology in Organi-
zations. Organization Studies, 13(1), 51–72.
Bingi, P., Sharma, M. K., & Godla, J. (1999). doi:10.1177/017084069201300106
Critical Issues Affecting an ERP Implementation.
Information Systems Management, 16(3), 7–14.
doi:10.1201/1078/43197.16.3.19990601/31310.2
1229
Duane, A., & Finnegan, P. (2003). Managing em- Ignatiadis, I., & Nandhakumar, J. (2007). The
powerment and control in an intranet environment. impact of Enterprise Systems on organizational
Information Systems Journal, 13(2), 133–158. resilience. Journal of Information Technology,
doi:10.1046/j.1365-2575.2003.00148.x 22(1), 36–43. doi:10.1057/palgrave.jit.2000087
Elmes, M. B., Strong, D. M., & Volkoff, O. Jones, M. R. (1999). Information Systems and the
(2005). Panoptic empowerment and reflective Double Mangle: Steering a Course between the
conformity in enterprise systems-enabled orga- Scylla of Embedded Structure and the Charybdis
nizations. Information and Organization, 15(1), of Strong Symmetry. In T. J. Larsen, L. Levine, J.
1–37. doi:10.1016/j.infoandorg.2004.12.001 I. DeGross (Eds.), Information Systems: Current
Issues and Future Changes (pp. 287-302). New
Gattiker, T. F., & Goodhue, D. L. (2004). Un-
York: OmniPress.
derstanding the local-level costs and benefits of
ERP through organizational information process- Kallinikos, J. (2004). Deconstructing infor-
ing theory. Information & Management, 41(4), mation packages: Organizational and behav-
431–443. doi:10.1016/S0378-7206(03)00082-X ioural implications of ERP systems. Infor-
mation Technology & People, 17(1), 8–30.
Gibson, J. J. (1979). The ecological approach
doi:10.1108/09593840410522152
to visual perception. Boston: Houghton Mifflin.
Kavanagh, J. F. (2004). Resistance as Motivation
Giddens, A. (1984). The Constitution of Society:
for Innovation: Open Source Software. Com-
Outline of the Theory of Structuration. Cambridge:
munications of the Association for Information
Polity Press.
Systems, 13, 615–628.
Giddens, A. (1990). The Consequences of Mo-
Koch, C. (2001). BPR and ERP: Realising
dernity. Cambridge: Polity Press.
a vision of process with IT. Business Pro-
Glaser, B. G., & Strauss, A. (1967). The discovery cess Management Journal, 7(3), 258–265.
of grounded theory: Strategies for qualitative doi:10.1108/14637150110392755
research. New York: Aldine.
Kossek, E. E., Young, W., Gash, D., & Nichol, V.
Hanseth, O., Ciborra, C. U., & Braa, K. (2001). (1994). Waiting for Innovation in the Human Re-
The Control Devolution: ERP and the Side Effects sources Department: Godot Implements a Human
of Globalization. The Data Base for Advances in Resource Information System. Human Resource
Information Systems, 32(4), 34–46. Management, 33(1), 135–160. doi:10.1002/
hrm.3930330108
Holland, C. P., & Light, B. (1999). A critical suc-
cess factors model for ERP implementation. IEEE Legare, T. L. (2002). The role of organizational
Software, 1999, 30–36. doi:10.1109/52.765784 factors in realizing ERP benefits. Information
Systems Management, 19(4), 21. doi:10.1201/1
Hong, K. K., & Kim, Y. G. (2002). The criti-
078/43202.19.4.20020901/38832.4
cal success factors for ERP implementation: an
organizational fit perspective. Information & Malone, T. W. (1997). Is Empowerment Just a
Management, 40, 25–40. doi:10.1016/S0378- Fad? Control, Decision Making, and IT. Sloan
7206(01)00134-3 Management Review, 38(2), 23–35.
1230
Marakas, G. M., & Hornik, S. (1996). Passive Newell, S., Huang, J. C., Galliers, R. D., &
Resistance Misuse: Overt Support and Covert Pan, S. L. (2003). Implementing Enterprise
Recalcitrance in IS Implementation. European Resource Planning and knowledge management
Journal of Information Systems, 5(3), 208–220. systems in tandem: fostering efficiency and
doi:10.1057/ejis.1996.26 innovation complementarity. Information and
Organization, 13(1), 25–52. doi:10.1016/S1471-
Markus, M. L. (1983). Power, Politics, and MIS
7727(02)00007-6
Implementation. Communications of the ACM,
26(6), 430–444. doi:10.1145/358141.358148 Norman, D. A. (1988). The Psychology of Every-
day Things. New York: Basic Books.
Markus, M. L., & Tanis, C. (2000). The Enterprise
Systems experience - From adoption to success. Orlikowski, W. J. (1991). Integrated Information
In R. W. Zmud (Ed.), Framing the domains of Environment or Matrix of Control? The Con-
IT management: Projecting the future through tradictory Implications of Information Technol-
the past. Cincinnati, OH: Pinnaflex Educational ogy. Accounting . Management and Information
Resources, Inc. Technologies, 1(1), 9–42. doi:10.1016/0959-
8022(91)90011-3
Martin, I., & Cheung, Y. (2000). SAP and
Business Process Reengineering. Busi- Orlikowski, W. J. (2000). Using Technology
ness Process Management, 6(2), 131–121. and Constituting Structures: A Practice Lens for
doi:10.1108/14637150010321286 Studying Technology in Organizations. Orga-
nization Science, 11(4), 404–428. doi:10.1287/
Martinko, M. J., Henry, J. W., & Zmud, R. W.
orsc.11.4.404.14600
(1996). An attributional explanation of indi-
vidual resistance to the introduction of informa- Orlikowski, W. J., & Baroudi, J. J. (1991). Study-
tion technologies in the workplace. Behaviour ing Information Technology in Organizations:
& Information Technology, 15(5), 313–330. Research Approaches and Assumptions. Informa-
doi:10.1080/014492996120094 tion Systems Research, 2(1), 1–28. doi:10.1287/
isre.2.1.1
Miles, M. B., & Huberman, A. M. (1994). Quali-
tative Data Analysis (2nd ed.). Thousand Oaks, Pickering, A. (1995). The Mangle of Practice:
California: Sage Publications. Time, Agency and Science. Chicago: University
of Chicago Press.
Nah, F. F., Lau, J. L., & Kuang, J. (2001). Critical
factors for successful implementation of enterprise Rose, J., & Jones, M. R. (2004). The Double
systems. Business Process Management Journal, Dance of Agency: a socio-theoretic account of how
7(3), 285–296. doi:10.1108/14637150110392782 machines and humans interact. Paper presented
at the ALOIS Workshop: Action in Language,
Nandhakumar, J., Rossi, M., & Talvinen, J.
Organizations and Information Systems, Linkop-
(2005). The dynamics of contextual forces of ERP
ing, Sweden.
implementation. The Journal of Strategic Infor-
mation Systems, 14(2), 221–242. doi:10.1016/j. Rose, J., Jones, M. R., & Truex, D. (2003).
jsis.2005.04.002 The problem of agency: How humans act, how
machines act. Paper presented at the ALOIS
Workshop: Action in Language, Organizations
and Information Systems, Linkoping University,
Linkoping, Sweden.
1231
Rose, J., & Truex, D. (2000). Machine agency Somers, T., & Nelson, K. (2001). The Impact
as perceived autonomy; an action perspective. of Critical Success Factors across the Stages of
In R. L. Baskerville, J. Stage & J. I. DeGross Enterprise Resource Planning Implementations.
(Eds.), Organizational and Social Perspectives on Paper presented at the Hawaii International Con-
Information Technology (pp. 371–390). Aalborg, ference on Systems Sciences.
Denmark: Kluwer.
Tang, M., Sia, S. K., Soh, C., & Boh, W. F. (2000).
Schrnederjans, M. J., & Kim, G. C. (2003). Imple- A Contingency Analysis of Post-bureaucratic
menting enterprise resource planning systems with Controls in IT-related Change. Paper presented at
total quality control and business process reengi- the 21st International Conference on Information
neering survey results. International Journal of Systems, Brisbane, Queensland, Australia.
Operations & Production Management, 23(3/4),
van Fenema, P. C., & van Baalen, P. J. (2005).
418–429. doi:10.1108/01443570310467339
Strategies for Dealing with Drift during Imple-
Shang, S. S. C., & Seddon, P. B. (2000). A com- mentation of ERP Systems. Rotterdam: Erasmus
prehensive framework for classifying the benefits Research Institute of Management.
of ERP systems. Paper presented at the Americas
Walsham, G. (1993). Interpreting Information
Conference on Information Systems, Long Beach,
Systems in Organizations. Chichester: John Wiley
California.
& Sons.
Shanks, G., Parr, A., Hu, B., Corbitt, B., Thana-
Walsham, G. (1995). The Emergence of Inter-
sankit, T., & Seddon, P. B. (2000). Differences in
pretivism in IS Research. Information Systems
critical success factors in ERP systems implemen-
Research, 6(4), 376–394. doi:10.1287/isre.6.4.376
tation in Australia and China: a cultural analysis.
Paper presented at the European Conference on Winner, L. (1977). Autonomous Technology.
Information Systems, Vienna, Austria. Technics-out-of-Control as a Theme in Political
Thought. London: MIT Press.
Sia, S. K., Tang, M., Soh, C., & Boh, W. F. (2002).
Enterprise Resource Planning (ERP) Systems as a Zuboff, S. (1988). In the Age of the Smart Ma-
Technology of Power: Empowerment or Panoptic chine: The Future of Work and Power. Oxford,
Control? The Data Base for Advances in Informa- UK: Heinemann Professional Publishing.
tion Systems, 33(1), 23–37.
Siriginidi, S. R. (2000). Enterprise Resource
Planning in Reengineering Business. Business
Process Management Journal, 6(5), 376–391.
doi:10.1108/14637150010352390
1232
1233
Chapter 5.4
The Impact of Enterprise
Systems on Business Value
Sanjay Mathrani
Mohammad A. Rashid
Dennis Viehland
A significant investment in resources is required for The implementation of enterprise systems has been
implementation of integrated enterprise systems as considered the most important development in cor-
technology solutions while the effectiveness of these porate use of information technology (Davenport,
systems to achieve business value remains unclear 1995, 1998; Davenport & Prusak, 1998; Deloitte,
and empirically largely unexplored. Enterprise 1998). Enterprise systems (ES) broadly include all
systems integrate and automate business processes, enterprise-wide systems. These include enterprise
but unarguably, the real business value can only be resource planning (ERP) systems or any extended
achieved from improvements through the transfor- modules such as supply chain management (SCM) or
mation of enterprise systems data into knowledge by customer relationship management (CRM) systems.
applying analytic and decision making processes. However, despite a few dramatic successes, many
This study explores a model of transforming ES organizations still fail to realize the benefits while
data into knowledge and results by comparing two incurring huge costs and schedule overruns (Dalal,
case studies that examine the impact of enterprise Kamath, Kolarik, & Sivaraman, 2004). It has been
systems information on organizational functions and estimated that half of all ES implementations fail to
processes leading to realization of business value. achieve the desired results (Jarra, Al-Mudimigh, &
Zairi, 2000). In most cases enterprise systems are
DOI: 10.4018/978-1-59904-859-8.ch010
The Impact of Enterprise Systems on Business Value
implemented to improve organizational effective- penetration capability into the competitive market
ness (Davenport, 1998, 2000; Marcus & Tanis, by improving their operations. Both organizations
2000). These software applications connect and had implemented ES for at least three years and
manage information flows across complex orga- so were in their mature phase of implementation.
nizations, allowing managers to make decisions One organization had achieved considerable suc-
based on information that accurately reflects the cess from their ES implementation whereas the
current state of their business (Davenport, Harris, other had achieved little success. The two cases
& Cantrell, 2002). are compared to identify reasons for their levels
These systems are implemented to bring about of success.
definite business benefits that justify the invest-
ment. Truly significant return on investment (ROI)
comes from the process improvements that ES bUsINEss bENEFIts
supports and not just from improved information
access. In most implementations, ES software The justification for adopting ES centres on an-
alone makes marginal improvement in business ticipated business benefits from the ES. To receive
performance. If organizations continue to follow benefit from an ES, there must be no misunder-
the same pre-ES business processes after imple- standing of what it is about, its usability, and,
mentation, they can expect the same or possibly even more importantly, organizational decision
worse performance. ES software can, however, makers must have the background and tempera-
enable and support many new and improved ment for this type of decision making coupled
processes, but not without the organization decid- with the right quality of information (Donovan,
ing what those processes are and committing to 1998). Many researchers have evaluated benefits
them. Positive ROI can come from changing the from ES investments (Cooke & Peterson, 1998;
way business was performed in the past to more Davenport et al., 2002; Deloitte, 1998; Donovan,
streamlined, faster and lower cost processes that 1998, 2001; Hedman & Borell, 2002; Ittner &
better serve the needs of the customer and, if that Larcker, 2003; Shang & Seddon, 2000; Yang &
is done well, the organization will be a winner Seddon, 2004). These studies have found that
(Donovan, 2003). ES’s are designed to help manage organizational
The focus of this chapter is to better understand resources in an integrated manner. Furthermore,
the effectiveness of enterprise systems technology the level of integration that is promoted across
in an organizational setting. A qualitative research functions in an enterprise closely relates to the
methodology is used to explore how firms can primary benefits that are expected as a result of
leverage ES technologies to realize improved their implementation. After adoption, improved
business value. Field studies were conducted in business performance should produce both opera-
two large manufacturing organizations in India tional and strategic benefits (Irving, 1999; Jenson
that have implemented ESs, in order to understand & Johnson, 2002; Nicolaou, 2004).
their experience in achieving growth by leveraging A study of 85 global companies (Deloitte,
data from their ES investment. Semi-structured 1998) found tangible benefits (e.g. cost savings,
interviews were conducted with senior managers faster processing) and intangible benefits (e.g.,
of the two organizations between January 2005 and improved information visibility, new/improved
February 2006. The empirical data were integrated processes, and improved customer responsive-
and analysed to formulate inferences presented ness) from ES implementation. In a survey of 163
in this paper. Both organizations had aggressive large firms (Davenport et al., 2002) key benefits
growth plans with an objective to achieve better realized by organizations adopting ES included
1234
better management decision making, improved to support a business decision. The third are the
customer service and retention, ease of expansion/ outcomes which are the events that change as a
growth, increased flexibility, faster and more ac- result of the analysis and implementation of the
curate transactions, cost reduction, and increased decisions made. As Davenport’s model shows, the
revenue. There have also been some studies on process of ES data transformation into knowledge
organizational benefits resulting from overlapping leads into organizational changes. The most basic
implementation of knowledge management (KM) potential outcome of this process is the changes
and ES in organizational settings (e.g., Newell, in behaviors of individual managers, employ-
Huang, Galliers, & Pan, 2003; Baskerville, ees, customers, suppliers, and all stakeholders
Pawlowski, & McLean, 2000; Bendoly, 2002). in the value chain. Another outcome from the
However, this paper focuses on the impact of ES decisions or the behavioral changes may be
on realizing business value through the process of new initiatives to bring about improvements in
ES data transformation into knowledge by apply- business or make changes in existing projects.
ing analytic and decision-making processes. This The results of decisions can also include process
study attempts to establish the link between data, changes– determining that an existing process is
decisions, and actions, its impact on functional and not working effectively can lead to changes in the
business processes, and their outcomes. existing process or design and implementation
of an entirely new process. The ultimate results
of all these activities are the business benefits
tUrNING Es DAtA INtO which lead to positive financial impacts for the
Es KNOWLEDGE organization. “Decisions lead to new behaviors,
new initiatives, and processes, which do not
A model conceptualized and used by Davenport matter unless they improve the bottom line and
(2000) and his team of researchers for turning the return to share holders” (Davenport, 2000) p.
ES data into ES knowledge is shown in figure 225). It may be difficult to draw a direct chain of
1. The model comprises three major steps. The influence from prerequisites to transformation to
first is the context. This includes the factors that non-financial outcomes to financial results, but
must be present for transformation of ES data into establishing that linkage should be the objective
knowledge and results. The second is transforma- of an organization that invests effort and resources
tion of ES data into knowledge which takes place in ES data transformation (Davenport, 2000) and
when the data are actually analyzed and then used is the focus of this study.
Figure 1. A model of how ES data are transformed into knowledge and results
1235
The two case studies are presented next and the context: strategic
discussion follows Davenport’s model- context,
transformation, and outcomes for both cases as Growel’s journey towards becoming an interna-
shown in Figure 1. tional e-business began in the late 1990’s. The
company wanted to grow exports, widen its global
footprint, secure new customers, become the
cAsE stUDy 1 world’s largest manufacturer of axle components,
and be a key global provider of engine components.
company Overview To achieve these goals, Growel planned to double
its manufacturing capacity by implementing a
Growel Limited1 is a U.S. $1 billion forging major capacity expansion program. However,
manufacturing company and is one of the world’s this involved large investments and risks. The
largest manufacturers and exporters of automotive company had to resort to major cost controls,
engine and suspension components. It has the improve operational efficiencies and optimize its
world’s largest single-location forging capacity business processes to counter the adverse financial
and one of the most technologically advanced effects of the major investments. Senior managers
commercial forge shops in the world. Growel has in the company decided to pursue a strategy of
been a publicly traded company whose stock has operational excellence. The company historically
appreciated more than 200 percent since March lacked integration between its order-to-cash, ship-
2004. With manufacturing facilities in India and ping, and accounts receivable processes. There
Germany, the company manufactures a wide were disputes on invoices and purchase orders
range of forgings and machined components for relating to price and terms of business. There was
the automotive, diesel engine, railway, earthmov- a lack of visibility into finished goods inventory
ing, cement, sugar, steel, coal, ship building, and and overall accuracy of inventory was poor. Vis-
oilfield industries. An ISO 9001:2000, ISO/TS ibility of material requirements and inventory
16949:2002 accredited company, Growel is inter- throughout the value chain was inadequate and
nationally reputed for its cutting edge technology, did not provide decision support at all stages of
established quality processes, and capabilities to operations.
meet the exacting standards of the most demanding The company had not integrated the design and
customers in the world. Growel Limited is a global development practices with the operational sys-
corporation with world class engineering capabili- tems; therefore the time lag between development
ties, state-of-the-art manufacturing facilities, and a and marketing of products was large and resulted
global customer base that includes General Motors, in poor customer service and dissatisfaction. The
Toyota, Ford, Daimler Chrysler, Honda, Renault, pre-production approval process was another
Volvo, Caterpillar – Perkins, and several others. aspect which required attention. The company
It is the largest manufacturer of axle components needed the ability to interactively participate
for heavy trucks and has a 35% global market with its customers at the early stages of product
share, with a 10% global market share in engine development and avoid rework at a later stage.
components. The following sections discuss how The product forecasting process also required im-
Growel leveraged knowledge-driven technologies provements. Managers only discovered that they
to improve business dynamics with considerable had a shortage of manufacturing capacity when
success. The discussion follows Davenport’s the line ran out. On the sales side, management
model – context, transformation, outcomes – as had limited visibility of who its most and least
shown in Figure 1. profitable customers and products were. They
1236
also did not have information on whether they limitations for analysis. They also have a thorough
were buying in the most cost effective manner. working knowledge of several analytic and data
The management team recognized what types of presentation software packages, along with strong
decisions had to be made to support their strate- interpersonal skills to train and support end users.
gic objectives but could not utilize the available
operational data. context: Data
context: Organizational and cultural Issues of data quality were less at Growel, where
transaction data captured in the SAP system were
A company is as good as its people and Growel created internally based on all transactions from
has the advantage of having a highly qualified and sales orders to shipping invoices. Monitoring and
motivated manpower base. Since its inception, updating of data was a regular feature at Growel
Growel has attached great significance to “people and the transaction data was made available in a
power” and considers its employees as important timely fashion to support decision making.
assets. With interactive communications at all
levels, Growel continues to provide a congenial context: technology
and peaceful working atmosphere to its employ-
ees. In the process of ES implementation, the Growel had historically been using a home-
organizational and cultural elements were aligned grown legacy system which provided disparate
to support the use of transaction data at Growel. information which lacked proper integration and
The compensation system was also changed to utilization. However, this lack of operational data
reward sales people for sales volume and profit to support decision making changed with the
to include a fixed and variable component of implementation of SAP’s R/3 in 2000-01. The
pay. The company created a friendly atmosphere modules implemented were finance, sales and
within the organization which fostered orientation distribution, materials management, production
to change. The organization also adopted a data- management, and human resources. The company
oriented culture and encouraged employees to use now had unprecedented visibility into its opera-
data to support any business decision. tions and customer base. SAP business intelligence
tools were extensively used to extract, analyze,
context: skills and Knowledge and develop adhoc reports.
Growel has always had a high quality, motivated transformation

work force. The company employs about two
thousand workmen of which over four hundred The transformation process at Growel was a result
are engineers with a high ability to learn and of putting knowledge-leveraging activities into
implement modern manufacturing methods us- action. How this happened is explained next. The
ing high tech equipment. The company provides value creation process was initially described in
extensive training both in house and externally, detail to gain an in-depth understanding of where
including overseas exposure. Within the group of and how Growel adds value for its customers. A
knowledge workers and analysts, the skills include critical success factor framework for each func-
detailed knowledge of the organization’s underly- tional area was developed and a strong linkage
ing business processes. They possess extensive between departmental performance indicators
skills for interpreting the SAP data, including and top-level metrics for gauging the effective-
understanding how key elements relate and their ness of company strategy was put in place. Task
1237
groups within each functional area translated the and purchase orders relating to price and terms
general framework into team-specific programs of business diminished. The online visibility of
to leverage innovations for achieving strategic demand and supplies with customers and suppliers
goals and plans. through the integrated supply chain management
The model was shared with all the relevant team (SCM) system led to less volatility in sudden spurts
players. Descriptive indicators of the improve- of demand which existed earlier. This resulted in
ment and corrective-action plans were identified more streamlined supplies and a dramatic change
to facilitate decision making. The implementation in customer and supplier behavior since Growel
plans for these decisions to achieve the desired could react better to change orders now and was
results along with the steps to create the results able to be more flexible in the manufacturing
were identified. Forms were designed to describe environment.
these plans and their measures. The indicators
were documented choosing the reference for Outcomes: New Initiatives
benchmarking and external validation along the
time-line. The analytical process was the means by The ability to analyze customer and product
which ES data became knowledge. The SAP data, profitability lead to a new initiative of value
required for each of the indicators were identified, engineering to improve or change the design to
extracted, and interpreted, to create useful infor- make the product more profitable. Unprofitable
mation for monitoring the progress for achieving products and customers were identified and the
the objectives. The signals and messages coming division’s existing unprofitable product lines were
from each indicator were analyzed and evaluated replaced by more profitable new product lines.
to support decision-making. The decision-making New initiatives towards implementing just-in-
process was based on high-quality, well-analyzed time inventory systems were undertaken, which
ES data on a multitude of factors. Some key areas decreased inventory costs substantially.
where ES data were utilized for improvements
were customer and product profitability analysis, Outcomes: Process changes
price/volume analysis, market and customer seg-
ment analysis, sales forecasting and operations Growel recognized that the SAP data created
planning analysis. All actions likely to improve opportunities for redesigning some business pro-
the likelihood that the result will be coherent with cesses which could create entirely new sets of
the strategic intent were identified, evaluated, and decisions. Growel is moving at full speed to
implemented. re-design some business processes and build e-
commerce applications with SAP as a backbone
Outcomes: changing behaviors for their legacy systems and other collaborative
software like SCM (Supply Chain Management)
One of the major outcomes from the initiatives de- and PLM (Product Lifecycle Management). SAP
scribed above was changing behaviors. Improved provides capabilities such as CRP (Capacity
information sharing, transparency, and openness Resource Planning), and BPR (Business Process
with customers, suppliers, employees, and all Re-engineering) which offer powerful links
stakeholders resulted in improved inter-personal within the entire value chain from customers to
and business relations. Having easy access to suppliers. The company has set up an integrated
invoice and purchase order data enabled Growel supply chain management system which enables
to improve price synchronization with customers real-time visibility of material requirements and
and suppliers. The earlier disputes on invoices inventory throughout the value chain and provides
1238
decision support at all stages of operations. With designs and manufactures diesel engines and
a majority of the company’s suppliers receiving gensets for industry sectors such as power, oil and
supply chain data, the company has a real-time gas, construction, earthmoving, and transporta-
total demand management system in place. A tion and supplies to various countries all over
virtual private marketplace has been created for the world. The unit employs about 1,200 people.
Growel through which the company engages in Originally Primemover only manufactured diesel
e-procurement and reverse auctions. The company engines. In 1993 the company collaborated with
has already started selling scrap online. a German company to include gensets in order to
In a development that will substantially reduce extend its range of products, to develop a more
product development time and bring the company complete supply capability to the engine manu-
closer to its customer, Growel is in the process of facturing industry, and to position the company
implementing a Collaborative Product Commerce for long-term growth. Both product ranges served
(CPC) module. CPC will enable the company as solution providers for their customers, provided
to work online with its customers to design and products with a large-range of kva and hp ratings.
develop products and share information and The synergies created by the integration of the
knowledge with the customers. This will reduce two product ranges enabled better designs and
product development time and costs and, more product offerings. Specifically, customers had
importantly, forge close ties with the customers more effective, one-stop access to a comprehensive
from early stages of product development. range of products as well as simplified commer-
cial relationships. By enabling improved service
to customers, the integration of the two product
cAsE stUDy 2 lines enhanced the market position of the power
generation group of Primemover. However, the
company Overview company realized that it would face challenges
to ensure its profitable growth in the long term.
Primemover2 is a multi-faceted engineering
enterprise. Established in 1859, Primemover context: strategic
is one of India’s leading and well-diversified
engineering companies with a US$500 million Primemover was one of the few companies that
turnover. The company’s core competencies are were full-line supplier of the entire product range
in diesel/petrol engines, power generating sets in industry markets however; the company was fac-
(gensets), agricultural and construction equip- ing growing challenges. Despite increased demand
ment. The business operations of the company are resulting from a healthy growing market, greater
divided into various business groups strategically competition in its core markets and high operating
structured to ensure maximum focus on each costs could inhibit the achievement of financial
business area and yet retain a unique synergy in returns expected from its operations. Primemover
the operations. The business groups are power had to overcome the competition, and leverage the
generation, agricultural equipment, light engines, new opportunities to ensure profitable growth for
and infrastructure equipment. Primemover has the future. To compete effectively, Primemover
six manufacturing plants located at several lo- needed to improve its supply chain performance
cations in India. The company has an extensive and cost; the business processes of the two product
sales and service network manned by a highly lines were not performing at the levels necessary
skilled and dedicated workforce. to grow profitably in the emerging competitive
The power generation group of Primemover environment. The time frames required to commit
1239
to delivery of finished goods to customers were on-going training and development to upgrade the
not competitive; customer order handling and skills and knowledge of its employees including
service processes were complex; operations were knowledge of the organization’s underlying busi-
not sufficiently flexible to enable rapid response ness processes.
to shifting demand; market share was not growing
rapidly; and inventory-carrying costs and other context: Data
expenses inhibited achievement of adequate finan-
cial returns. These challenges were compounded Quality of data was not an area of focus for
by the fact that 60% of Primemover diesel engine Primemover prior to ES implementation. The
and genset parts were externally sourced and there maintenance of data records was not consistent and
were long lead times for products such as pistons discrepancies were often encountered in the data
from Germany and turbochargers from UK. Fore- records. There was a lack of discipline in updating
cast accuracy was low, inventory data and related transactions in the warehouse which would lead
information were inaccurate. Thus, key goals to stock and other data discrepancies. This lack of
for Primemover included enhancing its ability accuracy and currency of information led to data
to respond to customer requirements, improving integrity issues amongst employees. The tools for
market share, and reducing costs throughout the data extraction were also inadequate. Availability
operation. Primemover determined that it must of transactional data and information was an issue.
improve its supply chain planning and execution Data extracts could not be made easily available
capabilities to improve its operations in order to and in time to support decision making.
achieve better penetration capability into the com-
petitive market, which was their prime objective. context: technology
context: Organizational and cultural Primemover realized a need to couple its execution
systems with a new class of supply chain plan-
Primemover has an overall employee base of ning software to address its requirements in the
about three thousand employees and has a change- future. After considering various software solu-
oriented organizational culture. Primemover tions and determining the business strategy, the
enjoys the enviable reputation of being one of the company selected SAP R/3 to be the foundation
few corporates that has successfully maintained of the planning system for its integrated opera-
harmonious human relations year after year and tions. Primemover had historically been using an
the credit for this goes to all its employees who internally-developed system which provided dis-
have always been open to and welcomed change. parate information that lacked proper integration
The organization structure is collaborative and and utilization. However, the situation changed
the compensation system is aligned to achieving with the implementation of SAP’s R/3 in 2000-
goals. 01. The company implemented financial, sales
and distribution, production management, human
context: skills and Knowledge resources along with Web-enabled supply chain
management systems. Primemover now had better
Out of the three thousand workmen Primemover visibility into its operations and customer base.
employs, over five hundred are qualified engineers Standard reports from SAP and Microsoft’s Excel
having technical and commercial expertise. The were used to analyze and report data.
company has hired mostly skilled workmen in the
last fifteen years. The company has been providing
1240
transformation of out-of-inventory events, but did not achieve

