You are on page 1of 29

 Introduction

 Objective (Business & Text Mining Problems)


 Literature Review (Related Work)
 Data Understanding
 Data Preparation in IBM SPSS Modeler
 Modelling using IBM SPSS Modeler
 Modelling Results & Evaluation
 Basic Recommendations
Tay Eng Soon Library at SIM University headquarter

Loaning resources General collection


Languages collection Electronic resources
Computers
Photocopy, printing and Journals
scanning Recommended course
Infoskills workshop reference
Course collection & reference
Multimedia collection
From textual responses generated from open-
ended questions in the Biennial Library User Survey
of SIM Library conducted in year 2013,
Identify:
• areas of improvement
• Recommendations

810 student respondents


4 schools - School of Business (SBIZ), School of
Science and Technology (SST), School of Arts and
Social Sciences (SASS) and School of Human
Development and Social Services (HDSS)
Student Survey Identify areas for
Opinion Mining in improvement and
(Unstructured IBM SPSS Modeler recommendations
Textual Responses) for SIM Library

Online Feedback
Comment: IBM Text Mining node IBM Text Mining node
Form
“during exams
Concept
period the opening Category
Type
hours should be
further extended”
 Michalski (2011) - opinion mining to find out the
reasons students withdrew course from the Florida
Stage College
 616 textual responses
 Data cleaning - (1) identification and removal of
'useless' comments such as "none", "n/a", etc., (2)
standardization of common terms and abbreviations
("FSC" and 'college" referred to Florida Stage
College, " etc.) and (3) checking for misspellings.
 11 categories - "health", "financial", "family“, “job-
work”
 "who", "when" and "how“ -> “why”
Attributes Description
School Identifier for the textual response
Responses from students Content of the textual response
Categories identified by librarian when
Categories for reference collating the data
IBM SPSS
Text 29 types
Analytics –
Text Mining
Node

43 types
Insignificant concept

Mistyped concept

Unrecognized synonym
Creating new terms

Insert into existing synonym

Create new synonym


Creating new types

Exclude list
Category Category Rule Docs
Aspect -study ( <Chairs> | <Cubicles> | <StudyArea> | <PowerSocket> |
area (tables <Desks> | <Seats> ) & ( <Negative> | <NegativeFunctioning> | 139
and chairs) <Laptops> )
Aspect -
( <Website> | <Search> | <LogIn> | <Mobility> | graphical user
website and 90
interface ) & ( <Positive> | <Negative> | <NegativeFunctioning> )
search engine
Aspect - ( <ProposedFacility> | <ProposedResource> | <ProposedService>
proposed | <Rooms> | small | <Signage> | <Partnership> |
88
facility, resource <Event/Activity> | <Communication> | <Internet> |
and service <TertiaryInstitution> ) & !( <Negative> )
Aspect - ( <Librarian> | <CustomerSupport> | <PositiveAttitude> |
customer <PositiveCompetence> | <NegativeAttitude> | 85
Service <NegativeCompetence> | unwilling to read )
( <AudioVideoResources> | <BorrowingService> ) & (
Aspect – offline
<ResourceVariety> | <ResourceVolume> | <LoanDuration> | 77
resources
<LoanQuota> )
(554-73)/554 =
86.8%
Study area
Seats
Operating Hours
Offline resource (hardcopy books,
etc.)
Chairs
• Refine modeling results based on suggestion
generated from this presentation

• Compare the final categories created with


those pre-identified by the librarian

• Generate more in-depth identification of


areas of improvement and
recommendations for SIM Library

You might also like