the reduction and streamlining of inventories
The transformation process at Primemover was anticipated. The lead times could not be reduced
based on the information from the SAP system. substantially and delivery performance improved
The available SAP data were utilized, interpreted, marginally. The supply issues from vendors and
and analyzed in various areas of operations to sup- customers continued and dramatic overall im-
port decision-making for achieving the company’s provement in behaviors was not achieved.
strategic objectives. Standard reports from the
SAP system were used and evaluated on a regular Outcomes: New Initiatives
basis to support decision-making. Many of these
reports were transformed into Excel spreadsheets With some ability to analyze demand and with
to facilitate application of analytical processes. the information now visible, the company imple-
The decision-making process was based on well- mented the following changes to its supply chain
analyzed ES data on a multitude of factors. processes. Primemover analyzed its various mar-
However the organization’s business strategy ket and customer segments, determined the service
needed more alignment into departmental and requirements for these segments, and classified
divisional strategies. The definition of informa- products according to volume and variability of
tion critical to the success of the organization was demand. Primemover established monthly sales-
missing. This was evident by a remark from one forecasting and operations-planning meetings
of the interviewees when he said “there was a lot where the supply-and-demand plans were formally
of confusion on what is to be achieved, which data reviewed; performance of locally produced and
needs to be analysed, where it is to be applied imported product was considered; exceptions
and to achieve what results”. The analytical and and unique requirements were brought up for
decision-making process was based on ES data discussion; and market intelligence and longer-
which were not very accurate. There was a lack term constraints in material and production were
of timely availability of data extracts to support factored into plans.
analytical decision making and the link between
data, decisions, and actions was missing. Some Outcomes: Process changes
key areas where ES data were utilized for im-
provements were market and sales forecasting In order to shorten lead times, reduce inventory,
and operations planning analysis, theory of con- and increase throughput, Primemover employed
straints in manufacturing analysis, raw material a “theory of constraints” model, establishing
cost analysis, inventory analysis, activity-based processes to identify critical material and capacity
costing, and process efficiency analysis. constraints – and to optimize these constraints in its
manufacturing operations on an ongoing basis. As
Outcomes: changing behaviors operations of the two product lines were unified,
customer account numbers were merged; SKUs
As a result of implementing an ES, the attitude were combined; and the number of warehouses was
of customers, suppliers, employees, and all stake- reduced from five to three. These consolidations
holders improved but did not reach the levels required considerable revamping of numbering
expected since inaccuracies were still present in schemes and physical storage strategies. Ware-
the historical data. The online visibility of demand house consolidation was coupled with improved
and supplies with customers and suppliers through policies associated with material movement,
the integrated SCM system reduced the number virtually eliminating the need to transfer material
1241
between locations once received. Due to better rEsULts AND FINANcIAL IMPActs
tracking of material and a substantial reduction of
the need to move material multiple times, costs for Growel has surpassed the last full year’s total rev-
damaged and lost material declined from 1.60% enue and exports in the first nine months. Today,
to 0.78% of sales. Also, strategically significant Growel has achieved the distinguished position as
and measurable improvements to inventory, cost, the largest manufacturer of axle components for
and customer service were achieved. heavy trucks world wide and one of the key global
Forecast accuracy improved as the company players for engine components. Primemover has
gained experience with sales forecasting and op- improved over last year’s total revenue and exports
erations planning. The new processes also gave (as shown in table 2). However, Primemover has
visibility into material requirements for sched- not fully achieved the anticipated objectives and
uled orders and further facilitated procurement results. Primemover had an expectation that the
planning. Primemover had earlier been placing inventory levels would reduce by 60-70% and cost
large, irregular orders monthly or bimonthly with efficiency would also improve by 40-50% how-
its suppliers but is now placing regular weekly ever, reduction in inventory levels achieved was
orders, improving its suppliers’ abilities to plan only 40% and improvement in cost efficiency also
and thereby improving Primemover’s negotiat- was only 20%. Another expectation of Primemover
ing position. The store manager at Primemover was that the human resources would become highly
remarked that the suppliers and customers were motivated as an end result but the motivation
now giving “fewer headaches” with the new sys- expected was not seen. Although, improvement
tem. Primemover’s customers are also able to plan in the areas of on-time delivery and time taken
better due to the company’s improved forecasts. for new product development, was achieved as
Resolution of manufacturing constraints has anticipated. On the financial side, improvement
improved production throughput by 20%. Daily expectation in the bottom line, profit after tax,
cycle counting has enabled inventory accuracy to was 20% by both Growel and Primemover. An
advance from 52% to more than 80%, which has improvement of only 9% was finally achieved
facilitated a reduction in the levels of raw materi- by Primemover whereas Growel achieved 26%.
als and finished goods. Finally, many employees
are undertaking what-if analyses with the range
of available data, further improving planning and DIscUssION
execution across the supply chain. Primemover
has chosen to integrate related product lines, These cases illustrated Davenport’s conceptual
establish supply chain channels to match supply- framework (Figure 1) of how business benefits
and-demand streams, and resolve constraints are achieved from ES data transformation into
in its manufacturing network. The company ES knowledge highlighting the effectiveness of
recognizes that its process designs are not static enterprise systems in the two organizations. The
and that its business and enabling systems will overall business benefits including the financial
continue to evolve in line with market demands. results obtained in Growel’s case surpassed those
Ongoing achievements in cost reductions and of Primemover. The reasons attributed to this are
customer service improvements have enabled given in the following section.
Primemover to make continual improvements in Growel and Primemover both had a business
market position, premium pricing opportunities, strategy and had identified their business objec-
and financial return. tives however; the business strategy in Growel’s
case was clearly articulated and aligned. Growel
1242
Table 1. Key successes achieved by Growel and Primemover
Growel Primemover
1.Product development time - Decreased to around three weeks 1.Product development time - Decreased to around two months compared
as compared to six months earlier; this closely matches industry to three months earlier; this closely matches industry benchmarks
benchmarks
2.On-time delivery - Increased to 98% on-time 2.On-time delivery - Increased to 85% on-time
3. Inventory - Reduced by 80% 3. Inventory - Reduced by 40%
4. Cost efficiency - Increased by 60% 4. Cost efficiency - Increased by 20%
5. Relationship management - Much improved relations with 5. Relationship management - Some improved relations with customers
customers and suppliers and suppliers
6. Human resources - Highly motivated 6. Human resources – Motivated
Table 2. Financial outcomes of Growel and Primemover as on March 31, 2005
Growel Primemover
1.Total revenues - Increased by 47% 1.Total revenues - Increased by 24%
2.Exports - Increased by 76%; contributes 48% of total revenue 2.Exports - Increased by 15%; contributes 8% of total revenue
3. Profit before tax - Increased by 37% 3. Profit before tax - Increased by 15%
4. Profit after tax - Increased by 26% 4. Profit after tax - Increased by 9%
worked out their value creation process and identi- Growel. The quality of management reviews also
fied the critical areas that required attention and improved at Growel because executives became
improvement. They understood the key drivers much more reliant on numbers in explaining their
and had the means to influence those drivers and performance. Issues of data quality did not exist
measure them. Growel was able to translate their at Growel. Monitoring and updating of data was a
business strategy into departmental or divisional regular feature and the transaction data were made
strategies. They knew what was to be achieved, available in a timely fashion to support decision
which data needed analysis, its area of applica- making. In the case of Primemover, the lack of
tion, and expected outcomes. Primemover had discipline in updating transactions in the ware-
identified their key business objectives however, house led to data discrepancies and data integrity
the definition of information critical to the suc- issues amongst employees. Data extracts could
cess of the organization was lacking. Primemover not be made in time to support decision making.
could not create the link between departmental The analytical and decision-making process was
performance indicators and top-level metrics for based on ES data which were unclean and inac-
gauging the effectiveness of the company strategy. curate and the link between data, decisions, and
Managers had data from their ES investment but actions was lacking.
could not leverage it to maximize realization of Growel had streamlined their supply chain
benefits and achieve their business strategies. and achieved a major change in customer and
Growel understood the complexity of problems supplier behavior since Growel could react better
requiring analytic support. More complex issues, to change orders and was able to be more flexible
requiring sophisticated modeling and data analy- in the manufacturing environment. However, in
sis, are better served when analysts and decision the case of Primemover the online visibility of
makers are closely linked which was the case in demand and supplies with customers and suppliers
1243
reduced the number of out-of-inventory events, but factor in this process. Success or failure is often
did not achieve the reduction and streamlining of attributed to enterprise systems or their implemen-
inventories anticipated. The lead times could not tation process. However, it is evident from this
be reduced substantially and delivery performance study that enterprise systems provide a platform
improved marginally. Their data management was of functionalities and information to an organiza-
inconsistent coupled with a lack of clarity on the tion. The ability of an organization to extract value
information required to drive their strategy which from data, distribute results from analysis, apply
were the reasons they could not achieve their full knowledge, and establish decisions for strategic
potential and anticipated objectives. organizational benefits will lead the path towards
business success which would eventually emerge
from the process of ongoing transformations over
cONcLUsION a period of time.
This study has examined the effectiveness of

enterprise systems on organizational functions rEFErENcEs
and processes for realizing business value. It
has highlighted that business results follow in Baskerville, R., Pawlowski, S., & McLean, E.
a culture where the business strategy is clearly (2000). Enterprise resource planning and organi-
articulated and defined. The organization works zational knowledge: patterns of convergence and
out their value creation process identifying the divergence. Paper presented at the International
critical areas that require attention and improve- Conference on Information Systems, Proceedings
ment. They understand the key drivers and have of the 21st International Conference on Informa-
the means to influence those drivers and measure tion Systems, Brisbane, Australia, p. 396-406.
them. They translate their business strategy into (2002, August). Bendoly. (2002). Theory and
departmental or divisional strategies, know what Support for Process Frameworks of Knowledge
is to be achieved, which data needed analysis, its Discovery and Data Mining from ERP Systems.
area of application, and expected outcomes. The Information & Management, 40(7).
organization has a culture that supports decision
makers who have the definition of the informa- Cooke, D. P., & Peterson, W. J. (1998). SAP
tion critical to the success of the enterprise and implementation: Strategies and Results (No.
the means to achieve it by linking data, decisions, 1217-98-RR). New York.
and actions. And, for achieving all of this, the Dalal, N. P., Kamath, M., Kolarik, W. J., &
organization must possess the necessary expertise Sivaraman, E. (2004). Toward an integrated
and skills in the usability of ES and its informa- approach for modeling Enterprise processes.
tion. Quality of data plays a vital role. Communications of the ACM, 47, 83–87.
In order to succeed in today’s competitive doi:10.1145/971617.971620
world, businesses must shift their focus from
improving efficiencies to increasing effectiveness. Davenport, T. H. (1995). SAP: Big Change Comes
Integrated access to pertinent information captured in Big Packages, from http://www.cio.com/ar-
by ES must be available so that effective decisions chive/101595_davenpor.html
can be made towards successfully implementing
Davenport, T. H. (1998). Putting the enterprise
strategies, optimizing business performance, and
into the enterprise system. Sloan Management
adding value for customers. Knowledge is a key
Review, (July-August): 121–131.
1244
Davenport, T. H. (2000). Transforming the Prac- Markus, M., & Tanis, C. (2000). The Enterprise
tice of Management with Enterprise Systems. In Systems Experience - From Adoption to Success.
Mission Critical (p. 203-235). Boston: Harvard In R. W. Zmud (Ed.), In Framing the Domains of IT
Business School Press. Research Glimpsing the Future Through the Past
(pp. 173-207). Cincinnati: Pinnaflex Educational
Davenport, T. H., Harris, J. G., & Cantrell, S.
Resources, Cincinnati, USA.
(2002). The Return of Enterprise Systems: The
Director’s Cut: Accenture Institute for Strategic Newell, S., Huang, J. C., Galliers, R. D., & Pan,
Change. S. L. (2003). Implementing enterprise resource
planning and knowledge management systems
Davenport, T. H., & Prusak, L. (1998). Working
in tandem: fostering efficiency and innovation
Knowledge. Boston, Harvard Business School
complementarity. Information and Organization,
Press.
13, 25–52. doi:10.1016/S1471-7727(02)00007-6
Deloitte, C. (1998). ERP’s Second Wave-Maxi-
Shang, S., & Seddon, P. (2000, August 10-13).
mizing the Value of ERP-Enabled Processes. New
A Comprehensive Framework for Classifying
York: Deloitte Consulting, ISBN 1-892383-42-X.
the Benefits of ERP Systems. Paper presented
Donovan, M. (1998). There is no magic in ERP at the 6th America’s Conference on Information
software: It’s in preparation of the process and Systems, Long Beach, California.
people. Retrieved September 8, 2/7/2002, from
Yang, S., & Seddon, P. B. (2004). Benefits and
http://wwwrmdonovan.com/pdf/perfor_98_9.pdf
Key Success Factors from Enterprise Systems
Donovan, M. (2001). Successful ERP Implementa- Implementations: Lessons from Sapphire 2003.
tion the first time. Retrieved July 25, 2001, from Paper presented at the 35th Australasian Confer-
www.mdonovan.com/pdf/perfor8.pdf ence in Information Systems, Hobart, Australia.
Donovan, M. (2003). Why the Controversy over
ROI from ERP? from www.refresher.com/ar-
chives19.html KEy tErMs AND DEFINItIONs
Hedman, J., & Borell, A. (2002). The impact of Business Process Re-Engineering (BPR):
Enterprise Resource Planning Systems on Organi- Rethinking and redesign of business processes to
zational Effectiveness: An Artifact Evaluation. In achieve performance improvements in terms of
F. F.-H. Nah (Ed.), Enterprise Resource Planning overall cost, quality and service of the business
Solutions & Management (p. 125-142). Hershey, Conceptual Framework: A basic conceptual
London: IRM Press. structure built from a set of concepts to outline
Ittner, C. D., & Larcker, D. F. (2003). Coming Up possible courses of action or to present a preferred
Short on Nonfinancial Performance Measurement. approach to solve a complex research problem.
Harvard Business Review, 81(11), 88–95. Customer Relationship Management
(CRM): Software systems that help companies
Jarra, Y. F., Al-Mudimigh, A., & Zairi, M. (2000). to acquire knowledge about customers and deploy
ERP Implementation Critical Success Factors - strategic information systems to optimize revenue,
The Role and Impact of Business Process Manage- profitability and customer satisfaction
ment. IEEE & ICMIT, 02/2000, 122–127. Enterprise Resource Planning (ERP):
Software systems for business management that
integrates functional areas such as planning,
1245
manufacturing, sales, marketing, distribution, Supply Chain Management (SCM): Soft-

accounting, finances, human resource manage- ware systems for procurement of materials,
ment, project management, inventory manage- transformation of the materials into products, and
ment, service and maintenance, transportation, distribution of products to customers, allowing the
and e-business enterprise to anticipate demand and deliver the
Knowledge Management (KM): The cre- right product to the right place at the right time
ation, organization, sharing, and flow of knowl- at the lowest possible cost to satisfy its customers
edge within and among organizations
Return on Investment (ROI): A performance
measurement used to evaluate the efficiency of ENDNOtEs
an investment. ROI is calculated as the annual
financial benefit (return) after an investment mi-
1
A pseudonym. The name was chosen to
nus the cost of the investment divided by the cost symbolize growth.
of the investment the result being expressed as a
2
A pseudonym. The name was chosen to
percentage or a ratio. symbolize power.
1246
1247
Chapter 5.5
People-Oriented Enterprise
Information Systems
Giorgio Bruno
Politecnico di Torino, Italy
Current notations and languages do not emphasize Enterprise Information Systems (EISs) were first
the participation of users in business processes conceived as systems providing a repository for
and consider them essentially as service provid- business entities and enabling users (subdivided
ers. Moreover, they follow a centralized approach into appropriate roles) to handle such entities.
as all the interactions originate from or end in a Then the process-oriented approach (Dumas,
business process; direct interactions between us- van der Aalst, & ter Hofstede, 2005) has pointed
ers cannot be represented. What is missing from out that business purposes are achieved through
this approach is that human work is cooperative coordinated work to be carried out by means of
and cooperation takes place through structured two kinds of activities: user tasks and automatic
interactions called conversations; the notion of procedures. User tasks are units of work that users
conversation is at the center of the language/action carry out with the help of a graphical interface
perspective. However, the problem of effectively in order to achieve a particular purpose. Placing
integrating conversations and business processes a purchase requisition or filling in the review
is still open and this chapter proposes a notation form for a conference paper are examples of
called POBPN (People-Oriented Business Pro- user tasks. Automatic procedures, instead, ac-
cess Notation) and a perspective, referred to as complish their function without requiring any
conversation-oriented perspective, for its solution. human intervention.
People-Oriented Enterprise Information Systems
However, as the activities are developed sepa- and the requester may withdraw the PR or submit
rately from the processes, current notations and a revised version. In the latter case, the approver
languages, such as BPMN (Object Management is expected to re-evaluate the revised version
Group, 2008), XPDL (Workflow Management Co- and to provide the final outcome (“accepted” or
alition, 2007), and BPEL (OASIS, 2007), consider “rejected”).
business processes essentially as orchestrators of If user tasks are abstracted away, a conversation
work, which accomplish this function by means of consists of a number of interactions organized in
interactions with external services; it makes little a control flow specifying their sequence. Interac-
difference if the interaction is directed to a user tions are defined in terms of three major attributes:
task or an automatic procedure. This approach, name (embodying the interaction meaning, for
referred to as orchestration-oriented perspective, example “request,” “acceptance,” or “rejection”),
makes the representation more homogeneous but direction (i.e., initiator-follower or follower-
at the expense of handling human activities as a initiator) and the business entity conveyed by the
special case of automatic procedures. If people interaction (such as a PR). If the business entities
are involved, they are considered as service pro- are left unspecified, what remains is the flow that
viders. Moreover, each interaction that logically characterizes a particular kind of conversation.
takes place between two users, say, A and B, This flow, referred to as conversation protocol, is a
such as A asking B for the approval of a certain template that can be specialized by providing the
request, is mediated by the process and therefore actual types of the business entities exchanged.
it results in two sequential actual interactions, the Several modeling approaches based on LAP
first between A and the process and the second have been proposed (a short survey is given in the
between the process and B. next section); however, they mainly focus on the
What is missing from the orchestration- nature of the underlying interactions and do not
oriented perspective is that, in most cases, hu- provide an adequate solution to the integration
man work is cooperative and therefore human of conversations and business processes. This
activities are not performed in isolation but in integration is a major purpose of this chapter,
structured frameworks referred to as conversa- which proposes a notation called POBPN (Peo-
tions (Winograd & Flores, 1986). Conversations ple-Oriented Business Process Notation) and a
are the basis of the language/action perspective perspective referred to as conversation-oriented
(Weigand, 2006), or LAP, which was proposed for perspective.
the design of information systems and business In POBPN, the top-level representation of a
processes. According to LAP, EISs are mainly business process clearly separates the contribution
tools for coordination. of the users involved (in terms of their roles) from
An example of conversation is the one govern- that of the automatic activities. The roles partici-
ing the approval of a purchase requisition (PR). pating in the process appear as building blocks and
It takes place between two users, denoted as their refinement (in second-level models) is given
requester and approver: the former is the initiator as a combination of the conversations they are
and the latter is the follower of the conversation. involved in. Such conversations are actualizations
The requester starts the conversation by sending of standard conversation protocols obtained by
a PR to the approver, who evaluates the PR and providing the actual types of the business entities
sends it back to the requester with three alternative exchanged during the interactions. These types
outcomes, that is, “accepted,” “rejected,” or “to be come from an information model that comple-
revised.” In the first two cases, the conversation ments the process model: this way the behavioral
ends, in the third one the conversation continues aspects embodied in a business process are inte-
1248
grated with the informational aspects provided by In particular, BPMN provides building blocks
the companion information model. The automatic to represent the actual user tasks. Such building
activities are meant to support the conversations blocks can be placed in swim lanes associated
for which the process itself plays the initiator role with the roles needed; from an operational point
or the follower one; in the top-level representation of view, they enable the process to activate a user
of a business process, they are represented by a task, by sending it input information along with
specific building block, called System. the indication of the actual performer(s) required,
The RAD approach (Ould, 2005) is based on and then to wait for output information, which
interacting roles as well, but these roles encompass notifies the completion of the user task.
tasks and control flow activities, while, in POBPN, This approach, being an attempt at incorporat-
roles are based on conversations. An extension of ing user tasks in the process structure, does not
RAD towards LAP has been proposed (Beeson & cope with all the situations taking place in prac-
Green, 2003): the basic suggestion is to introduce tical applications. It only covers one interaction
a middle-level construct in roles, with the purpose pattern between a process and a user task, that
of grouping the tasks that are logically involved is, the one in which the user task receives one
in a conversation. input message (on its activation) and delivers one
This chapter is organized as follows. Section output message (on its completion); this pattern
2 discusses the major limitations of the orchestra- is referred to as (1, 1). There are three more task
tion-oriented perspective and reports on current interaction patterns, as follows.
research on conversations. Section 3 introduces Pattern (1, *) indicates that the task sends a
conversation protocols, whereas sections 4 and number of intermediate output messages before the
5 illustrate the proposed notation, POBPN, with completion one. These additional output messages
the help of an example. Section 6 focuses on the signify that the task has produced information
differences between the conversation-oriented items requiring immediate attention. For example,
perspective and the orchestration-oriented one. task “reviewPapers” is started on the arrival of
Section 7 presents the conclusion and the future a folder of papers to be reviewed and allows the
work. reviewer to release the reviews one by one.
Pattern (*, 1) indicates that the task may receive
additional input messages, after its activation. For
bAcKGrOUND example, task “assignPapers” is activated with an
initial group of papers and then it may receive
Business processes can be addressed with dif- additional papers, before its completion.
ferent orchestration-oriented notations, such as The fourth pattern (*, *) denotes both a flow
BPMN (Object Management Group, 2008) and of input messages and a flow of output ones. For
UML activity diagrams (Object Management example, task “evaluatePapers” receives a flow of
Group, 2007); however, they all have a number of reviews and provides a flow of papers evaluated,
features in common, as follows. They place great each of which will trigger an immediate notifica-
emphasis on the control flow of business processes, tion to its author.
under the pressure of research on workflow pat- To overcome the above-mentioned limitations,
terns (van der Aalst, ter Hofstede, Kiepuszewski, the process, instead of incorporating the user
& Barros, 2003); in contrast, they overlook the tasks in its control flow, should emphasize the
information flow (although information items can underlying interactions taking place with them.
be included for documentation purposes) and tend The notion of interaction is important because
to incorporate user tasks as execution steps in the it leads to a clear separation between processes and
process structures.
1249
user tasks; however, interactions are part of larger control-flow logic handling their ordering. In the
structures encompassing all those exchanged be- hierarchical structure provided by DEMO, tasks
tween two parties for a common purpose. At this are compared to atoms, transactions to molecules,
point, the notion of conversation comes into play. and business processes to fibers (Dietz, 2003).
Winograd and Flores (1986) introduced the Transactional roles might not have a one-to-one
term “conversation for action” to indicate the mapping to organizational roles, because the
kind of conversation aimed at producing an ef- tasks included in a transactional role might be
fect on the real world through the cooperation of assigned to different organizational roles: this is
two parties. Taking advantage of the theory of needed to support delegation (i.e., the situation in
speech acts (Austin, 1976), which studies how which the actor who receives the order is not the
people act through language, they proposed a one who delivers the result) but could make the
novel perspective, referred to as language/action model more difficult to understand.
perspective (LAP), for the design of information BAT (Goldkuhl & Lind, 2004) draws upon
systems and business processes. According to the ideas of Action Workflow and DEMO and
LAP (Weigand, 2006), information systems are addresses business interactions in an inter-organi-
mainly tools for coordination. Conversations may zational framework. In particular, it analyzes the
be carried out for different purposes: Winograd relationships between a supplier and a customer
(1987) mentions conversations for clarification, in a number of situations including single busi-
for possibilities and for orientation in addition to ness transactions and frame contracting (with
conversations for action. embedded transactions).
Several modeling approaches based on LAP Although the above-mentioned LAP approach-
have been proposed, among which stand out Ac- es yield a deeper understanding of the nature of
tion Workflow, DEMO, and BAT. interactions, they do not provide an operational
In the Action Workflow approach (Medina- solution to the integration between business
Mora, Winograd, Flores, & Flores, 1992), a typical processes and user tasks through conversations.
conversation takes place between a requester and This integration is pursued by the conversation-
a performer and is made up of four major phases, oriented perspective illustrated in the next sections
that is, request, commitment, performance, and along with the related notation, POBPN.
evaluation, forming the so-called workflow loop.
The first two phases establish a commitment be-
tween the parties and the last two phases bring cONVErsAtION PrOtOcOLs
the parties to an agreement on the result. The
actual workflow might be complicated because This section illustrates the notion of conversa-
negotiations can be made in each phase. tion protocol and shows the notation adopted in
In the DEMO approach (Dietz, 2006), work- POBPN, with the help of a number of examples
flow loops are subdivided in three phases, that to be reused in the next sections.
is, order, execution, and result, and are encap- A conversation implies a flow of interactions
sulated in transactions. Business processes are between the parties, an interaction being the com-
compositions of roles and transactions, where munication of one party’s intention to the other
transactions connect two transactional roles. party. Intentions can be categorized and identi-
Transactional roles are compound entities includ- fied by specific symbols, as illustrated later on in
ing transactional tasks (i.e., the tasks being part this section. Examples of intentions are making
of the transactions the role is involved in) and the a request, accepting a request, and providing a
1250
Figure 1. Examples of conversation protocols

r, T
alt r+, T r, T1
n, T
r-, T alt
r-, T1
r~, T NotificationCP
rep, T2
alt -r, T
opt
opt ~r, T
~r, T ~rep, T2
alt r+, T r~, T
rep~, T2 alt
r-, T n, T
ApprovalCP BasicCfACP Notification1CP
reply. An interaction conveys the business entity interactions, while providing a simple hierarchi-
that is the object of the intention. This business cal structuring mechanism based on fragments.
entity is referred to as the business content of the Four examples of protocols are shown in Figure 1.
interaction. As a general rule, the lifelines of the two ac-
Conversations can be categorized on the basis tors involved in the conversation are not shown;
of their purpose: for example, there are conversa- however, the initiator is meant to be located on
tions for approval and conversations for action. If the left side of the diagram and the follower on the
the details of the business content are ignored, all right side. The first interaction of a conversation
the conversations with the same purpose turn out is called initial interaction and the business entity
to have the same interaction flow, and the structure that it conveys is referred to as initial business
of this flow, called conversation protocol, is the entity. Timing constraints are not considered so
major concern of this section. as to keep the examples reasonably simple.
A conversation protocol is an abstraction as it is The first protocol in Figure 1 addresses con-
defined independently of the types of the business versations for approval and is called “Approval
entities exchanged by the parties; it is a template conversation protocol” (ApprovalCP in short).
in which such types are given generic names. These conversations start with a request, indicated
When a protocol is actualized, generic types will by symbol “r”; the business content, that is, the
be replaced with actual types: as will be shown in entity for which the approval is asked, is denoted
the next section, an actualized protocol relies on by a generic type, such as “T.” The follower can
an information model providing the definition of accept the request (r+), reject it (r-), or ask for
the types of the business entities involved. some improvement (r~). The three alternatives are
State models are often used (Winograd, 1987) enclosed in a fragment marked with keyword “alt.”
to define conversation protocols graphically; PO- In the first two cases, the conversation is ended; in
BPN, instead, adopts UML sequence diagrams the third one, the initiator can then withdraw the
(Object Management Group, 2007), because request (-r) thus ending the conversation or he or
they focus on the essential elements, that is, the she can submit a revised version (~r), which will be
followed by an acceptance (r+) or a rejection (r-).
1251
There is only one generic type involved, that worked on in this section and in the following
is T, because the business entities exchanged ones. The process is referred to as PRBP and its
during this kind of conversation are meant to be requirements are as follows.
all of the same type. PRs are submitted by users entitled to do so;
Protocol BasicCfACP defines simple conver- such users play the Requester role with respect
sations for action. As in the first example, the to process PRBP. When a PR is submitted, it
conversation is started with a request (r) coming may be automatically approved or it may be sent
from the initiator. The follower may reject the for evaluation to the appropriate supervisor. The
request (r-) or may provide a reply. The interaction first case occurs when the amount of the PR is
conveying the reply is indicated with a specific less than the personal expenditure threshold as-
keyword, “rep.” It is accompanied by a different sociated with the requester; in this case, the PR
business entity, denoted by generic type T2, be- is considered to be an “inexpensive” one. When
cause the purpose of a conversation for action is the second case occurs, the (expensive) PR is di-
to make the follower deliver a new entity as the rected to the requester’s boss (who becomes the
result of the action caused by the initiator’s request. supervisor involved). The requester is informed of
Upon receiving the reply, the initiator may then the result (acceptance or rejection). The supervi-
ask for a clarification (~rep) to be followed by a sor may also ask for some improvement and the
clarification (rep~) after which the conversation requester may provide a revised version of the PR
is ended. The sequence of interactions ~rep and or may withdraw it.
rep~ is optional and, therefore, it is enclosed in a The PRs approved are sent to the purchasing
fragment marked with keyword “opt”. department where they are handled by a number
The protocol of conversations for action can of equivalent buyers. Buyers are assumed to pick
become more complicated (Winograd, 1987), if pending PRs from a common queue and to include
counter-offers are included. them in purchase orders. In case a buyer has
Protocol NotificationCP consists of one in- some doubts, he or she can ask the PR requester
teraction (n), interpreted as a notification (i.e., a to make the necessary clarification; when a buyer
simple communication requiring no reply). has included the PR in a purchase order, he or she
Protocol NotificationCP1 ends with a notifi- sends a confirmation to the PR requester.
cation, which may be preceded by an optional Sometimes supervisors may ask one or more
sequence consisting of a request for clarification reviewers to provide third-party reviews of PRs
(~r) followed by a clarification (r~). considered to be critical. How this is done can be
subjected to different rules: in this case study, the
conversation related to a single review is expected
PEOPLE-OrIENtED bUsINEss to conform to conversation protocol BasicCfACP
PrOcEssEs shown in Figure 1. In addition, if a supervisor
has started a review conversation for a certain
As mentioned in the introduction, the conversa- PR, he or she must wait for its conclusion before
tion-oriented perspective places great emphasis starting a new review conversation or taking the
on the participation of users in business processes, final decision, for the same PR.
and the notation proposed, POBPN, aims at provid- The analysis of the requirements above leads
ing an effective integration between conversations to the discovery that five roles are involved. Four
and business processes. To clarify the features of them are user roles: Requester, Supervisor,
of POBPN, an example concerning a business Reviewer, and Buyer. The fifth role involved,
process handling purchase requisitions (PRs) is the System role, is responsible for all the activi-
1252
ties to be performed automatically (such as the be illustrated later on in this section), and they can
approval of inexpensive PRs). These roles take be adequately expressed only if an information
part in several conversations, as follows. For the model is provided.
approval of a PR, its requester interacts with either The companion information model for pro-
his/her supervisor or the System; to get support cess PRBP is shown in Figure 2 as a UML class
for a critical PR, a supervisor may interact with model (Object Management Group, 2007). Only
a number of reviewers; when processing a PR, the major items are considered, for the sake of
a buyer interacts with the PR requester (at least simplicity.
to notify him/her that the PR has been included Users are represented by classes marked with
in a purchase order). In addition, since approved stereotype <<role>>. Such classes are not meant
PRs need to be processed by the purchasing to be the actual ones, because the decision on
department, notifications are sent by the System how to represent users in the information system
to the buyers. is postponed to the software development phase;
The above-mentioned four conversations fit they can be considered as interfaces to the actual
the patterns presented in the previous section; classes.
however, it is necessary to indicate which busi- The relationship between role Requester and
ness entities are involved. An information model role Supervisor is a consequence of the require-
is then needed, which defines the business entity ment that an expensive PR has to be approved by
types along with their most relevant attributes the requester’s boss who acts as the supervisor
and relationships. This information model is the for the PR. Associative attribute “boss” enables
companion of the process model for two major the navigation from a requester entity to the cor-
reasons: firstly, the actualization of conversation responding supervisor entity.
protocols into actual conversation types is car- A PR is related to one requester entity (which
ried out by replacing the generic types used in can be reached through associative attribute
the protocols with the actual types provided in requester) and possibly to one buyer entity (only
the information model. Secondly, in the process in case the PR has been approved). A PR may
model there are a number of data-driven aspects, also be linked to one purchase order (in case it
such as the selection of a conversation follower (to has been approved), to a number of reviewers, as
Figure 2. The companion information model of process PRBP

1 *
<<role>> <<role>>
Supervisor boss Requester
1 requester
*
0..1 + 1 *
POrder PR Review
pr
* * *
1
<<role>> * <<role>>
Buyer 0..1 Reviewer
reviewers
Attributes
PR: float amount. Requester: float threshold.
1253
well as to a number of reviews. Purchase Orders The approach adopted in POBPN is to repre-
are represented by class POrder. sent participating roles as the top-level building
A review is linked to the reviewer entity blocks (referred to as role participations or simply
representing the reviewer who has provided it. participations) of the process diagram and to show
Associative attribute “pr” provides the PR as- the conversations involved as ports, depicted as
sociated with a given review. small squares on the participation icons. Each
Relationships PR-Review and PR-Reviewer conversation gives rise to two ports, one in the
are both needed in that a reviewer could refuse to initiator role (referred to as initiator port) and the
provide the review he or she has been asked for. other in the follower role (follower port). Such ports
Associative attribute “reviewers” provides the list are given identical names but different colors;
of Reviewer entities associated with a given PR. initiator ports are white, while follower ones are
Two relevant attributes are also shown in Figure grey. Ports are connected through labeled links,
2, because they are mentioned in the requirements whose details are shown in the Links section.
of PRPB: they are the amount attribute in PR enti- The port names are names of conversation types.
ties and the threshold (for expenditure) attribute Section “Conversation types” declares con-
in Requester entities. versation type “a” to be based on conversation
On the basis of the analysis made so far, the protocol ApprovalCP with generic type T replaced
model of process PRBP can be worked out. Its top- with entity type PR, while conversation types
level portion is presented in Figure 3 and consists “b,” “c,” and “v” are actualizations of protocols
of three parts: the process diagram, the section NotificationCP, Notification1CP and BasicCfACP,
defining conversation types and the one (Links) respectively.
showing how the followers of the conversations Section “Links” contains a number of link
will be determined. definitions, whose purpose is to show how the
Figure 3. The top-level representation of process PRBP

Conversationt ypes:
a: ApprovalCP (T = PR); b: NotificationCP (T = PR);
c: Notification1CP (T = PR); v: BasicCfACP (T1 = PR, T2 = Review);
l1 l3
a a Supervisor v v Reviewer
l2
Requester a System b
l4
l5
c c Buyer b
Links:
l1 (PR pr): [pr.amount > pr.requester.threshold] pr.requester.boss.
l2 (PR pr): [pr.amount <= pr.requester.threshold].
l3 (PR pr): pr.reviewers.last(). l5: (PR pr): pr.requester.
1254
followers of the corresponding conversations are needed. In general, when the follower clause is
determined. missing and the follower role is a user role, such
A link definition consists of four elements: as Buyer in link l4, the assignment of a particular
name, formal parameter, guard (included between follower will take place at run-time; it is up to
square brackets), and follower clause. The formal the members of the role to decide which of them
parameter refers to the initial business entity, that will follow a new conversation, as it happens with
is, the entity conveyed by the initial interaction. buyers who autonomously pick pending PRs from
In all the definitions shown, the initial business a common pool. In general, the selection of the
entity is a purchase requisition (whose type is PR) performer is a critical issue and several patterns
and the formal parameter introduces the name, have been proposed (Russell, van der Aalst, ter
that is, pr, used (in the guard and in the follower Hofstede, & Edmond, 2005).
clause of the link definitions) to refer to this entity. The follower clause in link l3 specifies that
A conversation of type “a” is initiated by a the reviewer to be involved is the last added to
requester and may be followed by a supervisor the collection of reviewers associated with the PR
or by the System. There are two links in compe- that started the conversation. Link l5 indicates that
tition, l1 and l2, because they have the initiator the follower is the requester of the current PR.
port in common: the choice is made on the basis Process diagrams are mainly architectural
of their guards. models, as they show the roles participating
The guard of l1, “pr.amount > pr.requester. in the process and the conversations that may
threshold,” consist of two terms, “pr.amount” take place between them. Participation building
and “pr.requester.threshold.” The evaluation of blocks can be refined into second-level models,
the first term returns the amount of the current called participation models, and the next section
PR, and the evaluation of the second one gives is dedicated to their illustration.
the personal expenditure threshold of the PR
requester: if the guard is true, the follower will
be the requester’s boss. PArtIcIPAtION MODELs
In most cases, the follower of a conversation is
represented by a business entity related, through a Participation models define the involvement of the
certain path of associations, to the initial business users (identified by their roles) or the contribu-
entity: then, the follower clause is needed and it tion of the System, in the business process being
consists of a navigational expression indicating considered. Those related to users are called user
how to reach the follower business entity from the participation models and fulfill two major aims:
initial one. Two associative attributes are involved firstly, by looking at them, users get an insight
in the follower clause of l1, that is, requester and into the overall sequence of interactions they will
boss: in the information model shown in Figure be involved in; secondly, they are the basis of the
2, they determine a path from a PR to a Supervi- implementation of the actual user tasks.
sor entity. The System participation model (there is at
In the definition of link l2, the follower clause most one in a business process) is mainly meant
is missing, because the follower role is System: to handle the automatic activities.
when the guard of l2 is true, the conversation is Both types of models are described in this
followed by automatic activities and no supervi- section and the examples given refer to process
sor is involved. PRBP: its user participation models are shown
The definition of link l4 is missing because in Figure 4.
both the guard and the follower clause are not
1255
Basically, user participation models are com- participation model. According to the PRBP
binations of conversations. For this reason, their requirements presented in the previous section, a
structure is like the one of conversation protocols supervisor, during conversation “a,” and precisely
with interactions replaced by references to con- after receiving a new PR or a revised one, may
versation types. Such references are placed in carry out a number of sequential conversations
constructs similar to the “InteractionUse” items “v.” What is needed is the possibility of including
available in UML sequence diagrams (Object a number of conversations “v” into a conversation
Management Group, 2007); however, in POBPN “a.” The original conversations are not modified,
they are called “ConversationUse” items, since but they are combined so as to define the overall
their meaning is different. In these items, the con- behavior of the role. POBPN supports the nesting
versation type name appears in a small rectangle of ConversationUse items, as follows. Conversa-
at the upper left corner: the rectangle is white or tionUse items are not black boxes, but they can
grey depending on whether the current role is the include one or more extensions. An extension is
initiator or the follower, respectively. a partition made up of two parts: the first part
The Reviewer participation model is based only shows the interaction(s) preceding the extension
on conversation “v,” and reviewers are followers (referred to as extension points), and the second
of such conversations. part contains the conversations (represented by
The Buyer participation model is a simple their ConversationUse items) to be carried out
sequence of conversations “b” and “c”; buyers before resuming the enclosing conversation. If
are followers of conversations “b” and initiators the enclosed conversations are subjected to con-
of conversations “c.” straints, they are placed in appropriate fragments.
The Requester participation model is based on The Supervisor participation model extends con-
conversation “a,” which, if it ends successfully versation “a”: the extension points are interactions
(i.e., if it ends with r+), is followed by conversa- “r” or “~r” (referring to the receipt of a new PR or
tion “c.” In the optional fragment, operator opt of a revised one, respectively), which are followed
is followed by an informal expression describing by a loop fragment including conversation “v.”
when conversation “c” is carried out. The loop operator introduces a sequence (possibly
An important feature of user participation empty) of conversations “v.”
models is the ability of structuring conversa- The System participation model addresses
tions hierarchically, as shown in the Supervisor those conversations for which the process itself
Figure 4. The user participation models in process PRBP

Reviewer Buyer RequesterS upervisor
v b a a
r, ~r
c opt:
a ended with r+ loop
v
c
1256
Figure 5. The System participation model in process PRBP
PR PR
a.r Supervisor.a.r+
p1 p2
t1 t2
a.r+ b.n b.n
plays the initiator role or the follower one. In PRBP defined in the companion information model of
there are two such conversations: the process is the process being considered; the one related to
the follower of conversations “a” for inexpensive process PRBP is shown in Figure 2. This solution
PRs and the initiator of conversations “b.” The eliminates the need for complex data patterns
related interactions must be dealt with by means (Russell, ter Hofstede, Edmond, & van der Aalst,
of automatic activities and, moreover, control-flow 2005) as all the data needed are associated with
logic is needed to enforce ordering constraints the tokens: the information flow is integrated with
between the activities. the control flow.
The notation used in System participation Places may have input arcs and output ones:
models is different from the one adopted in user the input arcs of a place come from its input
participation models and follows an orchestration- transitions and the output ones are directed to its
oriented perspective. output transitions. Transitions have input arcs and
The orchestration-oriented notation used in output ones, as well: the input arcs of a transition
POBPN is an extension of colored Petri nets come from its input places and the output ones
(Kristensen, Christensen, & Jensen, 1998) and are directed to its output places.
is called “interaction nets” to emphasize that its Tokens are added to places by their input transi-
major purpose is to handle interactions, on which tions and are removed by their output transitions;
orchestrations are based. they can also be added by followed interactions
This article does not provide a formal defini- as described later on in this section.
tion of interaction nets; instead, it provides two Transitions are the automatic activities per-
examples, the first of which, shown in Figure 5, formed by the process; they are token-driven as
defines the System participation model in process they are activated when the proper tokens are
PRBP. The second example is given in the next available in their input places. The basic rule is that
section. a transition fires as soon as there is one token in
Interaction nets are made up of places, transi- each of its input places. When it fires, a transition
tions, arcs and interaction symbols. takes the input tokens (i.e., it removes the activat-
Places are containers of tokens, which refer to ing tokens from the corresponding input places),
business entities, such as PRs or reviews. They performs an action (which can operate on the
have a name and a label (or place type): the former business entities referred to by the input tokens),
is written in the circle representing the place and and delivers the appropriate (output) tokens to its
the latter outside the circle and close to it. The output places. When there is a correspondence
label is the name of the type of the business enti- between the types of the input places and those of
ties that are referred to by the tokens contained the output places, tokens are simply moved from
in the place. The business entity types are those the input places to the output ones: this behavior
1257
is referred to as automatic propagation rule. In the place p1 whose type, PR, coincides with the type
other cases, the output tokens must be generated of the incoming business entity. Transition t1 is
in the action. then triggered and its action consists in performing
The behavior of transitions can be comple- two interactions, “a.r+” and “b.n.” On the basis
mented by a textual description that is made up of of the automatic propagation rule, the business
two sections, that is, the selection rule (introduced entity taken from place p1 is sent through output
by keyword “s”) and the action (introduced by interactions “a.r+” and “b.n,” without the need of
keyword “a”). These sections may include opera- writing any code in the action of t1. The recipient
tional code, but the examples shown only include of interaction “a.r+” is the one that started con-
informal descriptions, for the sake of simplicity. versation “a,” whereas the recipient of interaction
If a transition is provided with a suitable selec- “b.n” is determined by link l4 shown in Figure
tion rule, it can take a variable number of tokens 3, because this interaction is the initial one of a
from its input places and the selection can be based conversation “b.”
on the contents of the business entities linked to Monitored interactions are those interactions
the tokens. If the selection rule is omitted, the the process has to be aware of so as to take ap-
basic rule is applied. propriate actions, although it is not their follower.
Interaction symbols are depicted as labeled They are shown dashed and their labels include the
arcs, which may be either input arcs of places or initiator role before the conversation type name
output arcs of transitions. They represent ordinary and the interaction name. One such interaction
interactions or monitored ones. appears in the right part of Figure 5: its purpose
Ordinary interactions are those in which the is to enable the activation of conversation “b”
process is directly involved and may be divided when conversation “a” for an expensive PR ends
into followed interactions and initiated ones. Their successfully. Conversations “a” for expensive
labels, such as “a.r,” consist of the name of a con- PRs are handled by supervisors, but they are not
versation type and an interaction name separated supposed to decide what has to be done after the
by a dot. The conversation types correspond to successful approvals of PRs. It is up to the process
the names of the ports that appear in the System to trigger the subsequent processing of approved
block in the process diagram. PRs; therefore, it must be informed of the suc-
Followed interactions are depicted as input cessful conclusion of a conversation “a” so as to
arcs of places, because the effect of a followed start a conversation “b” with the buyers—these
interaction is to add a new token (referring to two conversations operate on the same PR.
the business entity conveyed by the interaction) Monitored interactions can only provide to-
to its output place. Initiated interactions are de- kens to the process in that they are used to trigger
picted as output arcs of transitions: when a token complementary behavior.
flows along such an arc (this occurs by virtue of
the automatic propagation rule or by an explicit
command issued in the input transition) the cor- cOMPArIsON OF NOtAtIONs
responding interaction is started and the follower
is determined on the basis of the links defined in The aim of this section is to provide a comparison
the top-level process model. between the notations used in the conversation-
Ordinary interactions appear in the left part of oriented perspective and in the orchestration-
Figure 5. When an inexpensive PR is submitted oriented one.
by a requester, interaction “a.r” is performed and In the former approach, the emphasis is placed
its effect is to add a token (referring to the PR) to on the conversations between users and on how
1258
Figure 6. The conversation-oriented interaction flow for the scenario proposed

RS ystemS VB
a.r
v.r
v.rep
a.r+
b.n
c.n
they are organized in participations: the explicit cess starts a conversation between R and S in
contribution of the process is restricted to the conformity to link l1, and the initial interaction
definition of the automatic activities, as shown takes place. Its label, “a.r,” is the name of a con-
in Figure 5 for process PRBP. versational interaction, as the interaction name is
In the orchestration-oriented perspective, preceded by the name of the conversation type
processes are thought of as explicit orchestrators the interaction belongs to.
of all the activities, both the automatic ones and In the participation model of the supervisor
the user tasks. Conversations are not a first-class shown in Figure 4, interaction “a.r” is an exten-
notion of the approach; therefore, the process has sion point that can be followed by a number of
to make up for the lack of conversation-oriented conversations “v” with reviewers. In the scenario
constructs by directly handling the corresponding above, supervisor S asks reviewer V to provide
flow of interactions. a review of the PR and this leads to the pairs
Conversation-oriented notations provide a of interactions “v.r” and “v.rep” shown in the
higher-level representation than orchestration- sequence diagram. Then the PR is approved by
oriented ones and the difference between them supervisor S resulting in interaction “a.r+.” This
can be easily recognized on the basis of a simple interaction is a monitored one as indicated in the
scenario, as follows. The scenario refers to a par- System participation model shown in Figure 5,
ticular occurrence of process PRBP involving four and therefore it triggers interaction “b.n” between
users: requester R, his/her supervisor S, reviewer the System and buyer B. The dashed arc con-
V and buyer B. Requester R submits an expensive necting interaction “a.r+” and interaction “b.n”
PR, which is sent to S for evaluation. S asks the emphasizes that “a.r+” is a monitored interaction
support of V and then approves the PR. Buyer B and “b.n” is a consequence. Buyer B is supposed
receives the PR from the process and then notifies not to need any additional clarification from the
R of its inclusion in a purchase order. requester and therefore the scenario ends in no-
On the basis of the scenario proposed, process tification “c.n” sent by B to requester R.
PRBP shown in Figure 3 gives rise to a number With an interaction-oriented notation, the
of interactions that are graphically represented sequence diagram for the same scenario becomes
in the sequence diagram presented in Figure 6. more complicated, because interactions between
The interaction flow develops as follows. When users (from an initiator to a follower) are mapped
requester R submits an expensive PR, the pro- to pairs of sequential interactions involving the
1259
Figure 7. The orchestration-oriented interaction flow for the scenario proposed

RS SV B ystem
(R)a_r
(S)v_r a_r(S)
v_r(V)
(V)v_rep
v_rep(S)
(S)a_r+
a_r+(R)
b_n(B)
(B)c_n
c_n(R)
process: the first interaction takes place between an example, the orchestration-oriented version
the initiator and the process, the second one be- of process PRBP is shown in Figure 8. It is the
tween the process and the follower. second example based on the interaction nets
The orchestration-oriented sequence diagram introduced in the previous section, and includes
related to the scenario proposed is shown in the description of a number of transitions.
Figure 7. To reduce the complexity of the model, a
For example, interaction “a.r” from requester number of notational simplifications have been
R to supervisor S is mapped to interaction “(R) introduced, as follows.
a_r” from R to the process, and to interaction If there are two transitions in series (Murata,
“a_r (S)” from the process to S. The name of the 1989), the intermediate place (together with its
interaction is obtained from that of the correspond- input arc and its output one) can be replaced with
ing conversational interaction by replacing the a link that directly connects the two transitions
dot operator with character “_”. In addition, in a and is labeled with the name and type of the place
user-process interaction, the initiator is indicated eliminated. Examples of such links are the one
between parentheses before the interaction name, connecting transitions tb1 and tb2 and the one
whereas, in a process-user interaction, the fol- connecting tv1 and tv2.
lower is indicated between parentheses after the If a place receives tokens only from a followed
interaction name. The labels of the process-user interaction and it is connected only to one transi-
interactions are shown on the right of the System tion, then it can be eliminated and the followed
lifeline, while those of the user-process interac- interaction can be directly connected to the transi-
tions are shown above the corresponding lines. tion: this simplification is carried out with several
The notion of monitored interaction is not transitions, for example, tb1 and tb2.
needed, as the process is an explicit orchestrator The third kind of simplification makes use of
of all the interactions. ε-arcs: an ε-arc links two places of the same type
A business process defined with an interaction- and its effect is that the tokens included in the
oriented notation corresponds to a POBPN process source place also belong to the destination place
that only includes the System participation model, and can trigger its output transitions (the vice
as all the interactions between users are mapped versa does not hold); if two places, say p1 and p2,
to pairs of interactions involving the process. As are the source and the destination, respectively,
of an ε-arc, p1 is said to be joined to p2.
1260
Figure 8. The orchestration-oriented model of process PRBP

(R)a_r
a_r(S)
p1 ts1
PR
a_r+(R) PR
PR
t1
ps1 pv1
b_n(B)
t2 p2
PR (S)a_r~ (S)v_r
PR ts2 tv1
pb1 a_r~(R) v_r(V)
PR pv2, PR
(B)c_~r (R)a_-r (V)v_rep
ts3 ps4 tv2
c_~r(R) tb1 a_-r(S) v_rep(S)
PR
pb2, PR (R)a_~r PR
(R)c_r~ ts4 ps2 pv3
tb2 a_~r(S)
c_r~(B) (S)a_r- (S)v_~rep
PR pb3 tv3
ts5 ps3 v_~rep(V)
a_r- (R) pv4, PR
PR (V)v_rep~
(B)c_n
(S)a_r+ tv4
tb3 v_rep~(S)
ts6
c_n(R) a_r+ (R) PR pv5
The interaction labels shown in Figure 8 follow Place ps1 has two outgoing flows. The one
the same convention adopted in Figure 7. on the left handles the case in which the super-
The meaning of the model shown in Figure 8 visor requests some modifications to the PR:
is as follows. Submitted PRs enter place p1 and then ts2 fires and puts the PR in place ps4 until
then trigger either transition ts1 or transition t1 the requester makes his/her move, which can be
depending on which selection rule is satisfied. The the withdrawal of the PR or the submission of a
incoming PR is associated with the input token revised version: these moves cause ts3 or ts4 to
taken from place p1 and the selection rules and fire, respectively. The firing of ts4 puts the PR in
the actions of transitions ts1 and t1 use the place place ps2, where the left path ends. The left path
name (i.e., p1) to refer to this PR. is optional, because ps1 is joined to ps2 and ps2 is
Transition ts1 fires in case of an expensive PR, joined to both ps3 and pv1. Therefore the super-
as shown in its selection rule: its action consists visor may immediately approve or reject the PR
in initiating interaction “a_r (S)” with the proper (from place ps3) or ask for external support (from
supervisor. The follower-selection rules, which in place pv1). If the supervisor accepts or rejects the
POBPN are presented in section “Links” of the PR, ts6 or ts5 will fire, respectively; the firing of
top-level process model, with this notation, must ts6 puts the PR in place p2 where the interaction
be embedded in transitions. Primitive “set” is used flow involving the buyer starts.
to establish the follower of an initiated interaction. The path starting at place pv1 handles the in-
In the action of ts1, the boss of the requester of teractions with the reviewers. Transition tv1 fires
the input PR is set as “S”, because S denotes the if the supervisor requests a reviewer’s support and
follower of interaction “a_r”. initiates interaction “v_r” with the reviewer; the
Transition ts1 puts the PR into place ps1 indi- reviewer is the one represented by the last reviewer
cating that the PR is waiting for the supervisor’s entity added to the PR, as shown in the action of
move. tv1. The PR is then put in the intermediate place
1261
included in the link between transition tv1 and oriented one. The former is a centralized approach
transition tv2 (connected in series) and waits for in which all the interactions are explicitly medi-
the arrival of the corresponding review. When ated by the process. If a user looks at a process
this happens, tv2 fires: it relays the review to the description, he or she cannot immediately identify
supervisor (i.e., the boss of the PR requester, as the interactions he or she will be involved in, as
indicated in the action) and puts the PR in place all user activities are intermingled. In the ap-
pv3. Place pv3 is joined to place ps3 and place proach proposed, instead, conversation models
pv1, as the supervisor, at this point, may issue come first (from the actualization of standard
the final decision or ask for another review. In conversation protocols) and the process model is
addition, the supervisor may ask the reviewer built upon them. The explicit contribution of the
for some clarification: for this reason, place pv3 process is restricted to those conversations for
is also followed by transition tv3, which relays which the process itself plays the initiator role
the clarification request to the reviewer. The path or the follower one.
goes on with transition tv4, which is triggered by Current work is going on in two directions.
the reply from the reviewer and relays it to the One line of research is concerned with the inclu-
supervisor; the PR is then put in place pv5 joined sion of inter-organizational conversations. For
to places ps3 and pv1. this reason, the similarities between conversation
Place p2 starts the interaction flow with a protocols and B2B collaborations or choreogra-
buyer. Transition t2 is meant to assign the PR to phies need to be thoroughly analyzed.
a buyer. The PR is then put in place pb1, which The other line of research is more pragmatic
has two outgoing paths. The path on the left is and addresses the automatic mapping of POBPN
optional: it enables the buyer to ask some clari- models to orchestration-oriented ones; the pur-
fication of the requester and ends in place pb3 to pose is to take advantage of current technology
which place pb1 is joined. Transition tb3 waits while preserving the conceptual strength of the
for the confirmation from the buyer and relays it conversation-oriented perspective.
to the requester.
rEFErENcEs
cONcLUsION
Austin, J. L. (1976). How to do things with words.
This article has proposed a notation, POBPN, and Oxford, UK: Oxford University Press.
a conversation-oriented perspective to effectively
Beeson, I., & Green, S. (2003, April 9-11). Using
integrate conversations and business processes in
a language action framework to extend organiza-
an intra-organizational context. The involvement
tional process modelling. Paper presented at the
of users in business processes is clearly shown
UK Academy for Information Systems Confer-
by means of participation building blocks appear-
ence, University of Warwick.
ing in the top-level representation of a business
process. Such participation items can be refined Dietz, J. L. G. (2003). The atoms, molecules
into second-level models which are basically and fibers of organizations. Data & Knowledge
combinations of conversation models. Engineering, 47(3), 301–325. doi:10.1016/S0169-
An example has been presented in order to 023X(03)00062-4
stress the differences between the orchestration-
oriented perspective (followed by most of current
notations and languages) and the conversation-
1262
Dietz, J. L. G. (2006). The deep structure of busi- Object Management Group. (2007). Unified mod-
ness processes. Communications of the ACM, eling language: Superstructure, V.2.1.1. Retrieved
49(5), 59–64. doi:10.1145/1125944.1125976 September 29, 2008, from http://www.omg.org/
docs/formal/07-02-03.pdf
Dumas, M., van der Aalst, W. M. P., & ter Hofst-
ede, A. H. M. (2005). Process-aware information Object Management Group. (2008). Business
systems: Bridging people and software through process modeling notation, V.1.1. Retrieved
process technology. New York: John Wiley & September 29, 2008, from http://www.bpmn.org
Sons.
Ould, M. (2005). Business process management:
Goldkuhl, G., & Lind, M. (2004). The generics A rigorous approach. Swindon, UK: The British
of business interaction - emphasizing dynamic Computer Society.
features through the BAT model. In M. Aakhus
Russell, N., ter Hofstede, A. H. M., Edmond, D.,
& M. Lind (Eds.), Proceedings of the 9th Con-
& van der Aalst, W. M. P. (2005). Workflow data
ference on the Language-Action Perspective on
patterns: Identification, representation and tool
Communication Modelling (LAP 2004) (pp. 1-26).
support. In L. Delcambre (Ed.), Lecture Notes
New Brunswick, NJ: Rutgers University.
in Computer Science, 3716 (pp. 353-368). Berlin,
Kristensen, L. M., Christensen, S., & Jensen, K. Germany: Springer.
(1998). The practitioner’s guide to coloured Petri
Russell, N., van der Aalst, W. M. P., ter Hofstede, A.
nets. International Journal on Software Tools
H. M., & Edmond, D. (2005). Workflow resource
for Technology Transfer, 2, 98–132. doi:10.1007/
patterns: Identification, representation and tool
s100090050021
support. In O. Pastor & J. Falcão e Cunha (Eds.),
Medina-Mora, R., Winograd, T., Flores, R., & Lecture Notes in Computer Science, 3520 (pp.
Flores, F. (1992, November 1-4). The action work- 216-232). Berlin, Germany: Springer.
flow approach to workflow management technol-
van der Aalst, W. M. P., ter Hofstede, A. H. M.,
ogy. In J. Turner & R. Kraut (Eds.), Proceedings
Kiepuszewski, B., & Barros, A. P. (2003). Work-
of the 1992 ACM Conference on Computer Sup-
flow patterns. Distributed and Parallel Databases,
ported Cooperative Work (pp. 281-288). Toronto,
14, 5–51. doi:10.1023/A:1022883727209
Canada: ACM Publishing.
Weigand, H. (2006). Two decades of the
Murata, T. (1989). Petri nets: Properties, analysis
language-action perspective: Introduction.
and applications. Proceedings of the IEEE, 77(4),
514–580. doi:10.1109/5.24143
doi:10.1145/1125944.1125973
OASIS. (2007). Web services business process
Winograd, T. (1987). A language/action perspec-
execution language, V.2.0. Retrieved September
tive on the design of cooperative work. Human-
29, 2008, from http://docs.oasis-open.org/wsb-
Computer Interaction, 3, 3–30. doi:10.1207/
pel/2.0/OS/wsbpel-v2.0-OS.html
s15327051hci0301_2
1263
Winograd, T., & Flores, F. (1986). Understanding Workflow Management Coalition. (2007). XML
computers and cognition. Norwood, NJ: Ablex process definition language, V.2.1. Retrieved
Publishing Corporation. September 29, 2008, from http://www.wfmc.org
This work was previously published in Social, Managerial, and Organizational Dimensions of Enterprise Information Systems,
edited by Maria Manuela Cruz-Cunha, pp. 63-80, copyright 2010 by Information Science Reference (an imprint of IGI Global).
1264
1265
Chapter 5.6
A SOA-Based Approach to
Integrate Enterprise Systems
Anne Lämmer
Sandy Eggert
Norbert Gronau
This chapter presents a procedure for the integration Enterprise resource planning systems (ERP systems)
of enterprise systems. Therefore enterprise systems are enterprise information systems designed to
are being transferred into a service oriented architec- support business processes. They partially or com-
ture. The procedure model starts with decomposition pletely include functions such as order processing,
into Web services. This is followed by mapping purchasing, production scheduling, dispatching,
redundant functions and assigning of the original financial accounting and controlling (Monk et. al.,
source code to the Web services, which are orches- 2005). ERP systems are the backbone of information
trated in the final step. Finally, an example is given management in many industrial and commercial
how to integrate an Enterprise Resource Planning enterprises and focus on the management of master
System with an Enterprise Content Management and transaction data (Sumner 2005). Besides ERP
System using the proposed procedure model. systems, Enterprise Content Management Systems
(ECM systems) have also developed into compa-
DOI: 10.4018/978-1-60566-968-7.ch008 nywide application systems over the last few years.
A SOA-Based Approach to Integrate Enterprise Systems
ECM solutions focus on indexing all information and SOA are similar: different enterprise systems
within an enterprise (Rockley, 2003). They cover are connected through one interface, and a cross-
the processes of enterprise-wide content collec- system data transfer and the re-usage of objects
tion, creation, editing, managing, dispensing and or components is enabled. Thereby a service
use, in order to improve enterprise and cooperation represents a well defined function which is gener-
processes (CMSWatch, 2009). In order to man- ated in reaction to an electronic request (Burbeck,
age information independently, ECM combines 2000). The SOA approach offers a relatively easy
technologies such as document management, way to connect, add and exchange single services,
digital archiving, content management, workflow which highly simplifies the integration of similar
management, etc. The use of ECM systems is systems (e.g. enterprise take-over). Moreover,
constantly on the rise (Lämmer et. al 2008). This SOA offers a high degree of interoperability and
leads to an increasing motivation for enterprises modularity, which increases the adaptability of
to integrate the ECM systems within the exist- enterprise systems (Andresen et. al., 2008), (Läm-
ing ERP systems, especially when considering mer et.al. 2008).
growing international competition. The need for The SOA approach is based on the concept
integration is also eminently based on economical of service. The sender wants to use a service and
aspects, such as the expense factor in system run in doing so he wants to achieve a specific result.
time (Aier & Schönherr, 2006). For a cross-system Thereby the sender is not interested in how the
improvement of business processes, enterprise request is processed or which further requests
systems have to be integrated. are necessary. This is the idea of SOA, where
services are defined in a specific language and
referenced in a service index. Service request
rELAtED WOrK and data exchange occur via use of pre-defined
protocols (Erl, 2008), (Papazoglou, 2007).
service Oriented Architecture This service orientation can be used on different
as an Integration Approach levels of architecture. The grid architecture is a
common example of infrastructure level (Bermann
A number of integration approaches and concepts et. al., 2003), (Bry et. al., 2004). On the applica-
already exist. They can be differentiated by in- tion level an implementation usually takes place
tegration level (for example data, functions or in terms of web services.
process integration) and integration architecture The use of web services offers the possibility
(for example point-to-point, hub & spoke, SOA) of re-using raw source code, which is merely
(Aier & Schönherr, 2006). This paper presents an transferred to another environment (Sneed, 2000).
approach to integrating enterprise systems by way The benefit of this transfer is the re-usage of per-
of building up service oriented architectures. This fected (old) algorithms. The main disadvantage
integration approach is of special interest and will is the necessity of revising the raw source code
be described in more detail. in order to find possible dependencies (Sneed,
The concept of service orientation is currently 2000). This is also true for enterprise systems. It
being intensively discussed. It can be differenti- isn’t efficient to re-use the entire old system, but
ated from component orientation by its composi- rather only significant parts of it. To accomplish
tion and index service (repository). Additionally, this it is necessary to deconstruct the old enter-
SOA is suitable for a process oriented, distributed prise system and to locate the source code parts
integration (Aier & Schönherr, 2006). However, which can effectively be re-used. Our approach
the addressed goals of component orientation uses self-diagnosis for finding these source code
1266
locations. This analysis will be considered in the the system has to possess abilities of self-detection
third integration step. or self-diagnosis. Sensors to detect contextual and
the environmental changes as well as a monitor
self-Diagnosis for continuous analysis should be a central part of
a self-organizing information system and create
As just described, our approach uses self-diagnosis the basis for the following:
for location of useful source code. For this, the
method of self-diagnosis will be presented and • For self-configuration the possibilities and
the differences to other approaches will be shown. needs of configuration must be recognized.
Some approaches for transformation of legacy- • For self-healing the symptoms of an infec-
systems into a SOA exist already. However, these tion must be identified.
approaches see the whole system as one service. • For a self-optimization it is necessary to
The system gets a service description for using search for new calculations, algorithms
this service in a SOA. Our approach differs in that and solution space or process chains.
it deconstructs the system for a tailored need. For • The self-protection needs detectors for
this, the method of self-diagnosis is used. identification of attacks.
Self-diagnosis can be defined as a system’s
capacity to assign a specific diagnosis to a de- Figure 1 shows a systematization of selfware.
tected symptom. The detection of symptoms and Beyond this, self-diagnosis can be seen as
assignment are performed by the system itself an integral part of all self-organising systems.
without any outside influence (Latif-Shabgahi Because of the necessity to achieve all aims of
et. al., 1999). The mechanism of self-diagnosis self-organising systems, self-diagnosis has to be
has been detected surveying natural systems; it subordinated to self-management.
can partly be applied to artificial systems as well. Today some approaches for self-diagnosis in
Self-diagnosis can be seen as an integral part of computer systems already exist i.e. fault detec-
all self-organising systems. Self-organisation is a tion in networking communication, or in storage
generic term for self-x-capabilities also called self- and processor (Sun, 2004). Self-diagnosis also
ware. Self-organisation should lead information works in infrastructure and load sharing for per-
systems to become self-healing, self-protecting, formance (Mos, Murphy, 2001). In this article a
self-optimizing and self-configuring (Hinchey et. proposal for the implementation of self-diagnosis
al., 2006), (Garnek, 2007). For reaching these aims, of enterprise application systems and particularly
Figure 1. Systematization of selfware
1267
in the application layer of information systems is factor for a division between active and passive
presented without using case based reasoning or self-diagnosis. A fundamental basis for either
multi agent systems. alternative is an observer or monitor.
The first step of self-diagnosis is the detection Using passive self-diagnosis, the monitor
of symptoms. Usually the detection of one existing detects and collects symptoms and information.
symptom is not sufficient to make an indisputable It can either be activated automatically or manu-
diagnosis. In this case, more information and data ally (Satzger et. al., 2007). If you know which
have to be gathered. This can be described as items need to be observed and the point where
symptom collection. In a second step the symptoms this information can be gathered, you only have
are assigned to a specific diagnosis. Depending to monitor this point. This is what passive self-
on the diagnosis, corresponding measures can be diagnosis does. For example: if you want to know
taken (Horling et. al., 2001). how often a button is pressed, you have to find
Symptoms are a very abstract part of self- where the button-event is implemented in the code
diagnosis. These symptoms can be a high network and observe this button-event.
load in distributed systems, missing signals, or In active self-diagnosis, the program’s func-
buffer overload of the hardware layer. For en- tion or modules are the active elements. They
terprise systems the symptoms can be e.g. the send defined information to the monitor and act
frequency of usage of user interface elements by independently if necessary. The monitor is used
the user, dependencies of code parts or compo- as receiver and interprets the gathered informa-
nents. Other types of symptoms are possible. In tion and symptoms. The main advantage of ac-
general, the answer to questions concerning the tive self-diagnosis is the possibility of detecting
measure of interesting items provides hints for new symptoms, even if no clear diagnosis can
possible symptoms. be made before the problems become acute and
are forwarded to other systems. In contrast, us-
Differentiation of Collection ing passive self-diagnosis, the monitor can only
inquire about specific data. In this case, a response
Self-diagnosis can be categorized by symptom or further examination is only possible if the
acquisition method. Active and passive self- problem is already known. For example: if you
diagnosis must also be distinguished. In this don’t know the location of all the buttons and or
context, the program or source code is the crucial the code component for the button-event, you will
Figure 2. Kinds of acquisition of symptoms for self-diagnosis; a) passive; b) active
1268
have to recognise all events with their initial point system. The amount of markers is getting smaller
and filter them with the monitor. The monitor accordingly. This implementation is important for
doesn’t have to know how many buttons exist or the usage of dependencies (detection of causality)
where their code is located, but the buttons have about the different components of the system.
to “know” to register with the monitor. These are The third layer is the application layer. Thereby
the requirements of active self-diagnosis. one marker is used for one application. Usage for
The assembly of diagnosis points depends this implementation is mainly the surveying of
on application context and software system. The enterprise application landscapes.
required time and effort cannot be specified; it
depends on the design and implementation of the Differentiation of Patterns
software system.
Self-diagnosis can also be employed for the For self-diagnosis, data are transmitted to a moni-
examination of source-code usage and interdepen- tor either from the code or the application. De-
dences. Depending on the desired information, pending on the purpose, patterns can be identified
different points of diagnosis have to be integrated which refer to the frequency (frequency pattern)
into the source-code. or markers reporting the dependencies of code
Different points of diagnosis have to be deter- fragments (causality pattern).
mined in order to allow for the allocation of code The frequency pattern provides information
parts to various fields and functions. Therefore about the usage of parts of an application and al-
context, programming language, and software lows drawing conclusions on user as well as the
system architecture must be considered. temporal details. The monitor does not view the
content of the message. It counts the frequency
Differentiation of Layers of a message and saves the corresponding meta-
data, e.g. invoking user or instant of the request
Self-diagnosis can be employed for the examina- for information. The application context of the
tion of source code usage and interdependences. frequency pattern is particularly interesting. In
According to the desired information different order to supervise the use of code, frequency pat-
markers have to be integrated into the source-code. terns could be used. By using frequency pattern
Depending on the desired outcome, source it is easily possible to analyze the exact usage of
code usage and interdependences could be objects certain functions on the code layer. This analysis
of an analysis. could be used for maintenance. The maintenance
Three layers can be distinguished: (1) source overhead increases at the increasing complexity
code layer, (2) component layer and (3) applica- of operational standard software at any function
tion layer. which was taken over in the standard. In run time
For self-diagnosis of the source code layer the it is usually unknown which functions are really
markers have to be integrated into the functions used. By detecting the use of frequencies in the
or methods. This offers a very specific analysis runtime a slim down of the software system can
of the system’s bottom, but this is connected with be carried out to make sure that the maintenance
heavy efforts and a deep knowledge about the overhead is done only for the functions which
programming of the system. also are needed.
Self-diagnosis of the component layer inte- The causality pattern shows the dependencies
grates the markers into functional combined code of certain code fragments between each other.
fragments of the components of the information Again, the focus is on the metadata, like for
1269
example users and time. Unlike the frequency PrOcEDUrE MODEL

pattern, the causality pattern collects information
of the code fragment invoked before and the code In the following, a procedure model that integrates
fragment to be invoked. For this the monitor must general application systems within a company is
be able to determine and to save the predecessor presented. The procedure model begins with the
and the successor of every message. By analyz- deconstruction of systems into web services. This
ing this, dependencies about the code fragments is followed by a mapping of redundant functions
are possible. The causality pattern is particularly and the assignment of original source code to web
important for the examination of a system for services, which is orchestrated in the last step.
the connection to the business process or to the The process includes taking the old ERP sys-
encapsulation of code fragments. An application tem; deconstructing it into different abstraction
scenario for the causality pattern at the component levels such as functional entities or business ob-
layer is found in the area of the system integra- jects, searching for redundant entities, allocating
tion. Here the dependencies of the individual the code fragments dependent on these functional
components between each other are important entities, and encapsulating them. This results in
for the isolation or reprocess of components and many independent functional entities, which can
to wrap these into a web service (Sneed, 2000). be described as a service. They have different
The context pattern examines not only the abstraction levels and have to compose and or-
metadata but recognizes messages belonging chestrate with, e.g. BPEL-WS. This composition
together. This is a semantic evaluation of the and orchestration is the way of integration.
messages. In this case, the cost for developing
the monitor is much higher than within the other Deconstruction of systems
two detection models. This is due to the fact that
messages might have a different wording and be First, the systems which are to be integrated are
written in different languages. deconstructed into services. The challenge of this
Our approach uses this method to locate code step depends on the number of particular services,
parts that can be collected into components. As which could span the range from one single ser-
we will demonstrate later in this article, we need vice per system, up to a definition of every single
to locate functions and enterprise systems busi- method or function within a system as a service. In
ness objects. the case of a very broad definition, the advantages,
This method can be used for the detection of such as easy maintenance and reuse etc., will be
code parts which are possible services. Diagnosis lost. In case of a very narrow definition, disadvan-
points must thereby be integrated into the system tages concerning performance and orchestration
source code, and software dependencies analysed. develop; the configuration and interdependencies
As discussed earlier the main challenges in of the services become too complex.
integration of legacy enterprise systems like This paper proposes a hierarchical approach
ERP and ECM are, first, the deconstruction and which describes services of different granular
second, the allocation of code. To address these qualities on three hierarchical levels. Areas of func-
challenges, we have developed a procedure model tion of a system are described as the first of these
which will be described next. levels (figure 3, part 1). For example, an area of
functions could include purchase or sales in case
of ERP systems and, in the case of ECM systems,
archiving or content management. An area of
1270
Figure 3. Procedure model for the integration of application systems
1271
function can be determined on the abstract level them or the original source code. The data level
by posing questions about the general “assigned is not touched by this process.
task” of the system. The differences between the
three hierarchical levels can be discovered by Preparation and Mapping
answering the following questions:
The main advantage of web service architecture
1. Question: What are the tasks of the paris the high degree of possible reuse. By divi-
ticular system? The answers resulting from sion into three hierarchical levels, a detection of
this step correspond to services on the first similar functions is made possible, especially on
level, which constitute the general task. For the level of functionality and business objects.
example Sales, purchase, inventory manage- In some cases an adjustment of the functions is
ment or workflow management, archiving necessary in order to serve different contexts of
and content management. These tasks are use. Therefore, the next step consists of integration
abstract and describe main functionalities. on different levels and the mapping of identical
They consist of many other functions which functions (figure 3, part 2). This step poses the
are the objects of the next level. following questions:
2. Question: Which functionality derives from
every single task? The answers to this ques- 1. Question: Which tasks, functions and busi-
tion correspond to the services on the second ness objects appear more than once? For
level that are contributed by the various example: most applications contain search
functions. These functions are more detailed functions, some applications have functions
than the general tasks. They describe what for check in and check out, ERP systems
the tasks consist of and what they do, for calculate the time for many things with the
example: calculate the delivery time, identify same algorithm under different names.
a major customer, or constitute check-in and 2. Question: Are these multiple functions and
E-Mail functionalities. For these functions objects redundant, i.e. superfluous, or do they
the application needs data, which can found provide different services? Some functions
in the third level. may have the same name, but perform dif-
3. Question: Which business objects are ferent tasks.
utilised by both systems? This answer con- 3. Question: Can these multiple functions and
stitutes the number of business objects that objects be combined by way of appropriate
will be used as basic objects in both systems, programming? For the functions ascertained
e.g. article data, customer data or index data. in question 2 to be similar functions with
different names, the possibility of integrating
In this procedure model, all possible levels them into one has to be analysed.
of service deconstruction are addressed; yet the
realisation on all hierarchical levels constitutes The advantage of this mapping is the detection
an individual task. of identical functions, which may by only named
The result of this step is a 3-stage model differently while completing the same task. In
displaying the services of an application. The doing so, the benefit of reuse can be exploited to
data-level, i.e. the integration of databases, is not a high degree. Additionally, this part of the sur-
further examined at this point since it is not an vey allows for a minimisation of programming,
integral part of our model, the aim of which is to due to encapsulation of multiple functions. Only
wrap functions as web services without altering those functions which share a high number of
1272
similarities, but nevertheless complete different can be started. This step necessitates the usage
tasks, have to be programmed differently; they of the existing source code and the description
can be merged by reprogramming. of relevant parts with a web service description
It is important to note that this part the decon- language, making possible the reuse of source
struction consists of an abstract level and in the code in service oriented architecture.
functional view. In the following step, this will If redundant services have been detected in
change: from a functional view to a code view. part two which need to be reengineered, then the
reengineering happens now.
Detection and Assignment of
services to code Fragments Orchestration of Web services
The next step brings the biggest challenge, namely The results of stage three are the described web
the transformation of existing applications into services. These have to be connected with each
service oriented architecture. Until now, services other depending on the business process. This
have been identified by their tasks, but the cor- orchestration takes place in several steps (figure
relation to existing source code still needs to be 3, part 4).
done. This is going to be accomplished in the next First, the context must be defined; second, the
step (figure 3, part 3). service description language has to be selected;
Self-diagnosis is used at this point to integrate and third, the web services need to be combined.
earlier defined points of diagnosis into the source A four-stage procedure model for a service
code. These points of diagnosis actively collect oriented integration of application systems has just
usage data and facilitate conclusions concerning been described. This process holds the advantages
the fields and functions via their structure. The of a step-by-step transformation. The amount
structure of the points of diagnosis depends on the of time needed for this realisation is consider-
context of their application and on the software ably higher than in a “big bang” transformation,
system. It is not possible to describe the complexity however, a “big bang” transformation holds a
of the process, which also depends on the structure higher risk and therefore requires high-quality
and programming of the software systems. preparation measures. For this reason, a “big
As we discussed earlier in section 2.2, the bang” transformation is dismissed, in favour of
points of diagnosis depend on what needs to be a step-by-step transformation.
observed. Here we want to know which code There is yet another important advantage in
fragments share correspondences and execute the integration or deconstruction of application
the identified functions in the functional view. systems into services, when carried out in several
From this follows the necessity of a monitor. For steps. First, a basic structure is built (construction
example, the points can be every method call in of a repository, etc.). Next, a granular decom-
the source code of an ERP system. If the user calls position into web services occurs on the first
a function, the points of diagnosis have to inform level, thereby realising a basic transformation of
the monitor that they were called. The monitor has a service oriented concept. Following this, web
to recognise and to analyse which method calls services of the second and third hierarchical level
belong together. can be integrated step-by-step. This reduction into
Now the code fragments are analysed and as- services provides high quality integration.
signed to the functions identified in part 1, and The procedure model we just presented is
the wrapping of code fragments into web services very abstract. Therefore, a practical example for
1273
two enterprise systems, ERP and ECM, will be modules such as purchase, sales and article man-
given in part 4. agement. The ECM system consists of modules
such as document management, archiving and
workflow management. In the current situation
EXAMPLE OF APPLIcAtION a bi-directional interface between both systems
exists. One example for a business process in
It is necessary to develop a general usage approach which both systems are used is the processing of
and to test it on ERP and ECM systems, since no incoming mails and documents. In order to scan
concrete scenario of these technologies in regard and save the incoming invoices of suppliers, the
to praxis as of yet exists (Gronau, 2003). The aim module of the ECM System “document manage-
of this example of use is to describe the integration ment” is used. The access to the invoice is made
of both company-wide systems, ERP and ECM, possible through the ERP system.
using our presented approach. In the future, a SOA-based integration of both
In what follows, we present a case study of a enterprise systems can be reasonably expected
situation of integration of two systems: an ERP under the aspect of business process improvement.
and an ECM system. A German manufacturer of Referring to the example mentioned above, the
engines and devices administrate a complex IT “portal management” component could be used
landscape. This IT landscape includes, among to access, search, and check-in all incoming docu-
others, two big enterprise systems. One of them ments. What now follows is a description, in four
is the ERP system “Microsoft Dynamics NAV” parts, of the integration based on the procedure
and the other is the ECM system “OS.5|ECM” model we presented in part 3.
of Optimal Systems. The ERP System includes
Figure 4. Segmentation of services
1274
Part 1: segmentation of ErP and of analogy of different functions, particularly in

EcM systems into services enterprise systems where implementation is differ-
ent, lies in the expertise of the developer. Another
According to the procedure model (figure 3), the example is the service “check in/check out”. This
individual systems will be separated into indepen- service is a basic function of both ERP and ECM
dent software objects, which in each case complete systems and is now to be examined for possible
specified functions or constitute business objects. redundancy. After determining that the services
The segmentation is structured in three bottom-up “check in” or “check out” are equal, the service
steps (Figure 4). will be registered as a basic function only once.
Identification is based on the answers to Services which are not equal but related will be
questions concerning main tasks of specific checked in another step and either unified with
systems. The basic functions of an ERP system suitable programming or, if possible, spilt into
are purchase, sales, master data management, different services. The results of this step are the
article management und repository management. classification of services from ERP and ECM
Document management, content management, systems into similar areas and the separation of
records management, workflow management and redundant services. The following table shows
portal management are basic functions of ECM examples of separated services.
systems. Subsequently, the areas of functions By this separation of both enterprise systems,
are disaggregated into separate tasks. Business a higher degree of re-usage and improved com-
objects are classified such as the business object plexity-handling of these systems is achieved. For
“article” or “customer”. Thus, segmentation in the application of services, a service-oriented ar-
areas of functions, tasks of functions and business chitecture (SOA) which defines the different roles
objects is achieved and a basis for the re-usage of of participants is now required (Burbeck, 2000).
services is created.
Part 3: Detection and Assignment
Part 2: Preparation of of services to code Fragments
Integration/Mapping
As already described in the general introduction,
The results of the first step of the segmentation are the identification of functions to be segmented
separation of services of differentiated granularity
per system. According to the procedure model, the Table 1. Examples of separate services of ERP
mapping on the different areas will be arranged and ECM systems
in the second step. For that purpose, the potential
services described will be examined for similari- Separate services ERP ECM
ties. On every level of hierarchy, the functional Basic Functions Purchase content manage-
descriptions (answers to questions in part 1) of ser- ment
vices are checked and compared with each other. Sales archiving
If functions or tasks are similar, they will have to article manage- document manage-
ment ment
be checked for possibility of combination and to
repository man- workflow manage-
be slotted for later reprogramming. One example agement ment
of such similarity between functions is “create Areas of Functions check in email connection
index terms”. Most enterprise systems include identify delivery save document
the function “create index terms” for documents time
such as invoices or new articles. The estimation check out create index terms
1275
in the source code constitutes one of the biggest pose (Leymann & Roller 2000) new web services
challenges in a transfer to service-oriented archi- from existing web services with help of workflow
tecture. As part of this approach, the method of technologies (Leymann, 2003). In BPEL, a process
self diagnosis is suggested. Appropriate points is defined which is started by a workflow system
of diagnosis will be linked to the source code in in order to start a business process.
order to draw conclusions from used functions to Web services are addressed via a graphical
associated class, method or function in the original representation with a modelling imagery of WS-
source code. Through the use of aspect oriented BPEL. The business process is modelled inde-
programming, aspects can be programmed and pendently from the original enterprise systems.
linked to the classes and methods of the application Since in the first integration step, the systems
system. Necessary data, such as the name of the were separated by their tasks and functions, now
accessed method, can be collected by accessing all of the functions are available for the business
the respective classes and methods (Vanderperren process as well.
et. al., 2005).
Based on a defined service, “order transac-
tion”, all names of methods which are necessary cONcLUsION
for the execution of “order transaction”, must be
identified. To wrap the service “order transaction”, critical consideration
i.e. to combine it with a web service description
language, the original methods need be searched The procedure model for the integration of ap-
for and encapsulated. Additionally, the reprogram- plication systems as it has been presented in this
ming of redundant functions is part of the phase paper is an approach that has been successfully
of identification and isolation of services. This, deployed in one case. Currently the assignment
as well, is only possible if the original methods ability and the universality are being tested. The
are identified. self-diagnosis, i.e. the assignment of source code
to services via aspect oriented programming,
Part 4: Orchestration constitutes a bigger challenge.
of Web services A verification of costs and benefit cannot be
given sufficiently; however, several examples
The last integration phase is used to compile web show convincing results and suggest a general
services. The previous steps had to be completed transferability. The complexity in such a realisa-
in preparation for the procedure model. The web tion cannot be specified. Particularly for bigger
services now are completely described and have and complex systems, the cost to benefit ratio has
an URI to be accessed. Now, only the composition to be verified. Despite this, it must be recognised
and the chronology of requests of the specific web that the assignment of code fragments to functions
services are missing. For the orchestration the is not an easy task. If one observes every method
web service business process execution language call, a high number of calls must be analysed.
(WS-BPEL) is recommended. The WS-BPEL was Visualisation can be helpful for analysing, since
developed by the OASIS-Group and is currently method calls belonging together will build a clus-
in the process of standardisation (Cover, 2005). ter in the emerging network. The observation of
If the web services present a function with a busi- method calls is possibly not the most optimal way
ness process, the WS-BPEL is particularly suitable for very complex systems. If the functional view
for orchestration of web services (Lübke et. al., of services in part 1 is not part of the business
2006). Essentially, BPEL is a language to com- object layer, but only of the general task layer,
1276
one can reduce the numbers of diagnosis points. Andresen, K., Levina, O., & Gronau, N. (2008).
The possibilities depend on the programming Design of the evolutionary process model for
language and their constructs. adaptable software development processes. Paper
presented at the European and Mediterranean
resume Conference on Information Systems 2008 (EM-
CIS2008)
The approach presented above describes a pro-
Berman, F., Fox, G., & Hey, T. (2003). Grid
cedure model for service oriented integration of
computing. Making the Global infrastrucure a
different application systems. The integration pro-
Reality. Wiley.
ceeds using web services which thereby improve
the integration ability, interoperability, flexibility, Bry, F., Nagel, W., & Schroeder, M. (2004). Grid
and sustainability. The reusable web services computing. Informatik Spektrum, 27(6), 542–545.
facilitate the extraction of several functions and
Burbeck, S. (2000). The Tao of e-business services.
combination of these into a new service. This
IBM Corporation. Retrieved October 7, 2006,
allows for reuse of several software components.
from http://www.ibm.com/software/developer/
Altogether, Web services improve the adapt-
library/ws-tao/index.html
ability of software systems to the business pro-
cesses and increase efficiency (Gronau, 2003). To CMSWatch. (2009). The ECM Suites Report 2009,
give an example of the realisation of the procedure Version 3.1. CMS Works, CMS Watch Olney.
model, an integration of an ERP- and ECM-system
Cover, R. (2004). Web standards for business
was chosen. The reasons for this choice consist
process modeling, collaboration, and choreog-
in targeted improvement of business aspects and
raphy. Retrieved October 7, 2001, from http://
increasing complexity of both application systems.
xml.coverpages.org/bpm.html.
Dealing with this complexity makes integration
necessary. Through mapping, redundant functions Erl, T. (2008). Web service contract design and
can be detected and as a consequence, a reduction versioning for SOA. Prentice Hall International.
of the complexity is made possible. Regarding the
Ganek, A. (2007). Overview of autonomic comput-
adaptability and flexibility of affected application
ing: Origins, evolution, direction. In M. Parashar,
systems, web services are a suitable approach for
& S. Harir (Eds.), Autonomic computing (pp. 3-18).
integration. In particular, it is the reuse of services
New York: Taylor & Francis Group.
and an adaptable infrastructure which facilitate
the integration. Gronau, N. (2003). Web services as a part of
In addition to all of this, we expect to discover an adaptive information system framework for
additional advantages concerning maintenance concurrent engineering. In R. Jardim-Goncalves,
and administration of affected application systems. J. Cha, & A. Steiger-Garcao (Eds.), Concurrent
engineering: Enhanced interoperable systems.
Hinchey, M. G., & Sterritt, R. (2006). Self-
rEFErENcEs managing software. Computer, 40(2), 107–111.
doi:10.1109/MC.2006.69
Aier, S., & Schönherr, M. (2006). Evaluating
integration architectures – A scenario-based evalu- Horling, B., Benyo, B., & Lesser, V. (2001).
ation of integration technologies. In D. Draheim, Using self-diagnosis to adapt organizational
& G. Weber (Eds.), Trends in enterprise applica- structures (Tech. Rep. TR-99-64). University of
tion architecture, revised selected papers (LNCS Massachusetts.
3888, pp. 2-14).
1277
Kuropka, D., Bog, A., & Weske, M. (2006) Se- Mos, A., & Murphy, J. (2001). Performance moni-
mantic enterprise services platform: Motivation, toring Of Java component-oriented distributed
potential, functionality and application scenarios. applications. Paper presented at the IEEE 9th
In Proceedings of the tenth IEEE international International Conference on Software, Telecom-
EDOC Enterprise Computing Conference, Hong munications and Computer Networks - SoftCOM
Kong (pp. 253-261). 2001.
Lämmer, A., Eggert, S., & Gronau, N. (2008). A Papazoglou, M. P. (2007). Web services: Principles
procedure model for a SOA-Based integration and technology. Prentice Hall.
of enterprise systems. International Journal of
Rockley, A. (2003). Managing enterprise content.
Enterprise Information Systems, 4(2), 1–12.
Pearson Education.
Latif-Shabgahi, G., Bass, J. M., & Bennett, S.
Satzger, B., Pietzowski, A., Trumler, W., &
(1999). Integrating selected fault masking and
Ungerer, T. (2007). Variations and evaluations
self-diagnosis mechanisms. In Proceedings of
of an adaptive accrual failure detector to enale
the Seventh Euromicro Workshop on Parallel
self-healing properties in distributed systems.
and Distributed Processing (pp. 97-104). IEEE
In P. Lukowicz, L. Thiele, & G. Tröster (Eds.),
Computer Society.
Architecture of computing systems - ARCS 2007
Leymann, F., & Roller, D. (2000). Production (LNCS 4415, pp. 171-184).
workflow - Concepts and techniques. Prentice
Sneed, H. M. (2000). Encapsulation of legacy
Hall International.
software: A technique for reusing legacy software
Leymann, F., & Roller, D. (2006). Modeling components. Annals of Software Engineering, 9,
business processes with BPEL4WS. Informa- 293–313. doi:10.1023/A:1018989111417
tion Systems and E-Business Management, 4(3),
Sumner, M. (2005). Enterprise resource planning.
265–284. doi:10.1007/s10257-005-0025-2
NJ: Pearson Education.
Lübke, D., Lüecke, T., Schneider, K., & Gómez,
Sun (2004). Predictive self-healing in the Solaris
J. M. (2006). Using event-driven process chains
10 operation system. A technical introduction.
fo model-driven development of business appli-
Retrieved from http://www.sun.com/bigadmin/
cations. In F. Lehner, H. Nösekabel, & P. Klein-
content/selfheal/selfheal_overview.pdf
schmidt (2006), Multikonferenz Wirtschaftsinfor-
matik 2006 (pp. 265-279). GITO-Verlag. Vanderperren, W., Suvée, D., Verheecke, B.,
Monk, E., & Wagner, B. (2005), Concepts in Cibrán, M. A., & Jonckers, V. (2005). Adaptive
enterprise resource planning (2nd ed.). Boston: programming in JAsCo. In Proceedings of the
Thomson Course Technology. 4th international conference on Aspect-oriented
software development. ACM Press.
This work was previously published in Organizational Advancements through Enterprise Information Systems: Emerging Ap-
plications and Developments, edited by Angappa Gunasekaran and Timothy Shea, pp. 120-133, copyright 2010 by Business
1278
1279
Chapter 5.7
Achieving Business Benefits
from ERP Systems
Alok Mishra
Atilim University, Turkey
AbstrAct quisition and implementation of ERP systems in

organizations in the present global context.
Enterprise resource planning (ERP) systems are
becoming popular in medium and large-scale
organizations all over the world. As companies INtrODUctION
have to collaborate across borders, languages,
cultures, and integrate business processes, ERPs Enterprise resource planning (ERP) systems have
will need to take globalization into account, be become very important in modern business opera-
based on a global architecture, and support the tions as ERP has played a major role in changing
features required to bring all the worldwide players organizational computing for the better. One
and processes together. Due to the high cost of study found that more than 60% of Fortune 500
implementation for these systems, organizations companies had adopted an ERP system (Stewart,
all over the world are interested in evaluating their Milford, Jewels, Hunter, & Hunter 2000). These
benefits in the short and long terms. This chapter systems have been credited with reducing inven-
discusses various kinds of business benefits in tories, shortening cycle times, lowering costs, and
a comprehensive way in order to justify the ac- improving supply chain management practices.
ERP has been credited with increasing the speed
DOI: 10.4018/978-1-59904-531-3.ch005 with which information flows through a company
Achieving Business Benefits from ERP Systems
(Davenport, 1998). ERP has also been credited and respond to market changes, and implement
with creating value through integrating activities new business strategies to develop and maintain
across a firm, implementing best practices for each a competitive edge. ERP is designed to help
business process, standardizing processes within businesses succeed in the global marketplace by
organizations, creating one-source data that results supporting international legal and financial com-
in less confusion and error, and providing online pliance issues, and enabling organizations to adapt
access to information (O’Leary, 2000). All of these internal operations and business processes to meet
features facilitate better organizational planning, country-specific needs. As a result, organizations
communication, and collaboration (Olson, 2004). can focus on improving productivity and serving
Applied Robotics increased on-time deliveries their customers instead of struggling to ensure they
40% after implementing ERP, and Delta Electron- are in compliance with business and legal require-
ics reduced production control labor requirements ments around the world. Companies that automate
by 65% (Weil, 1999). Therefore, in the last decade, and streamline workflows across multiple sites
organizations around the world have made huge (including suppliers, partners, and manufacturing
investments in enterprise systems. According to sites) produced 66% more improvement in reduc-
a report by Advanced Manufacturing Research, ing total time from order to delivery, according
the ERP software market was expected to grow to Aberdeen’s 2007 study of the role of ERP in
from $21 billion in 2002 to $31 billion in 2006, globalization. Those companies that coordinate
and the entire enterprise applications market which and collaborate between multiple sites, operat-
includes customer relationships management and ing as a vertically integrated organization, have
supply chain management software will top $70 achieved more than a 10% gain in global market
billion (AMR Research, 2002). It was estimated share. The majority of companies studied (79%)
that in the 1990s about $300 billion was invested view global markets as a growth opportunity, but
in ERP systems worldwide and that this was ex- of those companies, half are also feeling pressures
pected to grow to $79 billion annually by 2004 to reduce costs (Jutras, 2007). Those companies
(Carlino, Nelson, & Smith, 2000). that coordinate and collaborate between multiple
Enterprise systems include enterprise resource sites, operating as vertically integrated organiza-
planning, customer relationship management tions, have achieved more than a 10% gain in
(CRM), supply chain management (SCM), product global market share (Marketwire, 2007).
lifecycle management (PLM), and e-procurement
software (Shang & Seddon, 2002). ERP software
integrates management information and processes, bUsINEss bENEFIts
such as financial, manufacturing, distribution, FrOM ErP systEMs
and human resources, for the purpose of enabling
enterprise-wide management of resources (Dav- With the growing proliferation of ERP systems,
enport, 1998; Deloitte Consulting, 1998; Klaus, including midsize companies, it becomes critical
Rosemann, & Gable, 2000). to address why and under what circumstances one
ERP helps organizations to meet the challenges can realize the benefits of an ERP system (Gefen
of globalization with a comprehensive, integrated & Ragowsky, 2005). ERP systems can provide the
application suite that includes next-generation organization with competitive advantage through
analytics, human capital management, finan- improved business performance (Hitt, Wu, &
cials, operations, and corporate services. With Zhou, 2002; Kalling, 2003) by, among other things,
support for industry-specific best practices, ERP integrating supply chain management, receiving,
helps organizations improve productivity, sense inventory management, customer orders manage-
1280
ment, production planning and managing, ship- 2003), on average there is a $10 gain in company
ping, accounting, human resource management, valuation for each dollar invested in IT (Brynjolfs-
and all other activities that take place in a modern son et al., 2002). Showing such a positive relation-
business (Gefen & Ridings, 2002; Hong & Kim, ship is important because it affects MIS funding.
2002; Kalling, 2003). Thus, business benefits from According to Parker and Benson (1988), in
ERP systems use are multidimensional, ranging order for enterprise to gain competitive advantage,
from operational improvements through decision- the way in which IT is justified financially must
making enhancement to support for strategic goals change. Classical quantitative techniques (e.g.,
(Davenport, 2000; Deloitte Consulting, 1998; cost-benefit analysis) are not adequate for the
Markus & Tanis, 2000; Ross & Vitale, 2000; evaluation of IT applications, except when dealing
Irani & Love, 2001; Wilderman, 1999; Cooke & with cost-avoidance issues, which generally occur
Peterson, 1998; Campbell, 1998). Gartner Group at the operational level. If these methodologies are
(1998) also mentions enterprise systems benefits in to be enhanced, additional measures—such as per-
these areas, including both tangible and intangible ceived value to the business, increased customer
benefits. While some companies claim to have satisfaction, and the utility of IT in supporting
reduced their cycle time, improved their financial decision making—must be considered (Katz,
management, and obtained information faster 1993). Investment in ERP systems payoff remains
through ERP systems, in general they still have a controversial question (Hitt et al., 2002; Sarkis &
a high initial implementation failure rate (Hong Sundarraj, 2003; Kalling, 2003). ERP systems are
& Kim, 2002; Songini, 2004). Many prior studies very complicated software packages that support
examining the relationship between investing in the entire set of organizational activities. Hence,
IT and the performance level of the organization it is possible that there are many unknown factors
(Weil, 1999) dealt with the ratio of total IT invest- that impact the relationship between investment in
ment (i.e., software, hardware, personnel) to the ERP and organizational productivity. This chapter
entire organization’s performance (the total profit observes the manager’s perception of the benefits
of the organization). their organization gains from using ERP systems
Many early studies found no positive rela- and what impacts this benefits.
tionship between the two variables. Strassmann ERP system investments are strategic in nature,
(1985) examined service-sector firms but found with the key goal often being to help a company
no significant relationship between investment in grow in sales, reduce production lead time, and
IT and high-performing firms. Berndt and Mor- improve customer service (Steadman, 1999). In
rison (1992) even found a negative relationship IT, evaluation costs are hard to quantify in post-
between the growth in productivity and investment implementation audits and benefits are harder to
in high-tech, although, as they point out, this may identify and quantify (Hochstrasser & Griffiths,
have been the result of measurement problems. As 1991; Willcocks & Lester, 1999; Irani, Sharif, &
Brynjolfsson (1993) summarizes, positive returns Love, 2001; Seddon, Graeser, & Willcocks, 2002).
from investing in IT may not have shown up in Management of organizations that adopt ERP
previous research because of the inadequate way expects many benefits from the systems. These
it was measured. When measuring IT investment expectations are often difficult to meet. ERP can
on a per-user basis, there is a positive correlation be seen to provide more responsive information
between IT investment and overall productivity to management. There also is more interaction
(Brynjolfsson, 2003). Although there is a large across the organization and more efficient finan-
variance among companies in the benefit they cial operation (Olson, 2004). There is weaker
achieve from their IT investment (Brynjolfsson, perceived benefit from operational performance,
1281
such as improved operating efficiency, inventory turing, and replicated in Sweden by Olhager and
management, and cash management. While more Selldin (2003). Both studies used a 1 to 5 scale,
information is available at higher quality, this with 1 representing “not at all” and 5 representing
does not directly translate to cost efficiencies “to a great extent.” Average ratings are given in
across the board. Tables 1 and 2.
SAP-ERP delivers business benefits where they Here, the results are very similar. ERP systems
matter most—to the bottom line—and addresses were credited with making information more
the internal and external business requirements available, at higher quality, and with integrating
of global enterprise. Organizations can invest in operations (Olson, 2004). There was neutral sup-
mySAP ERP with confidence that expansion or port for crediting ERP with providing benefits in
change in any country or division will be sup-
ported. The solution provides global businesses
Table 1. Expected benefits of ERP systems (Mabert
with concrete benefits that enable success, includ-
et al., 2000; Olhager & Selldin, 2003)
ing the following (My SAP ERP, 2007):
ERP Performance Outcomes United Sweden
• Improved productivity for greater efficien- States
cy and responsiveness Quicker information response time 3.51 3.81
• Increased insight for more assured deci- Increased interaction across the 3.49 3.55
sion making enterprise
• Advanced flexibility and adaptability to Improved order management/order 3.25 3.37

cycle
cut costs and speed change
Decreased financial close cycle 3.17 3.36
• A partner for long-term growth
Improved interaction with customers 2.92 2.87
Improved on-time delivery 2.83 2.82
In addition, businesses can reduce the costs
Improved interaction with suppliers 2.81 2.78
associated with compliance and administration,
Lowered inventory levels 2.70 2.60
in part by creating flexible processes that balance
Improved cash management 2.64 2.54
global demands with local needs and that can be
Reduced direct operating costs 2.32 2.74
adapted quickly as regulations change. Compre-
hensive financial and reporting features ensure
that globally consolidated financial reports can be
Table 2. Areas benefiting from ERP systems
generated quickly. Support for internal controls
improves financial management and reduces the Area United Sweden
risk of noncompliance. In conjunction with the States
Collaboration Folders (cFolders) application, Availability of information 3.77 3.74
employees can work in seamless virtual project Integration of business operations/ 3.61 3.42
process
teams with other departments, partners, and sup-
pliers around the world. Analytical capabilities Quality of information 3.37 3.31
help organizations improve strategic insight Inventory management 3.18 2.99
and performance through better identification Financial management 3.11 2.98
of global market opportunities and drivers (My Supplier management/procurement 2.99 2.94
SAP ERP, 2007). Customer responsiveness/flexibility 2.67 2.95

Benefits perceived from adopting an ERP Decreased information technology 2.06 2.05
costs
system were studied by Mabert, Soni, and Ven-
Personnel management 1.94 2.06
kataraman (2000) in Midwestern U.S. manufac-
1282
specific materials management and financial • Real-time functional system enhancement

functions. The ratings of support for customer capability
response and personnel management were quite
low (although the Swedish rating for customer In Table 3, the first three categories are based
response was very close to neutral). Interestingly, on Anthony’s (1965) much cited work on planning
both surveys found low support for crediting ERP and control systems. Many IS benefit analyses
systems with decreasing information technology and frameworks have been organized around
costs. With an ERP, an organization can better Anthony’s trinity of operational, managerial and
negotiate with suppliers and reduce the cost of strategic levels of management. One example is
raw materials by as much as 15% (Schlack, 1992). Shang and Seddon (2000):
Hence, the higher the cost of raw material, the
higher the value of raw material cost reduction • Weil (1990) evaluated the payoff from
out of the cost of the product. three types of IS investment—in transac-
tional, informational, and strategic sys-
tems—in the U.S. valve industry. He found
ErP bENEFIts FrAMEWOrK that the greatest benefits came from invest-
ment in transactional level IT.
Shang and Seddon (2000) provided a comprehen- • Gorry and Scott Morton (1971) and oth-
sive framework of the benefits of ERP systems. ers (Silver, 1990; Demmel & Askin, 1992)
In their survey of 233 vendor success stories and reported significant benefits from using IT
34 follow-up phone interviews from three major for managerial decision suport.
ERP vendor Web sites, they found that all orga- • Porter and Miller (1985) and others
nizations derived benefit from at least two of the (McFarlan, 1984; Rackoff, Wiseman,
five categories, and all the vendors’ products had & Ullrich, 1985; Clemons, 1991;
returned customer benefit in all five categories. Venkataraman, Henderson, & Oldach
In the beginning of 1997 during the re- 1993) noted significant benefits from the
engineering process, most of the multinational use of IT in pursuing strategic goals.
organizations perceived the following benefits • Mirani and Lederer (1998) adapted
of implementing ERP system: Anthony’s framework to build an instru-
ment for assessing the organizational ben-
• Common processes across the globe efits of IS projects.
• Centralized operations • Hicks (1997), Reynolds (1992), and
• Multi-language and currency capabilities Schultheis and Sumner (1989) also used
• Better tracking of inventory Anthony’s categories in classifying IT ben-
• Improved utilization of raw materials efits as operational, tactical, and strategic.
• Tighter integration of production with The categories were also used as frame-
sales and distribution works for analyzing the benefits of general
• Tax advantages through improved asset and enterprise-wide information systems
management (Wysocki & DeMichiell, 1997; Irani &
• Removal of a number of existing legacy Love, 2001).
systems • Willcocks (1994) and Graeser, Willcocks,
• Improved development and support and Pisanias (1998) adapted Kaplan and
environment Norton’s (1996) balanced scorecard ap-
proach in assessing IS investment in finan-
1283
Table 3. ERP benefits framework and extent of tangibility and quantifiability (adapted from Shang &
Seddon, 2000)
Dimension/Sub-Dimensions Tangible Quantifiable

1. Operational
1.1 Cost reduction Full Full
1.2 Cycle time deduction Most Full
1.3 Productivity improvement Most Full
1.4 Quality improvement Some Most
1.5 Customer services improvement Some Most
2. Managerial
2.1 Better resource management Some Most
2.2 Improved decision making and planning Some Some
2.3 Performance improvement Most Most
3. Strategic
3.1 Support business growth Some Full
3.2 Support business alliance Low Most
3.3 Build business innovations Some Some
3.4 Build cost leadership Some Some
3.5 Generate product differentiation Some Low
3.6 Build external linkages Low Some
4. IT infrastructure
4.1 Build business flexibility for current and future changes Low Low
4.2 IT costs reduction Full Full
4.3 Increased IT infrastructure capability Some Some
5. Organizational
5.1 Support organizational changes Low Low
5.2 Facilitate business learning Low Low
5.3 Empowerment Low Low
5.4 Build common visions Low Low
cial, project, process, customer, learning, INtANGIbLE bENEFIts IN

and technical aspects, and measured orga- It AND ErP PrOJEcts
nizational performance along Anthony’s
three levels of business practice. Webster (1994) defines a tangible item as “some-
thing that is capable of being appraised at an actual
Therefore, there are very strong precedents in or approximate value.” But the ‘value’ is monetary
the IS literature for attempting to classify the worth, or some other measure like customer satis-
benefits of enterprise systems in terms of organi- faction is not certain. According to Hares and Royle
zational performance along Anthony’s three (1994), “an intangible is anything that is difficult
levels of business practice.(Shang & Seddon, to measure,” and the boundary between tangible
2000). and intangible is fuzzy at best. Determining the
1284
intangible benefits derived from information & Simon, 2002). According to Hares and Royle
systems implementation has been an elusive goal (1994), there are four main intangible benefits in
of academics and practitioners alike (Davern & IT investment:
Kauffman, 2000). Remenyi and Sherwood-Smith
(1999) pointed out that there are seven key ways 1. Internal improvement: This includes pro-
in which information systems may deliver direct cesses, workflow, and information access.
benefits to organizations. They also indicated that 2. Customer service: This ensures quality,
information systems deliver intangible benefits delivery, and support.
that are not easily assessed. Nandish and Irani 3. Foresight: This is vision regarding markets,
(1999) discussed the difficulty of evaluating IT products, and acquisitions in the future.
projects in the dynamic environment, especially 4. Adaptability: This is the ability to adapt
when intangibles are involved in the evaluation. change in rapidly changing industry.
Tallon, Kraemer, and Gurbaxni (2000) cited a
number of studies indicating that economic and The third and fourth sets of intangibles are
financial measures fail to assess accurately the future oriented and include spotting market trends
payoff of IT projects and suggested that one means and the ability to adapt to change.
of determining value is through the perception of Hares and Royle (1994) stated that the first set
executives. They focused on the strategic fit and of ongoing intangible benefits are those concerned
the contributions of IT projects, but indicated that with internal improvement of company operations
researchers need somehow to capture or represent or performance. These include changes in produc-
better the intangible benefits of IT. In the technol- tion processes, methods of operations manage-
ogy arena, as in the business areas, many projects ment, and changes to production value and process
deliver benefits that cannot be easily quantified chains with resulting benefits in increased output
(Murphy & Simon, 2002). Many benefits related or lower production costs. The second group of
to the information technology projects cannot be ongoing benefits, customer-oriented intangibles, is
easily quantified, for example better information more difficult to measure because their effective-
access, improved workflow, interdepartmental ness is determined by external forces. The benefits
coordination, and increased customer satisfac- of improving customer service are greater retention
tion (Emigh, 1999). These are also the features of customers and customer satisfaction. The third
that are listed as key attributes of ERP systems group of intangibles embodies the spotting of new
(Mullin, 1999; Davenport, 2000). ERP systems market trends. If new trends can be anticipated,
are implemented to integrate transactions along then technology may be able to transform or create
and between business processes. Common busi- products, processes, or services to gain new sales
ness processes include order fulfillment, materials and market position. The final group of intangible
management, production planning and execution, benefits is the ability to adapt to change. As with
procurement, and human resources (Murphy & the identification of market trends, the benefits
Simon, 2002). ERP systems enable efficient and derived include adapting products and services to
error-free workflow management and account- market trends and the modification of production
ing processes including in-depth auditing. These processes—a critical ability for firms in rapidly
systems feature a single database to eliminate changing industries.
redundancy and multiple entry errors, and they ERP system investments are strategic in nature,
provide in-depth reporting functionality. ERP with the key goal often being to help a company
systems provide information for effective deci- grow in sales, reduce production lead time, and
sion making on all organizational levels (Murphy improve customer service (Steadman, 1999). Or-
1285
ganizations turned up an average value of -$1.5 all relevant organizational goals. The most ap-
million when quantifiable cost savings and revenue propriate management level is business managers
gains were calculated against system implementa- (middle-level management control), as they have a
tion and maintenance costs. Improved customer comprehensive understanding of both the capabili-
service and related intangible benefits such as ties of ES and the business plans for system use. It
updated and streamlined technical infrastructure is not expected that all organizations will achieve
are important intangible benefits that organizations benefits in all 25 sub-dimensions, or even in all
are often seeking when making these investments. five main dimensions, but it provides an excellent
The development and implementation of ERP checklist of benefits that have been accomplished
systems is longer in duration and cost intensive. It in organizations using enterprise systems.
is difficult to quantify in monetary terms because
of the intangible nature of many of the derived 1. Operational Benefits:
benefits, for example, improved customer service 1.1 Cost reduction:
(Murphy & Simon, 2002). ▪ Labor cost reduction in cus-
The literature suggests that intangibles can be tomer service, finance, human
converted into monetary terms through the ability resources, purchasing, IT ser-
to take care of the following observations: vices, and training.
▪ Inventory cost in inventory
1. Maintain and increase sales turns, dislocation costs, and
2. Increase prices warehousing costs.
3. Reduce costs ▪ Administrative expenses reduc-
4. Create new business tion in printing and other busi-
ness supplies.
Hares and Royle (1994) give a procedure to 1.2 Cycle time reduction:
quantify intangible benefits. The major steps are: ▪ Customer support activities in
order fulfillment, billing, deliv-
1. Identify benefits to be quantified ery, and customer enquiries.
2. Make intangible benefits measurable ▪ Employee support activities in
3. Predict the benefits in physical terms month-end closing, requisition,
4. Evaluate cash flow terms HR and payroll, and learning.
▪ Supplier support activities in or-
der processing, information ex-
ENtErPrIsE systEM change, and payment.
bENEFIt FrAMEWOrK 1.3 Productivity improvement:
▪ Production per employee, pro-
According to Shang and Seddon (2002), the follow- duction by labor hours, produc-
ing five-dimensional framework, which is built on tion by labor costs, increased
a large body of previous research into IT benefits, work volume with same work-
has been organized around operational efficiency force, and reduced overtime.
and managerial and strategic effectiveness, as the 1.4 Quality improvement:
outlook of strategic managers are too broad to ▪ Error rate, data reliability to data
identify casual links between enterprise system accuracy.
investment and benefit realization, and those of 1.5 Customer service improvement:
operational managers are too narrow to consider ▪ Ease of data access and inquiries.
1286
2. Managerial Benefits 2.3 Performance improvement in a

2.1 Better resource management: variety of ways in all levels of the
▪ Better asset management for organization:
improved cost, depreciation, ▪ Financial performance by lines
relocation, custody, physical of business, by product, by cus-
inventory, and maintenance re- tomers, by geographies, or by
cords control, both locally and different combinations.
worldwide. ▪ Manufacturing performance
▪ Better inventory management monitoring, prediction, and
in shifting products where they quick adjustments.
were needed and responding ▪ Overall operational efficiency
quickly to surges or dips in de- and effectiveness management.
mand. Managers are able to see 3. Strategic Benefits
the inventory of all locations in
their region or across boundar- Strategic benefits are in a wide spectrum of
ies, making possible a leaner activities in internal and external areas in terms
inventory. of general competitiveness, product strategies,
▪ Better production management strategic capabilities, and competitive position
for coordinating supply and de- of the organization.
mand, and meeting production
schedules at the lowest cost. 3.1 Support business growth:
▪ Better workforce management ◦ In transaction volume, processing ca-
for improved workforce allo- pacity and capability.
cation and better utilization of ◦ With new business units.
skills. ◦ In products or services, new divi-
2.2 Improved decision making and sions, or new functions in different
planning: regions.
▪ Improved strategic decisions for ◦ With increased employees, new poli-
greater market responsiveness, cies and procedures.
fast profit analysis, tighter cost ◦ In new markets.
control, and effective strategic ◦ With industry’s rapid changes in
planning. competition, regulation, and markets.
▪ Improved management deci- 3.2 Support business alliance by:
sions for flexible resource man- ◦ Efficiently and effectively consoli-
agement, efficient processes, dating newly acquired companies
and quick response to opera- into standard business practice.
tional changes. ◦ Building consistent IT architecture
▪ Improved customer decisions support in different business units.
with flexible customer servic- ◦ Changing selling models of new
es, rapid response to customer products developed by a merged
demands, and prompt service company.
adjustments. ◦ Transiting new business units to a
corporate system.
1287
◦ Integrating resources with acquired 3.8 Generating or sustaining competitiveness:

companies. ◦ Maintaining competitive efficiency.
3.3 Building business innovation by: ◦ Building competitive advantage with
◦ Enabling new market strategy. quick decision making.
◦ Building new process chains. ◦ Staying ahead of competitors for bet-
◦ Creating new products or services. ter internal business support.
3.4 Building cost leadership by: ◦ Using opportunities generated by
◦ Building a lean structure with stream- enterprise systems to pull abreast
lined processes. of world leaders by using the same
◦ Reaching business economies of software and being compatible with
scale in operation. customers.
◦ Shared services. 4. IT Infrastructure Benefits
3.5 Generating product differentiation by: 4.1 Building business flexibility by rapid
◦ Providing customized product or response to internal and external
services, such as early preparation changes at lower cost and providing a
for the new EMU currency policy, range of options in reacting to changing
customized billing, individualized requirements.
project services to different customer 4.2 IT cost reduction in:
requirements, and different levels of ▪ Total cost of maintaining and
service appropriate for various sizes integrating legacy systems by
of customer organizations. eliminating separate data cen-
◦ Providing lean production with make- ters and applications, as well as
to-order capabilities. their supporting costs.
3.6 Enabling worldwide expansion: ▪ IT staff reductions.
◦ Centralized world operation. ▪ Mainframe or hardware
◦ Global resource management. replacement.
◦ Multicurrency capability. ▪ System architecture design and
◦ Global market penetration. development.
◦ Cost-effective worldwide solution ▪ System upgrade maintenance.
deployment. ▪ System modification and future
3.7 Enabling e-commerce by attracting new changes.
customers or getting closer to customers through ▪ Technology research and
the Web integration capability. The Web-enabled development.
ES provides benefits in business to business and 4.3 Increase IT infrastructure capability:
business to individual in:
◦ Interactive customer service. Stable and flexible support for the current and
◦ Improved product design through future business changes in process and structure.
customer direct feedback. Stability:
◦ Expanding to new markets.
◦ Building virtual corporations with • Reliable platforms.
virtual supply and demand consortia. • Global platforms with global knowledge
◦ Delivering customized service. pipeline.
◦ Providing real-time and reliable data • Transformed IS management and increased
enquiries. IS resource capability.
1288
• Continuous improvement in process and 5.5 Shifting work focus:

technology. ◦ Concentrate on core work.
◦ Focus on customer and market.
Flexibility: ◦ Focus on business process.
◦ Focus on overall performance.
• Modern technology adaptability. 5.6 Increased employee morale and satisfaction:
• Extendibility to external parties. ◦ Satisfied users with better decision-
• Expandability to a range of applications. making tools.
• Customizable and configurable. ◦ Satisfied users with increased work
efficiency.
5. Organizational Benefits ◦ Satisfied users in solving problems
efficiently.
Organizational benefits can be evaluated in ◦ Satisfied users in increased system
individual attitudes, employee morale and moti- skills and business knowledge.
vation, and interpersonal interactions. ◦ Increased morale with better business
performance.
5.1 Changing work pattern with shifted focus: ◦ Satisfied employees for better em-
◦ Coordination between different inter- ployee service.
disciplinary matters.
◦ Harmonization of interdepartmental
processes. DIscUssION
5.2 Facilitating business learning and broaden
employee skills: The above benefits were reported by all selected
◦ Learned by entire workforce. cases as mentioned by Shang and Seddon (2002);
◦ Shortened learning time. also, examples of each benefit dimension were
◦ Broadened employee skills. found in cases from each ES vendor. Every busi-
◦ Employees with motivation to learn ness achieved benefits in at least two dimensions.
the process. Operational and infrastructure benefits were the
5.3 Empowerment: most quoted benefits: 170 cases (73% of 233)
◦ Accountability, more value-added claimed to have achieved operational benefits,
responsibility. and 194 cases (83%) claimed IT infrastructure
◦ More proactive users in problem benefits (Shang & Seddon, 2002). Operational
solving, transformed from doers to benefits such as cost, speed, and error rates are
planners. measurable in many cases. Managerial benefits,
◦ Working autonomously. although less tangible, are linked directly with
◦ Users with ownership of the system. information used at different decision-making
◦ Greater employee involvement in levels and with different resources. The most
business management. useful information on both these dimensions
5.4 Building common visions: was provided by business managers or process
◦ Acting as one and working as a com- owners, who had a clearer picture of the impact
mon unit. of the adoption of ES on the overall organization,
◦ Consistent vision across different lev- including their own and their colleagues’ decision
els of organizations. making. Strategic benefits appear to flow from
1289
a broad range of activities in internal and exter- cONcLUsION

nal areas, and are described in terms of general
competitiveness, product strategies, and other Assessing whether investment in enterprise sys-
strategic capabilities. Organizational benefits tems pays off is an important issue. Organizations
are mainly reflected in individual attitudes (e.g., can achieve a number of tangible and intangible
employee morale) and interpersonal interactions. benefits due to successful implementation of ERP
Operational benefits may come with increased systems. These benefits can be derived globally,
managerial effectiveness, strategic benefits rely and in the context of globalization it is important
on operational efficiency, and organizational ben- to understand an organization’s managerial people
efits can be realized in parallel with managerial and shareholders as well. ERP helps organiza-
benefits (Shang & Seddon, 2002). tions meet the challenges of globalization with a
Regardless of tangible or intangible benefits, comprehensive, integrated application suite that
it is progressively more difficult to measure includes next-generation analytics, human capital
managerial, organizational, and strategic benefits management, financials, operations, and corpo-
than infrastructure or operational benefits; this rate services. ERP is designed to help businesses
has been an issue of debate since information succeed in the global marketplace by supporting
systems advanced beyond transaction processing international legal and financial compliance is-
systems (Murphy & Simon, 2002). With ERP sues, and enabling organizations to adapt internal
systems, success has been determined based on operations and business processes to meet country-
the organization’s acceptance of the changes specific needs. This will be helpful for decision
that the system introduces. Further, Murphy and makers (managerial people) of organizations to
Simon (2002) observed that organizational and evaluate various available ERPs in acquisition
managerial classification benefits are not only the and implementation. This will also a further aid
most difficult to obtain, but also are the hardest managers in assessing the benefits of their exist-
to quantify. ing ERPs in the organization in a more objective
way all over the world .
FUtUrE rEsEArcH DIrEctIONs

rEFErENcEs
Empirical studies of ERP benefits assessments
in different organizations and their comparisons Anthony, R. N. (1965). Planning and control
might be an interesting area for further work in systems: A framework for analysis. Graduate
this direction. Furthermore, assessment of ERP School of Business Administration, Harvard
benefits can be performed at two levels: first at an University, USA.
enterprise level, where the entire ERP system can Berndt, E. R., & Morrison, C. J. (1992). High-tech
be assessed regarding different types of benefits capital formation and economic performance in
derived from the ERP; and second at a specific U.S. manufacturing: An exploratory analysis.
module (application) level, which offers interest- Economics, Finance and Accounting Working
ing areas for future research. Future research ef- Paper #3419, Sloan School of Management, Mas-
forts should focus on managerial, organizational, sachusetts Institute of Technology, USA.
and strategic benefits, which are still unexplored
in terms of intangible benefits measurement to Brynjolfsson, E. (1993). The productivity paradox
quantify. of information technology. Communications of the
ACM, 36(12), 67–77. doi:10.1145/163298.163309
1290
Campbell, S. (1998). Mining for profit in ERP Gefen, D., & Ragowsky, A. (2005). A multi-level
software. Computer Reseller News, (October), 19. approach to measuring the benefits of an ERP sys-
tem in manufacturing firms. Information Systems
Carlino, J., Nelson, S., & Smith, N. (2000). AMR
Management, (Winter): 18–25. doi:10.1201/107
Research predicts enterprise applications market
8/44912.22.1.20051201/85735.3
will reach $78 billion by 2004. Retrieved from
http://www.amrresearch.com/press/files/99518. Gefen, D., & Ridings, C. (2002). Implementa-
asp tion team responsiveness and user evaluation
of CRM: A quasi-experimental design study of
Clemons, E. K. (1991). Evaluation of stra-
social exchange theory. Journal of Management
tegic investments in information technol-
Information Systems, 19(1), 47–63.
ogy. Communications of the ACM, 34, 22–36.
doi:10.1145/99977.99985 Gorry, A., & Scott Morton, M. S. (1971). A frame-
work for management information systems. Sloan
Cooke, D. P., & Peterson, W. J. (1998, July). SAP
Management Review, 13, 49–61.
implementation: Strategies and results. R-1217-
98-RR, The Conference Board, New York, USA. Graeser, V., Willcocks, L., & Pisanias, N. (1996).
Developing the IT scorecard. London: Business
Davenport, T. H. (1998). Putting the enterprise into
Intelligence.
(July/August): 121–131. Hares, J., & Royle, D. (1994). Measuring the value
of information technology, 7, 109-122.
Davenport, T. H. (2000). Mission critical—Real-
izing the promise of enterprise systems. Boston: Hicks, J. O. (1997). Management information
Harvard Business School. systems: A user perspective. Minneapolis/St.
Paul: West.
Davern, M. J., & Kauffman, R. J. (2000). Discover-
ing potential and realizing value from information Hitt, L. M., Wu, D. J., & Zhou, X. (2002). ERP
technology investments. Journal of Information investment: Business impact and productivity
Management Information Systems, 16, 121–143. measures. Journal of Management Information
Systems, 19(1), 71–98. doi:10.1201/1078/43199
Deloitte Consulting. (1998). ERP’s second wave—
.19.1.20020101/31479.10
Maximizing the value of ERP-enabled processes.
New York. Hochstrasser, B., & Griffiths, C. (1991). Control-
ling IT investment: Strategy and management.
Demmel, J., & Askin, R. (1992). A multiple-
London: Chapman & Hall.
objective decision model for the evaluation of
advanced manufacturing systems technology. Hong, K., & Kim, Y. (2002). The critical success
Journal of Manufacturing Systems, 11, 179–194. factors for ERP implementation: An organizational
doi:10.1016/0278-6125(92)90004-Y fit perspective. Information & Management, 40(1),
25–40. doi:10.1016/S0378-7206(01)00134-3
Emigh, J. (1999). Net present value. Computer-
world, 33, 52–53. Irani, Z., & Love, P. E. D. (2001). The propaga-
tion of technology management taxonomies for
Gartner Group. (1998). 1998 ERP and FMIS
evaluating investments in information systems.
study—Executive summary. Stamford, CT.
17, 161–178.
1291
Irani, Z., Sharif, A. M., & Love, P. E. D. (2001). McFarlan, F. W. (1984). Information technology
Transforming failure into success through organi- changes the way you compete. Harvard Business
sational learning: An analysis of a manufacturing Review, (May/June): 98–103.
information system. European Journal of Infor-
Mirani, R., & Lederer, A. L. (1998). An instru-
mation Systems, 10, 55–66. doi:10.1057/palgrave.
ment for assessing the organizational benefits
ejis.3000384
of IS project. Decision Sciences, 29, 803–838.
Jutras, C. (2007). The role of ERP in globaliza- doi:10.1111/j.1540-5915.1998.tb00878.x
tion. Retrieved from http://www.aberdeen.com/
MIT. (n.d.). Economic performance in U.S. manu-
summary/report/benchmark/RA_ERPRoleinGlo-
facturing: An exploratory analysis. Economics,
balization_CJ_3906.asp
Finance and Accounting Working Paper #3419,
Kalling, T. (2003). ERP systems and the strategic Sloan School of Management, Massachusetts
management processes that lead to competitive Institute of Technology, USA.
advantage. Information Resources Management
Mullin, R. (1999). ERP users say payback is passé.
Journal, 16(4), 46–67.
Chemical Week, 161, 25–26.
Kaplan, R., & Norton, D. P. (1996). Using the
Murphy, K. E., & Simon, S. J. (2002). Intangible
balance scorecard as a strategic management
benefits valuation in ERP projects. Information
system. Harvard Business Review, (January/
Systems Journal, 12, 301–320. doi:10.1046/
February): 75–85.
j.1365-2575.2002.00131.x
Katz, A. I. (1993). Measuring technology’s busi-
MySAP ERP. (2007). Globalization with localiza-
ness value: Organizations seek to prove IT benefits.
tion. Retrieved from www.sap.com/usa/solutions/
Information Systems Management, 10, 33–39.
grc/pdf/BWP_mySAP_ERP_Global_Local.pdf
doi:10.1080/10580539308906910
Nandish, V. P., & Irani, Z. (1999). Evaluating
Klaus, H., Rosemann, M., & Gable, G. G. (2000).
information technology in dynamic environ-
What is ERP? Information Systems Frontiers, 2,
ments: A focus on tailorable information.
141–162. doi:10.1023/A:1026543906354
Logistics Information Management, 12, 32.
Mabert, V. M., Soni, A., & Venkataraman, N. doi:10.1108/09576059910256231
(2000). Enterprise resource planning survey of
O’Leary, D. E. (2000). Enterprise resource
U.S. manufacturing firms. Production and Inven-
planning systems: Systems, life cycle, electronic
tory Management Journal, 41(2), 52–58.
commerce, and risk. Cambridge: Cambridge
Marketwire. (2007). Thinking global? Don’t University Press.
lose sight of profitable growth. Retrieved from
Olhager, J., & Selldin, E. (2003). Enterprise
http://www.marketwire.com/mw/release_html_
resource planning survey of Swedish manufac-
b1?release_id=224493
turing firms. European Journal of Operational
Markus, L. M., & Tanis, C. (2000). The enterprise Research, 146, 365–373. doi:10.1016/S0377-
systems experience—From adoption to success. 2217(02)00555-6
In R.W. Zmud (Ed.), Framing the domains of IT
Olson, D. L. (2004). Managerial issues of enter-
research: Glimpsing the future through the past.
prise resource planning systems. McGraw-Hill
Cincinnati, OH: Pinnaflex Educational Resources.
International.
1292
Parker, M., & Benson, R. (1988). Information Seddon, P. B., Graeser, V., & Willcocks, L. (2002).
economics: Linking business performance to Measuring organizational IS effectiveness: An
information technology. London: Prentice Hall. overview and update of senior management
perspectives. The Data Base for Advances in
Porter, M. E., & Miller, V. E. (1985). How infor-
Information Systems, 33, 11–28.
mation gives you competitive advantage. Harvard
Business Review, 63, 149–160. Shang, S., & Seddon, P. B. (2002). Assessing and
managing the benefits of enterprise systems: The
Rackoff, N., Wiseman, C., & Ullrich, W. A. (1985).
business manager’s perspective. Information Sys-
Information systems for competitive advantage:
tems Journal, 12, 271–299. doi:10.1046/j.1365-
Implementation of a planning process. MIS Quar-
2575.2002.00132.x
terly, 9, 285–294. doi:10.2307/249229
Shang, S., & Seddon, S. (2000). A comprehensive
Remenyi, D., & Sherwood-Smith, M. (1999).
framework for classifying the benefits of ERP
Maximize information systems value by
systems. Proceedings of Americas Conference
continuous participative evaluation. Logis-
on Information Systems.
tics Information Management, 12, 14–25.
doi:10.1108/09576059910256222 Silver, M. (1990). Decision support systems:
Direct and non-directed changes. Information Sys-
Research, A. M. R. (2002). AMR Research pre-
tems Research, 1, 47–88. doi:10.1287/isre.1.1.47
dicts enterprise applications market will reach
$70 billion in 2006. Retrieved from http:www. Songini, M. C. (2004, August 23). Ford aban-
amrresearch.com dons Oracle procurement systems, switches
back to mainframe apps. Retrieved from http://
Reynolds, G. W. (1992). Information systems for
www.computerworld.com/softwaretopics/erp/
managers. Minneapolis/St. Paul, MN: West.
story/0.10801,95404,00.html
Ross, J. W., & Vitale, M. (2000). The ERP
Steadman, C. (1999). Calculating ROI. Comput-
revolution. Surviving versus thriving. Infor-
erworld, 33, 6.
mation Systems Frontiers, 2(2), 233–241.
doi:10.1023/A:1026500224101 Stewart, G., Milford, M., Jewels, T., Hunter, T.,
& Hunter, B. (2000). Organizational readiness
Sarkis, J., & Sundarraj, R. P. (2003). Managing
for ERP implementation. In Proceedings of the
large scale global enterprise resource planning sys-
Americas Conference on Information Systems
tems: A case study at Texas Instruments. Interna-
(pp. 966-971).
tional Journal of Information Management, 23(5),
431–442. doi:10.1016/S0268-4012(03)00070-7 Strassmann, P. (1985). Information payoff: The
transformation of work in the electronic age.
Schlack, M. (1992). IS has a new job in manufac-
London: Collier Macmillan.
turing. Datamation, (January 15), 38-40.
Tallon, P., Kraemer, K., & Gurbaxni, V. (2000).
Schultheis, R., & Sumner, M. (1989). Manage-
Executives’ perception of business value of infor-
ment information systems: The manager’s view.
mation technology: A process-oriented approach.
Boston: Irwin.
16(4), 145–173.
1293
Venkataraman, N., Henderson, J., & Oldach, S. ADDItIONAL rEADING

H. (1993). Continuous strategic alignment: Ex-
ploiting IT capabilities for competitive success. Bingi, P., Sharma, M. K., & Godla, J. K. (1999).
European Management Journal, 11, 139–149. Critical issues affecting an ERP implementation.
doi:10.1016/0263-2373(93)90037-I Information Systems Management, (Summer):
121–131.
Weil, M. (1999). Managing to win. Manufacturing
Systems, 17(November), 14. Litecky, C. R. (1981). Intangibles in cost/benefit
analysis. Journal of Systems Management, 32,
Wilderman, B. (1999). Enterprise resource man- 15–17.
agement solutions and their value. Stanford, CT:
MetaGroup. Motwani, J., Mirchandani, D., Madan, M., &
Gunasekaran, A. (2002). Successful implemen-
Willcocks, L. (Ed.). (1994). Information manage- tation of ERP projects: Evidence from two case
ment: Evaluation of information systems invest- studies. International Journal of Production
ments. London: Chapman & Hall. Economics, 75.
Willcocks, L. P., & Lester, S. (Eds.). (1999). Olson, D. L. (2004). Managerial issues of enter-
Beyond the IT productivity paradox. Chichester: prise resource planning systems (international
John Wiley & Sons. ed.). Singapore: McGraw-Hill.
Wysocki, R., & DeMichiell, R. L. (1997). Manag- Simms, J. (1997). Evaluating IT: Where cost-
ing information across the enterprise. New York: benefit can fail. Australian Accountant, (May),
John Wiley & Sons. 29-31.
van Everdingen, Y., van Hellegersberg, J., &
Waarts, E. (2000). ERP adoption by European
midsize companies. [t]. Communications of the
ACM, 43(4), 27–31. doi:10.1145/332051.332064
This work was previously published in Enterprise Resource Planning for Global Economies: Managerial Issues and
Challenges, edited by Carlos Ferran and Ricardo Salim, pp. 77-92, copyright 2008 by Information Science Reference
1294
1295
Chapter 5.8
Experiences of Cultures in
Global ERP Implementation
Esther Brainin
Ruppin Academic Center, Israel
AbstrAct limitations of technological implementation and

use policies to improve the benefits generated
The chapter considers the complexities of cultural by the technology. Topics of explicit concern
differences for global enterprise resource planning to ERP implementation in global organizational
(ERP) implementation. An extensive review of economies related to organizational and societal
the literature related to societal and organizational culture are discussed, and suggestions for manage-
culture is followed by a delineation of the stages rial mechanisms for overcoming major obstacles
of ERP implementation and the actors involved in this process are proposed.
in each stage, reflecting the basic assumption that
global ERP systems are not universally acceptable
or effective, and that testing the cross-cultural INtrODUctION
generalizability of ERP systems in organizations
will produce a managerial agenda that facilitates Enterprise resource planning (ERP) systems
the implementation process. The recognition impose high demands on virtually all organi-
and discussion of these differences can provide zational members since these process-oriented
a stimulus for identifying and modifying the technologies are designed to standardize business
procedures across the enterprise. In addition to
DOI: 10.4018/978-1-59904-531-3.ch010 its potential benefits, the introduction of an ERP
Experiences of Cultures in Global ERP Implementation
system into an enterprise becomes the means power of integrative examination is in produc-
for achieving organizational standardization and ing helpful insights to poor global technology
integration. Hence, it can be viewed as an orga- implementation antecedents. Technology does not
nizational “boundary crossing” channel, and as constitute an entity per se, and despite its standard
such, ERP integration processes are more than engineering components, its preferred pattern
likely to face resistance to change. It is not surpris- of use can differ as a result of what Orlikowski
ing, then, that while companies worldwide have (2000) described as the diverse attributions of
made substantial investments in the installation meanings and power relation conceptions within
of ERP systems, difficulties in implementation and between organizations. The social construc-
and uncertain bottom-line benefits (Davenport, tion of technological systems (Bijker, Hughes, &
1998; Kumar & Hillegersberg, 2000; Robey, Pinche, 1987) as an interpretive theory maintains
Ross, & Boudreau, 2000) may be attributed to a that different types of user groups differ in how
failure in the implementation process (Brainin, they conceptualize, interpret, and exploit tech-
Gilon, Meidan, & Mushkat, 2005; Klein, Conn, nologies and their potential. Accordingly, the
& Speer-Sorra, 2001) rather than a deficiency in effective utilization of a technology is the result
the technology itself. of explicit and implicit “negotiations” between
The complexity of ERP implementation is groups of users regarding the desired use of the
exacerbated when ERP systems are globally technology and its organizational contribution and
implemented since the integrative nature of this significance. “Technology should be treated as
technology calls for the crossing of national or simultaneously social and physical and examine
regional, in addition to organizational boundaries. the interplay between the material characteristics
Since countries, regions, and organizations differ of technology and the social context within which
in their absorptive capacity (Lane, Koka, & Pathak, it is designed and deployed” (Grant, Hall, Wailes,
2006) and their societal and organizational culture & Wright, 2006, p. 4).
(Hall & Hall, 1990; Hofstede, 1980; Javidan, In the literature on culture, technology is
House, & Dorfman, 2004a; Trompenaars, 1996), viewed as a “cultural artifact” (Schein, 1992),
implementation of ERP systems in global enter- thus constituting an integral and inseparable part
prises requires a detailed examination of potential of organizational culture, which is reproduced in
gaps or inconsistencies in the interaction between everyday working routines. Consequently, global
new technologies, end users, and organizations in ERP systems must not be viewed only as large off-
different countries and/or regions (e.g., Boerma & the-shelf software solutions that provide integrated
Kingma, 2005; Dube & Robey, 1999; Krumbholz business and software systems to a customer,
& Maiden, 2001; Leidner & Kayworth, 2006). but also as cultural tools, mostly designed and
The approach taken here blends interpretive invented by the Western world but implemented
and positivist theories (Lee, 1991). The use of in diverse local/regional settings, all having dif-
interpretive theories enables portrayal of tech- ferent cultural characteristics (Davison, 2002).
nological systems as “cultural tools,” ascribed In fact, cultural differences may impede ERP
with different interpretations. Utilizing positivist implementation even when its diffusion occurs
theories opens the way for mapping cultural differ- within Western countries. Boerma and Kingma
ences between countries, scoring countries in ac- (2005) described Nestlé’s ERP implementation as
cordance with their societal culture characteristics, an example of a misfit between the decentralized
and examining the relationship between societal organizational culture of Nestlé and the central-
culture and various outcomes such as leadership ized culture imposed on the conglomerate by the
style, and economic and social indicators. The adoption of the ERP system. Their example high-
1296
lights differences in cultural attributes between sure. Consequently, the findings of the GLOBE
a preexisting organizational culture and the new project, an impressive research effort conducted
technological culture after ERP implementation. by 170 investigators from 62 countries, will be
The Nestlé study is an example of how ERP presented. This project measured culture at dif-
software packages, implemented in different ferent levels through both practices and values,
organizational contexts, force local cultures to and explored the relationship between culture
surrender to ERP mechanisms and logic that dis- and societal, organizational, and leadership ef-
regard local cultures and leadership styles. Such fectiveness (House, Hanges, Javidan, Dorfman,
an implementation process may result in poor & Gupta, 2004).
and delayed exploitation due to a misfit between The basic assumption here is that global
the cultural characteristics that are embedded in ERP systems are not universally acceptable
and represented through the new technology, and or effective, and that testing the cross-cultural
the target organization’s culture and its end user generalizability of ERP systems in organizations
practices and perceptions. will produce a managerial agenda that facilitates
Accordingly, transfer of technology from one the implementation process. Topics of explicit
region/country to another necessitates the addition concern to ERP implementation in global orga-
of the cultural dimension to Bijker et al.’s (1987) nizational economies related to organizational
theory of the social construction of technology and societal culture, and proposal for managerial
among end users. Leidner and Kayworth (2006), mechanisms for overcoming major obstacles in
who reviewed 86 studies on the cultural dimen- this process will be addressed. The first section
sions of ERP implementation, suggested the term of this discussion provides a brief overview of the
“technology cultural conflict” to “…lend insights dimensions of organizational and societal culture.
into the understanding of the linkages between IT This is followed by a description of the stages
and culture” (p. 357). of ERP implementation and the actors involved
Cultural dimensions are entwined in technol- in the process at every stage. Building upon this
ogy implementation on different levels of analysis review, recommendations are proposed for the
and operation (e.g., societal, organizational). relationships between ERP implementation and
This chapter outlines the various stages of ERP cultural dimensions. The conclusion summarizes
implementation, indicating the potential for cul- the key points drawn from the analysis.
tural discordance or concordance at every stage.
Managers and end users are defined as actors
during this long process. They are likely to be “sOcIEtAL cULtUrE”
involved in ERP implementation from the initial AND Its IMPAct ON
stage of strategic planning and decision making, OrGANIZAtIONAL cULtUrE
and continue their involvement by leading and
championing the implementation using goal set- Cultural factors take center stage in the discussion
ting, feedback, and reward techniques. Clearly, of global ERP implementation. Implementation
global ERP implementation positions manage- of the same technology in different countries
rial and leadership practices in a multicultural produces various cultural encounters that may
context. Past research relating societal culture facilitate or inhibit its exploitation (Ford et al.,
to IT implementation was based on Hofstede’s 2003). Thus, the concept of organizational and
pioneering work (e.g., Ford, Connelly, & Meister, societal culture should be addressed. Work orga-
2003). However, it would be unwise to assume any nizations are distinguished by social experiences
aspect of reality is quantifiable by a single mea- that can be called “cultures.” Such experiences,
1297
however, do not necessarily represent the orga- that develops its own unique dimensions because
nization as a whole. In this sense, organizational of differences between tasks, expertise/occupa-
culture is itself organized into various work set- tions, and activities (Huang, Newell, Galliers, &
tings (Frost, Moore, Louis, Lundberg, & Martin, Pan, 2003; Rousseau & Cooke, 1988; Sackmann,
1985) and produces a form of locally recognized 1992; Trice, 1993). It is suggested that these dif-
and common social knowledge, similar to the com- ferent views be addressed as two perspectives that
mon knowledge that develops within a clan. The complement each other.
organizational culture is the glue holding a group
of people together. It is produced over a period Methodological Issues related to
of time, and helps solve the group’s problems of the Examination of Organizational
internal integration and survival in an external and societal culture
environment (Schein, 1992). At the same time, on
a different level of analysis, employees working Organizational culture has different layers of
in the same local societal culture share the same depth, ranging from the most visible layers of
societal meaning system (Hofstede, 2001), which artifacts, practices, and behavior, to the less visible
supports their adaptation to their local organiza- layers of values and basic assumptions (Schein,
tion. A shared meaning system can be formed at 1992). The multiplicity of organizational cultural
different levels, from the micro level of the group layers and the range of areas that they cover make
or team, the meso level of the organization, up to them difficult to study. Societal culture differences
the macro level of nations and beyond (Shokef in various organizations are visible only if the
& Erez, 2006). same research methodology is used; but a study
The question arises whether organizational that examined how societal culture is reflected in
culture is a reflection of societal culture, or orga- an organization’s culture encountered immense
nizations are culture-producers. The open system problems when researchers from many different
theory builds on the principle that organizations countries were required to cooperate and use the
are ‘open’ to their environment, and devote a great same metrics, as was the case at the GLOBE
deal of attention to understanding their immedi- research project (House et al., 2004).
ate tasks or business environment, through direct Other researchers have argued that the descrip-
interactions with their customers, competitors, tion of organizational culture as a set of shared
suppliers, labor unions, and government agencies. assumptions is rather oversimplified and mislead-
All these stress the importance of being able to ing. Myers and Tan (2002) asserted that even the
bridge and manage critical boundaries and areas concept of a “national culture” is problematic since
of interdependence, and develop appropriate the nation-state is a relatively recent phenomenon,
operational and strategic responses. Thus, the has continued to change in its form and makeup,
surrounding societal culture is an external source and is composed of more than one cultural group.
of influence on organizational culture through Martin (2002) proposed that the attribution of
the behavior of organizational members who meaning (which is a central part of the cultural
introduce their beliefs, norms, and values into process) is complicated because of cognitive and
the organization. Thus, consistent with the open normative diversity within an organization, and
systems theory, we expect to find systematic so- leads to integration as well as fragmentation, to
cietal variations over and above within-societal unity as well as diversity. It is the actor’s reality
differences. Yet individual organizations may also that forms the basis for further action—people
have their own unique culture: organizational produce and reproduce organizations by means of
culture can be examined as an independent entity actions and interactions on a daily basis (routine).
1298
Furthermore, culture research has been difficult understanding of how attributes of societal and
to conduct (Straub, Loch, Evariso, Karahanna, & organizational cultures affect selected organiza-
Srite, 2002) due to the lack of clear concepts or tional practices. This assumption is the key issue
measures of culture. Another issue is what particu- in global ERP implementation.
lar level of culture one should study (Pettigrew, Researchers who have explored the link be-
1990). Some researchers argue that culture cannot tween information technology and organizational
be objectively analyzed at a single level. Straub culture claim that a firm-level study of culture’s
et al. (2002), for example, suggested a more real- influence on the use of information systems
istic view of culture. In these researchers’ views, should not only examine organizational culture,
individuals are simultaneously influenced by an but also its possible interaction with national or
array of cultural values on the national, ethnic, organizational sub-culture values and how these
organizational, or even sub-cultural levels. interactions potentially influence behaviors (Dube
A final difficulty in studying organizational & Robey, 1999; Kaarst-Brown, 2004; Leidner &
culture relates to the content under investiga- Kayworth, 2006). At the same time, they assert
tion, which is exceptionally large, since culture that in studying cross-cultural differences, research
refers to the process of meaning construction and should address three types of methodological
sense-making. Thus, any attempt to measure and biases that have not been sufficiently taken into
describe culture in general, and organizational account (e.g., Hofstede, 1980): construct bias,
culture in particular, must focus on certain parts method bias, and item bias. However, the GLOBE
of the specific culture and disregard other ele- project’s very adequate dataset was able to repli-
ments. Nonetheless, the necessity to improve cate and extend Hofstede’s landmark study. It was
research methodologies regarding cross-cultural able to test hypotheses relevant to relationships
differences does not reduce the impact that among societal-level variables, organizational
culture has on differences in work behaviors in practices of three different industries’ sectors
general (Erez & Earley, 1993) and in information (financial, food, and telecommunications) in ev-
systems implementation in particular (Leidner & ery country, and leader attributes and behavior.
Kayworth, 2006). Furthermore, the data was sufficient to replicate
middle-management perceptions and unobtrusive
societal culture and Globalization measures.
The project consisted of three phases related
Many firms struggle with the interpretation, to three empirical studies: Phase One was devoted
implementation, and impact of globalization on to the development of scales that assess organiza-
their everyday operations. Adopting a global tional and societal culture, and culturally shared
mindset challenges managers and companies to implicit theories of leadership. Evidence for con-
look beyond their own operations so that they struct validity of the culture scales was provided
may improve their practice of global management from several sources such as Hofstede (1980)
beyond the experience provided by their local busi- and Schwartz (1992) (for a detailed description,
nesses. The importance of international thinking see Hanges & Dickson, 2004). Phase Two was
lies in its “ability to serve as a bridge between devoted to the assessment of nine core attributes
the home country (i.e., the head office’s country) of societal and organizational cultures. The nine
and the local sites’ environments, playing the role cultural dimensions that served as independent
of cultural interpreter for both sides” (Jeannet, variables in the GLOBE program are: uncertainty
2000, p. 37). The road map to strategic global avoidance (Hofstede, 1980), power distance (Hof-
competitiveness assumes that managers have an stede), institutional collectivism (Triandis, 1995),
1299
in-group collectivism (Triandis, 1995), gender knowledging and appreciating cultural values,
egalitarianism (based on Hofstede’s Masculinity practices, and subtleties in different parts of the
index), assertiveness (based on Hofstede’s Mas- world. McDonald’s is an illuminating example
culinity Index), future orientation (Kluckhohn of cultural sensitivity. In France, McDonald’s
& Strodtbeck, 1961), performance orientation serves wine and salads with its burgers. In India,
(McClelland, 1961), and human orientation where beef products are taboo, it created a mut-
(Kluckhohn & Strodtbeck, 1961; Putnam, 1993). ton burger named Maharaja Mac. To succeed in
When quantified, these attributes were referred global business, managers need the flexibility to
to as cultural dimensions, and 62 societal cultures respond positively and effectively to practices and
were ranked accordingly, testing hypotheses about values that may be dramatically different to what
the relationship between these cultural dimen- they are accustomed. However, it is not easy for
sions and several important dependent variables one to understand and accept practices and values
such as economic prosperity, success in basic that are different from one’s personal experiences.
science, societal health, life expectancy, and so The GLOBE research project has shown that the
forth. Phase Two also investigated the interactive status and influence of leaders vary considerably
effect of societal-cultural dimensions and industry as a result of the cultural force in the countries or
(finance, food processing, and telecommunica- regions in which the leaders function.
tion) on organizational practices and culturally
endorsed implicit theories of leadership (House et
al., 2004). By measuring these cultural dimensions stAGEs AND ActOrs IN
across 62 countries, the GLOBE project liber- tEcHNOLOGIcAL cHANGE
ated organizational behavior research from U.S.
hegemony in theory and practice. It is a valuable Technological change entails a long process in-
database that can serve as an important tool for volving different types of actors in each of several
designing global project interventions. stages. Beyond having significant implications for
In Phase Three, the impact and effectiveness of the organization’s form and function, the decision
specific leadership behaviors and styles of CEOs to purchase an ERP system for global implementa-
on subordinates’ attitudes and performance over tion necessitates a heavy investment of resources.
three to five years were investigated. This phase The uncertainty surrounding the decision and its
also included testing of the moderating effects of implications, as well as the implementation itself,
culture on relationships between organizational lengthens and complicates the process. In the
practices and organizational effectiveness. developing body of academic literature related
It seems clear that an ERP system that is to ERP project implementation, researchers un-
developed in one specific region or state and derline that it is important to divide ERP project
implemented in another region (the regions may implementation into several phases or levels,
differ only in the language spoken) constitutes suggesting ‘key activities’ to be included in every
an inter-cultural encounter that generally ends in phase (Al-Mudimigh, Zairi, & Al-Mashari, 2001;
conflict on one level or another. These encounters Markus & Tanis, 2000; Parr & Shanks, 2000) and
must be managed and implementation processes using critical success factors (CSFs) for planning
must be planned to cope with the diverse gaps and and monitoring ERP implementation. Dividing
expectations. The latter can create antagonism the process into stages and checking to see that
among end users at different levels, leading to the target is met at each stage may alleviate some
partial or deficient implementation. Therefore, degree of uncertainty and allow proper oversight,
one of the most important challenges is ac- including clear definition of starting and ending
1300
points for every stage. Parr and Shanks (2000) Initiative stage
related to the latter as “realistic milestones and
end date…” (p. 293). The need for a new global ERP system in an or-
Although the purchase of the system can be ganization can emerge as a result of institutional
viewed as the first stage in the implementation isomorphism, a technology crisis or inadequacy,
process, there is a consensus among researchers new technology boost, or global business de-
that organizational events and actions that precede velopment. In listing the reasons for adopting
the purchasing stage impact system selection enterprise systems, Markus and Tanis (2000)
and its eventual implementation. These stages differentiate between small companies’ and large
are referred to by Al-Mudimigh et al. (2001) companies’ reasons, and identify them as techni-
as “strategic levels,” including current system cal vs. business reasons (p. 180). In most cases,
evaluation, having a clear sense of vision and actors, within or from outside of the organization,
objectives, and implementation strategy. While identify a need or decide to promote technological
Parr and Shanks (2000) relate to the first stage as change by initiating the implementation process.
the “Planning” stage, including clarification of the As a result of such actions, a careful review of
system rationale and determination of high-level the technological change should be conducted
project scope, Markus and Tanis (2000) call this through strategic planning before any decision
stage the “Chartering Phase” and it includes the is made. ERP implementation calls for a major
use of a sound business case for a sound assess- organizational transformation, which must be
ment of business conditions and needs. The fol- planned strategically and implemented thoroughly.
lowing proposed implementation stages can serve The process of strategic planning starts with
as a common denominator to the above models. identifying the impetus for changes in the com-
They are proposed to address the implementation pany’s business and IT systems, and their expected
process as composed of five stages (most models strategic and operational benefits. This process
pertain to four stages only)—three of them occur helps people understand the need for change;
before the new technology is introduced into the it sparks their interest in it and promotes their
organization, and the final two stages occur fol- commitment to the process (Adams, Sarkis, &
lowing the introduction of the system. Although Liles, 1995; Al-Mashari, 2003). Strategic planning
several stages may overlap, it is correct to address associated with ERP implementation relates to
each as a separate stage and manage the progress process design, process performance measure-
from one stage to the next based on achieving the ment, and continuous process improvement,
targets in each stage. also known as business process reengineering
Many organizational actors in the organization (Hammer & Champy, 1993). Additionally, it deals
and its environment are involved in the implemen- with critical success factors in early phases of
tation process, yet their level of involvement may ERP implementation (Nah, Zuckweiler, & Lau,
vary from stage to stage. While the organization’s 2003). In classical strategic planning, employees
senior managers are the dominant and central or representatives from all organizational ranks
actors in the initial stages of the process, respon- are commonly co-opted into the process. The case
sibility passes to project supervisors appointed by of ERP implementation at Texas Instruments (TI)
management, and to end users, as the implementa- described by Sarkis and Sundarraj (2003) illus-
tion process proceeds. Stages that occur before trates how market forces compelled TI to make a
the system is delivered to the organization include radical shift in its business. TI’s strategic response
the following. to these changes identified flexibility and time
(speed) as the key strategic performance metrics
1301
that had to be stressed for them to be competitive at this stage, ensuring the involvement of users in
in their business. As a result, this is what they system development is an important mechanism
focused on when designing and implementing a for improving system quality and utilization
new ERP system for the company. Al-Mudimigh (Baroudi, Olson, & Ives, 1986). Sounder and
et al. (2001) suggested that a company’s decision Nashar (1995) referred to the matching process
whether to engage in business process reengineer- between new technology and end user needs as
ing before, during, or after ERP implementation the “transfer of technology between a developer
depends on the company’s specific situation. and a user” (p.225), and suggested that “failures
(of transfer) often occur as a consequence of
Decision stage many natural barriers within the transfer process
(e.g., a developer’s choice of an inappropriate
This stage leads to a decision to invest resources technology or a user’s risk aversion).” Leonard-
in acquiring and assimilating a new system. When Barton and Sinha (1993) stressed that effective
deciding to allocate resources for the purchase internal technology transfer—the implementation
of a global ERP system, a cost assessment that of technical systems developed and disseminated
includes cost estimations made by representatives to operational sub-units within a single organiza-
of all operational departments (e.g., logistics, tion—depends not only upon the cost, quality, and
training) must be performed. In addition to the compatibilities of the technology, but also upon
cost of the system, this must also include the two processes of interaction between developers
costs of implementation processes. It is standard and users: user involvement in development, and
practice today to assume that 18% of an ERP adaptation by the developers and users of both the
implementation budget should be reserved for technical system itself and the workplace. A link
implementation and consulting (Gartner Group, between producers and users is needed to ensure
2006). The assured availability of resources for a good fit of any global new ERP system to the
the entire implementation process is an important organizations that must implement it.
variable in predicting implementation quality Global ERP technology is based on “best
(Klein et al., 2001). practices” solutions, an ideal recipe for the most
Between the second and third stages, it is effective performance of business functions.
important to examine which system best answers Still, the organization must ask whether standard
the organizational needs defined in the strategic solutions fit its organizational requirements. Off-
planning process. Organizational suitability is the-shelf ERP software packages are implemented
a compulsory condition, but it is not enough to in different organizational contexts, which often
ensure successful implementation. Transforming deviate considerably from the context in which
an off-the-shelf product into one that is appropriate these packages were originally designed and
for a specific organization requires customization. developed. Boersma and Kingma (2005) noted
In this process it is essential that developers meet that the structure of ERPs typically requires the
users, therefore potential end users of the new redefinition of work from the actor’s point of view.
system should be involved in the selection process. At this point it is important to nominate the
project leader who will manage the fourth and
selection and Leader fifth stages of the implementation process and
Nomination stage achieve the organizational improvements that the
new system was designed to accomplish. Hammer
A development or selection process leads to an and Champy (1993) suggested that the implemen-
order to produce a new technological system. Thus, tation project leader establish a number of teams
1302
composed of potential end users, representing all of integration and the selective incorporation of
the operational units where the new system will be existing socio-technical practices with new ones”
implemented. The number of teams depends on the (p. 225). Thus it is important to acknowledge that
quantity and quality of operational processes that organizational actors can respond and influence
will be forced to undergo massive changes during ERP implementation.
system implementation. The teams should analyze
the future changes and find solutions to problems routine Operations stage
that are likely to occur during the implementation
stage. The leader operates as an advisor, assisted In this final stage of ERP implementation, the new
by a steering committee comprising the senior system becomes established and the employees
managers who are in charge of decision making who operate it adopt stable work patterns. At this
during the process. Two of their most important stage, the organizational improvements resulting
tasks are to define priorities among the various from system operation should be tested, and project
change processes and to arbitrate the conflict of leaders should conduct a post-project review of the
interest disputes that are almost certain to arise. entire process in order to achieve what Markus and
The final two stages occur after the new system Tanis (2000) called “technological and business
is delivered to the organization. However, several flexibility for future developments.”
activities must be performed between the third and Cultural dimensions impact the entire change
fourth stages, in order to ensure a smooth transfer process yet play varying roles in the different
of responsibility for the implementation and the stages of the ERP implementation process. Since
new system operation to the end users. implementation methods are culturally dependent,
they should be planned in advance after careful
basic Operational stage scrutiny of organizational and societal cultures.
In this stage, the new system is installed in the

workplace and employees “…ideally become tHE INtErFAcE bEtWEEN
increasingly skillful, consistent, and committed in cULtUrE AND GLObAL
their use of the innovation” (Klein & Speer-Sorra, ErP IMPLEMENtAtION
1996, p. 1057). Management attention should be
directed to setting priorities for ERP implementa- In global ERP implementation, the initiation,
tion and influencing the implementation process decision, and selection stages are usually domi-
and project duration (Sheu, Chae, & Yang, 2004). nated by the parent company and may reflect a
This is the stage for end user training, which—to- desire for more control and standardization of
gether with goal setting, feedback, and rewards—is work processes. Thus, the new system is imposed
a fundamental condition for system operation. The ‘top-down’ on most local managers and all end
focus in this stage is on system customization; users, turning this process into a deterministic
developers should interact with end users in fine- one, with no leeway for local impact related to
tuning the system. This is the stage that allows structural and/or cultural adjustment. Indeed,
explicit and implicit ‘negotiations’ between groups corporations’ different national and organizational
of users regarding the desired use of the technol- cultures have been shown to be associated with
ogy and its organizational contribution. Light and problems during ERP implementation (Krumbholz
Wagner (2006) found in their study that “the use & Maiden, 2001).
of customization to enable socio-technical integra- Using the contingency theory of organization,
tion allows for the recognition of existing forms Donaldson (1987) presented a model wherein
1303
incompatibility between technology and organi- that the analysis of the link between the imple-
zational structure results in lowered performance. mentation phases and cultural influences relates
Donaldson believed that in such cases the organiza- to specific cultural manifestations, since any at-
tion will eventually make an adjustment to restore tempt to measure and describe culture in general,
compatibility between the new technology and and organizational culture in particular, must be
the organizational structure (SARFIT—structural concerned with certain parts of the culture only.
adjustment to regain fit). Applying Donaldson’s In the context of information technology, an
argument to global implementation, cultural important point to be made is that information
adjustment must also be addressed to achieve ad- technologies are not culturally neutral and may
equate compatibility (Krumbholz & Maiden, 2001; come to symbolize a host of different values driven
Sheu et al., 2004). Implementation of global ERP by underlying assumptions and their meaning,
systems implies that management must have an use, and consequences (Coombs, Knights, &
international perspective, or what Jeannet (2000) Willmott, 1992; Robey & Markus, 1984; Scholz,
referred to as a “global mindset.” 1990). Using a value-based approach reveals the
In the various stages of global ERP implemen- types of ‘cultural conflicts’ that might arise from
tation, different encounters with cultural mani- the development, adoption, use, and management
festations are to be expected. Thus, a cross-level of IT (Leidner & Kayworth, 2006). Thus, some
analysis of cultural issues is indispensable. To un- implementation phases are influenced by the
derstand the interface between culture and global international, national, regional, and/or business-
ERP implementation, the difference between level culture, while other phases are influenced by
treating organizations as “culture producers” the culture an organization produces. It should be
(Rousseau & Cooke, 1988) or as the reflections of emphasized that implementation of a global ERP
the national-societal culture surrounding the orga- system constitutes a force for cultural change,
nization must be examined. These two approaches and new cultural dimensions may be expected to
can be seen as complementary. Although organi- emerge in the assimilating organization at the end
zational culture is an internal attribute of an orga- of the process. In this sense, organizational culture
nization, Erez and Earley (1993) pointed out that is conceptualized as both a dependent and inde-
studies that focus only on the internal dimensions pendent variable. The purpose of portraying the
of organizational culture, without considering the cultural interface in global ERP implementation is
broad cultural context in which the organization to disclose the discrepancies and inconsistencies
operates, are deficient. The question of the impact that must be addressed to restore fit or reduce cul-
of the broader culture when examining the culture ture conflict, with the end result being heightened
of a specific organization has been considered in global ERP implementation effectiveness.
numerous studies (e.g., Hofstede, 1980; Schwartz, Mapping the link between the societal and or-
1992; Trompenaar, 1996). These studies ranked ganizational culture and the global implementation
approximately 60 countries according to cultural of ERP systems requires us to relate separately to
characteristics such as individualism/collectivism, the three initial stages, as described before, which
power distance, uncertainty avoidance, universal- take place prior to the installation of the system,
ism/particularism, affective/neutral culture, and and to the last two stages, which occur after the
specific/diffuse culture. organization has received the system. When a
In line with this, the following analysis of the global organization decides to introduce an ERP
interface between culture and ERP implementation system into its various worldwide branches, senior
relates both to organizational and societal-national management is responsible for executing the first
culture manifestations. It is important to stress three phases, as outlined above. The decision to
1304
acquire a new system is one that is forced upon the sequence of countries in which the system is
the organization’s distant branches, which do implemented.
not have the authority to make decisions of this
type. In contrast, the fourth and fifth stages are The Initiating Stage
within the range of authority and responsibility of
the individual sites, and consequently, the latter At this stage it is important to discover the de-
must plan and implement these stages effectively. gree of receptivity of companies in a particular
Only the local office (in which the organizational culture (represented by a nation) to certain types
management resides) carries out all five of the of innovations, compared to companies in other
stages. Assuming that the implementation is to cultures. This knowledge reflects the overall
be executed in a number of countries, the cultural ‘technological leap’ that could be expected to
implications of the global change involved in occur as a result of global ERP implementation.
the first three phases need to be discussed by the This course of action is supported by empirical
central senior management. Stages four and five findings from several studies. Van Everdingen and
are assumed to be executed by the organization’s Waarts (2003) compared ERP implementation in
local subsidiaries. 10 European countries and found that “national
culture has a significant influence on the country
the Importance of societal adoption rate” (p. 217). The extent of cultural open-
culture in the First three stages ness (accommodation of another’s culture) has a
of ErP Implementation strong positive influence on the degree to which
the technology transfer is successful (Hussain,
The product or end result of the first three stages 1998). IT is less readily adopted in risk-averse
is: (1) the decision to invest resources in acquiring and high power distance cultures since technol-
a global information system, and (2) the selection ogy is perceived as inherently risky (Hasan &
of a suitable system. The central axis around Ditsa, 1999; Png, Tan, & Wee, 2001; Srite, 2000;
which these decisions are made is the process of Thatcher, Srite, Stephina, & Liu, 2003). Thus, the
strategic thinking. impetus for acquiring a global ERP system will
There are long-term implications of the strate- most likely come from countries that have the
gic planning by firms’ intent on becoming global cultural characteristics that allow them to cope
enterprises. Such strategic planning requires a with change and uncertainty. Findings from the
balanced choice between strategic decisions in- project GLOBE related to future orientation can
volving numerous variables such as values and be used to assess a country’s openness to change
problem-solving methods of owners, managers, and flexibility. Societies that score higher on
and end users; formal recruitment procedures; future orientation tend to achieve economic suc-
reward systems; and regulation and control process, have organizations with a longer strategic
cesses. As a company expands internationally, it orientation, have flexible and adaptive organiza-
needs to fit its corporate culture to the various tions and managers, and so forth. The two most
societal cultures of its overseas operations to future-oriented countries are Singapore and Swit-
obtain the maximum benefits of the new imple- zerland, and the lowest are Argentina and Russia
mented system. (Ashkanasy, Gupta, Mayfield, & Trevor-Roberts,
Given that we are discussing a complex pro- 2004, p. 304). Studies on India and Southeast and
cess of change, the cultural issues in this process East Asia showed a process of strategic planning
take on critical significance since they may affect quite distinct from the rational approach dominant
in the West. Haley and Tan (1999) observed that
1305
“strategic planning in South and Southeast Asia appropriate action to sell the system to end users
has developed into a process which is ad-hoc and convince them of its benefits.
and reactive, highly personalized, idiosyncratic
to the leader and which uses relatively limited The System Selection and
environmental scanning” (p. 96). Leader Nomination Stage
The Decision Stage In the third global ERP implementation stage, two
pivotal events must take place: system selection
The decision to invest resources in a new system and nomination of a leader to oversee the project.
must be made only after a strategic process is
undertaken by the parent company. This process system selection
will also help the parent company decide how
and in what order the system is to be distributed At this stage, managers are shown a demonstration
to different countries. Such decisions are cultur- of the system under consideration. Differences
ally endorsed. Moreover, target country selection in perceptions between developers of the system
has two important implications. The first relates and managers of the purchasing organization
to the examples given in our description of the constitute an important consideration in the stra-
first phase. The parent company must anticipate tegic process of choosing a system. Leidner and
differences among the implementing countries in Kayworth (2006) claimed that variations across
terms of their abilities to cope with change. The cultural values may lead to differing perceptions
higher the implementing country is in uncertainty and approaches in the manner in which information
avoidance and power distance, the more resources systems are developed. Thus managers should be
should be allocated to them for the implementa- aware that the system reflects the developers’ view
tion process. Support by the parent company may of the cultural organization, which might deviate
greatly alleviate problems that such countries considerably from that of their own organization
encounter in the implementation process. or its strategic needs. These differences were
Second, different countries may have differ- conceptualized by Hazzan and Dubinsky (2005)
ent perspectives on the strategic goals of the new in their discussion of the connections between
system based on differences in societal values. For a national culture and the culture inspired by
instance: Service quality dimensions function dif- software development methods. According to
ferently across national cultures (Kettingen, Lee, their proposed model of the tightness of Software
& Lee, 1995). Information privacy concerns were Development Methods (SDM) and the tightness
found to vary across countries: countries high in of a national culture, the fitness of a given SDM
uncertainty avoidance and power distance ex- and a national culture can predict the degree to
hibited higher levels of government involvement which a given SDM will be accepted by a specific
in privacy regulation (Milberg, Burk, Smith, & national culture. Thus, the system developers and
Kallman, 1995). the organization’s IT people must be involved in
Since agreement on strategic goals is an impor- the ERP selection process.
tant condition for implementing and exploiting the Since IT developers are in charge of technical
new IT system, early mapping of the differences in assistance during the implementation processes,
the perception of these goals is imperative. Gaps it is important to understand that this assistance is
will appear as a result of differences in national also culturally dependent: Hassan and Ditsa (1999)
cultures, but knowing about them beforehand found that IT staff are able to give advice to IT
will allow the system’s implementers to take managers in countries with low power distance.
1306
This finding has implications for the fourth stage of the organization worldwide. The products of
of implementation as well, in which end users these stages should include: the exploitation of
operating the system require technical assistance the new system, final customization of the system
in coping with problems that emerge. However, to the needs of end users and to the special needs
when cultural differences between the developer’s of specific organizations, a post-project review
support personnel and end users are significant, process, and a cultural change in the organization
they may impede the implementation process. (if required).
The central axis around which these stages
Project Leaders for Global revolve relates to employee management, and it
ErP Implementation involves classic areas in organizational behavior
and human resource management such as training,
Global ERP implementation implies confront- learning, and performance goal setting and reward.
ing work situations charged with dynamic In these stages of the process, the relevant
cultural issues. Elaborating effective solutions cultural dimensions concern the differences within
for global implementation activities, influenced the organizational culture. An organizational
by nuances of culture, pose difficult challenges culture is not uniform; it comprises horizontal as
that may confound even the most skilled leader. well as vertical sub-cultures. A horizontal sub-
Brake, Walker, and Walker (1995) portrayed the culture may be created on the basis of professional
ideal global implementation leader as a strategic practices—a sub-culture of physicians, nurses,
architect-coordinator who is able to recognize op- and managers (Trice, 1993). Similarly, a vertical
portunities and risks across national and functional sub-culture may emerge, based on the organiza-
boundaries, and is sensitive and responsive to tional structure—a sub-culture of the production
cultural differences. The implementation project department, the R&D department, or the HR
leader should have a “global mindset” since global department (Schein, 1996), or a sub-culture of
ERP system implementation requires fundamental technology-intensive departments as opposed to
changes in managerial practices in domestic as low-technology departments (Brainin et al., 2005).
well as international organizations. The GLOBE These sub-cultures imply that there will be differ-
project offers detailed examples of the cultural ences in the implementation of the new IT system
manifestation of countries and leadership attribu- because each sub-culture perceives the role and
tion. The GLOBE’s major empirical contribution function of the IT system in its work differently.
to this stage in the implementation process was its This process of ascribing meaning to technology
identification of universally desirable and cultur- is labeled the “social construction of technology.”
ally contingent attributes of leadership. GLOBE’s Furthermore, different occupational sub-cultures
findings showed the relationships among cultural have entirely different cultural interpretations of
dimensions, organizational practices, and cultur- proposed technologies, and experience conflict
ally endorsed leadership dimensions. and resistance to adopting certain technologies
(Von Meier, 1999). Researchers found that clash-
the Importance of Organizational ing values among organizational sub-cultures
culture in the Fourth and Fifth hinder the information sharing and collaboration
stages of ErP Implementation needed to effectively integrate technology.
In this stage it is important to appoint champi-
Transfer of technology from the manufacturer to ons of technological change from within the orga-
the end-user begins when the decision is made nization (Howell & Higgins, 1990) to effectively
to implement the system in different branches lead the implementation process. Management
1307
style also influences the implementation approach will likely amplify. As a means to integrate work
and project duration. Management style relates to processes, ERP implementation becomes a method
the attitude toward setting priorities for implement- for organizational boundary crossing and requires
ing an ERP system (Sheu et al., 2004). It must be special handling to overcome the resistance this
stressed that societal culture also has an influence change induces. The above discussion provides a
on the organization’s activities, and therefore the detailed examination of the cultural discrepancies
techniques used in human resource management, that arise at various stages of the implementation
as explained above, must be adapted to the local process. Cultural boundaries to be crossed as a
culture, for example, societal culture was found result of global ERP implementation within the
to influence instruction processes (Earley, 1994), organization include: organizational sub-cultures
goal setting (Erez, Earley, & Hullin, 1985), and as a result of organizational structure, and orga-
many other areas. nizational sub-cultures as a result of different
There are organizational culture characteristics professions and roles. Further, between organiza-
that facilitate technology implementation. Organi- tions this includes societal culture as a result of
zations with a high learning orientation are better cultural differences between countries.
able to adjust to changes on the whole (Lipshitz, To overcome the problems arising from global
Popper, & Oz, 1996), and to technological changes ERP implementation, it was suggested that cross-
in particular (Brainin & Erez, 2002; DeLong & ing the cultural boundary be treated as a process of
Fahey, 2003), because they have learning chan- cultural exchange, instead of as a cultural conflict.
nels and can learn from experience. This is a In order to restore fit and to prevent poor global
very important condition for implementing new ERP implementation and resistance to change,
technology. Global ERP implementation can result managers must harness the ERP systems to their
in culture transformation over time. needs by adapting it to their set of beliefs, thereby
breaking the link between technology and Western
logic. It requires them: to simultaneously examine
cONcLUsION their positions and behaviors and those of end users
regarding ERP use, and analyze their reciprocal
A Managerial Agenda for a Positive impact; to help formulate policy regarding utiliza-
cultural Experience during tion of ERP systems in different organizational
Global ErP Implementation settings across countries or regions; to coordinate
expectations between themselves and end users
The preceding discussion set forth to explain how in order to narrow the cultural gap; and to raise
variations in cultural aspects, which surface as the consciousness of ERP engineers and designers
a result of global ERP implementation, may be regarding the impact of differences in the socio-
handled, based on the assumption that global ERP cultural attributes of potential managers and end
systems are not universally acceptable and effec- users, and hence increase their effective fit into
tive. The increasing interrelations among countries different societies.
and the globalization of corporations do not imply The following recommendations are proposed:
that cultural differences are disappearing or dimin- Organizational leaders must be made aware of
ishing. On the contrary, as economic boundaries cultural differences and be prepared to cope with
are eliminated, cultural barriers may present new them. Financial resources must be available to
challenges and opportunities in business. When allow investment in a complete mapping of the
different cultures come into contact, they may cultural gaps that may manifest themselves in
converge in some aspects, but their idiosyncrasies the second implementation stage. Multicultural
1308
groups of project managers—glocal teams—and of cultural dimensions for each society, it does
team leaders (see Appendix) can overcome the not present a behavioral profile. Future research
lack of communication between the host and related to differences in societal culture is required
subsidiaries that often results in mistrust, proj- to build up in-depth understanding of how people
ect delay, and budget overruns. Moreover, such actually function and manifest different cultural
groups will foster informal communication among attributes, to investigate how different cultural
representatives of different nationalities, which dimensions interact, and what the relative impor-
is critical for the success of global ERP projects, tance of each dimension in understanding each
and will limit formal documentation channels that culture is (Javidan et al., 2004b). The suggested
people tend to use in the absence of such groups. managerial mechanisms proposed for formulating
Solutions may include: policy regarding utilization of ERP systems in
different organizational settings across countries
1. Multi-national teams for developing and or regions was the establishment of multicultural
implementing as a key to success groups of project managers—glocal teams—and
2. Local examination of the technological leap appointing team leaders. However, multicultural
3. Flexibility in implementation stages—even teams can face high levels of conflict and mis-
if the strategic stages were carried out in understandings. Such teams need to develop a
a deterministic fashion (were imposed on shared meaning system through socialization to
the organization), the later stages may al- the team, and through contacts and interactions
low participative actions and be culturally among team members. Their ability to serve as
appropriate. ‘mediators’ and represent their countries is con-
4. Involving users in the design of global ERP tingent upon collaboration dynamics that must
systems—although determining the key ac- overcome cultural differences. Research related
tors in user groups is especially challenging to multicultural teams is scarce (Earley & Gib-
in international settings, their involvement son, 2002; Earley & Mosakowski, 2000; Erez &
may partially assuage subsequent perception Gati, 2004). Shokef and Erez (2006) suggested
conflicts since the greater the extent to which that a ‘glocal’ identity represents both the global
a user’s group values are embedded in a identity and a strong local identity, and it seems to
system, the less vision conflict is expected. enable individuals to shift from one social context
to another. Future research is needed to explore
The chapter addressed aspects of organiza- the coexistence of multiple identities and its con-
tional and societal culture that are of explicit tribution to glocal team work. Another research
concern to ERP implementation in global organi- challenge relates to the issue of team leaders for
zational economies, and recommended managerial multicultural teams—what are adequate selection
mechanisms for overcoming major obstacles in and training processes? Current theories in man-
this process. agement and psychology do not provide sufficient
frameworks to explain the successes or failures of
people working and managing in foreign cultures.
FUtUrE rEsEArcH DIrEctIONs It is suggested that the measure of one’s ‘cultural
intelligence’ (CQ) be used as a predictor of an
Cultural differences may impede global ERP outsider’s natural ability to interpret and respond
implementation. Although the GLOBE project, to unfamiliar signals in an appropriate manner
used in this chapter to recognize and discuss dif- (Earley, Ang, & Joo-Seng, 2006). Accordingly,
ferences in societal cultures, provides a profile a manager with high CQ can enter new cultural
1309
settings—national, professional, organizational, Boerma, K., & Kingma, S. (2005). Develop-

regional—and immediately understand what is ing a cultural perspective on ERP. Business
happening and why, confidently interacting with Process Management Journal, 11(2), 123–136.
people and engaging in the right actions. How- doi:10.1108/14637150510591138
ever, there is very little empirical research that
Brainin, E., & Erez, M. (2002, April). Technology
explores this issue. The future research options
and culture: Organizational learning orientation
presented in this chapter should serve to stimulate
in the assimilation of new technology in orga-
and challenge researchers to explore new research
nizations. In Proceedings of the 3rd European
areas that will contribute to work experiences in
Conference on Organizational Learning (OKLC),
a global economy.
Athens, Greece.
Brainin, E., Gilon, G., Meidan, N., & Mushkat, Y.
rEFErENcEs (2005). The impact of intranet integrated patient
medical file (IIPMF) assimilation on the quality of
Adams, S., Sarkis, J., & Liles, D. H. (1995). The medical care and organizational advancements.
development of strategic performance metrics. Report # 2001/49 submitted to the Israel National
Engineering Management Journal, 1, 24–32. Institute for Health Policy and Health Services.
Al-Mashari, M. (2003). A process change-ori

Enterprise Information Systems Concepts, Methodologies, Tools and Applications

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Enterprise Information Systems Concepts, Methodologies, Tools and Applications

Uploaded by

Copyright:

Available Formats

Enterprise Information

Information Resources Management Association

Business science reference

Published in the United States of America by

and in the United Kingdom by

British Cataloguing in Publication Data

Editorial Advisory Board

Craig van Slyke

Electronic Business: Concepts, Methodologies, Tools, and Applications

Electronic Commerce: Concepts, Methodologies, Tools, and Applications

Electronic Government: Concepts, Methodologies, Tools, and Applications

Knowledge Management: Concepts, Methodologies, Tools, and Applications

Information Communication Technologies: Concepts, Methodologies, Tools, and Applications

Intelligent Information Technologies: Concepts, Methodologies, Tools, and Applications

Information Security and Ethics: Concepts, Methodologies, Tools, and Applications

Medical Informatics: Concepts, Methodologies, Tools, and Applications

Mobile Computing: Concepts, Methodologies, Tools, and Applications

Multimedia Technologies: Concepts, Methodologies, Tools, and Applications

Virtual Technologies: Concepts, Methodologies, Tools, and Applications

Free institution-wide online access with the purchase of a print collection!

information science reference

Chapter 1.2. Evolution of Enterprise Resource Planning ..................................................................... 21

Chapter 1.3. Exploring Enterprise Information Systems ...................................................................... 35

Chapter 1.4. Enterprise Systems in Small and Medium-Sized Enterprises .......................................... 53

Chapter 1.8. E-Government and ERP: Challenges and Strategies...................................................... 112

Chapter 1.9. Enterprise Application Integration (EAI) ....................................................................... 128

Chapter 1.10. Enterprise Tomography: An Efficient Approach for Semi-Automatic Localization

Chapter 1.12. From ERP to Enterprise Service-Oriented Architecture .............................................. 169

Chapter 1.13. Data Reengineering of Legacy Systems....................................................................... 181

Chapter 1.17. An Overview of Executive Information Systems ......................................................... 235

Chapter 1.19. Mix, Match, Rediscovery: A Mashup Experiment of Knowledge Organization in

Section II. Development and Design Methodologies

Chapter 2.4. Extending Enterprise Architecture with Mobility .......................................................... 356

Chapter 2.6. The Enterprise Systems Approach ................................................................................. 397

Chapter 2.10. An ERP Adoption Model for Midsize Businesses........................................................ 487

Chapter 2.13. Development and Design Methodologies in DWM ..................................................... 533

Section III. Tools and Technologies

Chapter 3.4. Transforming Compensation Management Practices through Web-Based Enterprise

Chapter 3.5. Business Process Management as a Critical Success Factor in EIS

Chapter 3.9. Towards a Model-Centric Approach for Developing Enterprise Information

Chapter 3.13. In-House vs. Off-the-Shelf e-HRM Applications......................................................... 752

Chapter 3.17. Mobile Technologies Extending ERP Systems ............................................................ 817

Chapter 3.21. Mining Association Rules from XML Documents....................................................... 879

Section IV. Utilization and Application

Chapter 4.1. QoS-Oriented Grid-Enabled Data Warehouses .............................................................. 901

Chapter 4.7. Consistency and Modularity in Mediated Service-Based Data Integration

Chapter 4.9. Implementation of ERP in Human Resource Management ......................................... 1021

Chapter 4.10. A Study of Information Requirement Determination Process of an Executive

Chapter 4.13. Multiple-Step Backtracking of Exception Handling in Autonomous Business

Chapter 4.14. A Resource-Based Perspective on Information Technology, Knowledge

Chapter 4.17. Specifying Software Models with Organizational Styles........................................... 1141

Chapter 4.18. Mobile Strategy for E-Business Solution ................................................................... 1162

Chapter 4.19. Application of Software Metrics in EPR Projects ...................................................... 1172

Section V. Organizational and Social Implications

Chapter 5.3. Enterprise Systems, Control and Drift ......................................................................... 1209

Chapter 5.5. People-Oriented Enterprise Information Systems ........................................................ 1247

Chapter 5.6. A SOA-Based Approach to Integrate Enterprise Systems............................................ 1265

Chapter 5.8. Experiences of Cultures in Global ERP Implementation ............................................. 1295

Chapter 5.11. Information Technologies as a Vital Channel for an Internal E-Communication

Section VI. Managerial Impact

Chapter 6.3. Managing Temporal Data ............................................................................................. 1461

Chapter 6.5. Identifying and Managing Stakeholders in Enterprise Information System

Chapter 6.6. Understanding Information Technology Implementation Failure: An Interpretive