Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Download
Standard view
Full view
of .
Save to My Library
Look up keyword
Like this
1Activity
0 of .
Results for:
No results containing your search query
P. 1
A Role-Oriented Content-based Filtering Approach: Personalized Enterprise Architecture Management Perspective

A Role-Oriented Content-based Filtering Approach: Personalized Enterprise Architecture Management Perspective

Ratings: (0)|Views: 44 |Likes:
Published by ijcsis
In the content filtering-based personalized recommender systems, most of the existing approaches concentrate on finding out similarities between users’ profiles and product items under the situations where a user usually plays a single role and his/her interests persist identical on long term basis. The existing approaches argue to resolve the issues of cold-start significantly while achieving an adequate level of personalized recommendation accuracy by measuring precision and recall. However, we investigated that the existing approaches have not been significantly applied in the context where a user may play multiple roles in a system simultaneously or may change his/her role overtime in order to navigate the resources in distinct authorized domains. The example of such systems is enterprise architecture management systems, or e-Commerce applications. In the scenario of existing approaches, the users need to create very different profiles (preferences and interests) based on their multiple /changing roles; if not, then their previous information is either lost or not utilized. Consequently, the problem of cold-start appears once again as well as the precision and recall accuracy is affected negatively. In order to resolve this issue, we propose an ontology-driven Domain-based Filtering (DBF) approach focusing on the way users’ profiles are obtained and maintained over time. We performed a number of experiments by considering enterprise architecture management aspect and observed that our approach performs better compared with existing content filtering-based techniques.
In the content filtering-based personalized recommender systems, most of the existing approaches concentrate on finding out similarities between users’ profiles and product items under the situations where a user usually plays a single role and his/her interests persist identical on long term basis. The existing approaches argue to resolve the issues of cold-start significantly while achieving an adequate level of personalized recommendation accuracy by measuring precision and recall. However, we investigated that the existing approaches have not been significantly applied in the context where a user may play multiple roles in a system simultaneously or may change his/her role overtime in order to navigate the resources in distinct authorized domains. The example of such systems is enterprise architecture management systems, or e-Commerce applications. In the scenario of existing approaches, the users need to create very different profiles (preferences and interests) based on their multiple /changing roles; if not, then their previous information is either lost or not utilized. Consequently, the problem of cold-start appears once again as well as the precision and recall accuracy is affected negatively. In order to resolve this issue, we propose an ontology-driven Domain-based Filtering (DBF) approach focusing on the way users’ profiles are obtained and maintained over time. We performed a number of experiments by considering enterprise architecture management aspect and observed that our approach performs better compared with existing content filtering-based techniques.

More info:

Published by: ijcsis on Nov 02, 2010
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

11/02/2010

pdf

text

original

 
A Role-Oriented Content-based Filtering Approach:Personalized Enterprise Architecture Management Perspective
Imran Ghani, Choon Yeul Lee, Seung Ryul Jeong,Sung Hyun Juhn
(School of Business IT, Kookmin University, 136-702, Korea)E-mail: imransaieen@gmail.com;{cylee, srjeong, juhn} @kookmin.ac.kr
Mohammad Shafie Bin Abd Latiff 
(Faculty of Computer Science and Information Systems UniversitiTeknologi Malaysia, 81310, Malaysia)E-mail: shafie@utm.my
 Abstract
-
In the content filtering-based personalizedrecommender systems, most of the existing approachesconcentrate on finding out
similarities between users’
profiles and product items under the situations where auser usually plays a single role and his/her interestspersist identical on long term basis. The existingapproaches argue to resolve the issues of cold-startsignificantly while achieving an adequate level of personalized recommendation accuracy by measuringprecision and recall. However, we investigated that theexisting approaches have not been significantly appliedin the context where a user may play multiple roles in asystem simultaneously or may change his/her roleovertime in order to navigate the resources in distinctauthorized domains. The example of such systems isenterprise architecture management systems, or e-Commerce applications. In the scenario of existingapproaches, the users need to create very differentprofiles (preferences and interests) based on theirmultiple /changing roles; if not, then their previousinformation is either lost or not utilized. Consequently,the problem of cold-start appears once again as well asthe precision and recall accuracy is affected negatively.In order to resolve this issue, we propose an ontology-driven Domain-based Filtering (DBF) approach
focusing on the way users’ profiles are obtained and
maintained over time. We performed a number of experiments by considering enterprise architecturemanagement aspect and observed that our approachperforms better compared with existing contentfiltering-based techniques.
 Keywords:
role-oriented content-based filtering,recommendation, user profile, ontology, enterprisearchitecture management 
1
 
INTRODUCTIONThe existing content-based filtering
approaches
 
(Section 2) claim determining the similarities
 between user‟s
interests and preferences with productitems available in the same category. However, weinvestigated that these
approaches
 
achieve soundresults under the situations where a user normallyplays a particular role. For instance, in e-Commerceapplications a user may upgrade his/her subscriptionpackage from normal customer to a premiumcustomer or vice versa. In this scenario, the existingrecommender systems usually manage to recommend
the information related to a user‟s
new role. However,if a user wishes the system to recommend him/herproducts as a premium as well as a normal customerthen the user needs to create different profiles(preferences and interests) and has to login based onhis/her distinct roles. Likewise, EnterpriseArchitecture Management Systems (EAMS)emerging from the concept of EA [18] deals withmultiple domains whereas a user may performseveral roles and responsibilities. For instance, asingle user may hold a range of roles such as aplanner, analyst and EA managers or a designer anddevelopers or constructors and so on. In addition, auser
s role may change over time creating a chain of roles from current to past. This setting naturally leadsthem to build up very different preferences andinterests corresponding to the respective roles. On theother hand, a typical EAMS manages enormousamount of distributed information related to severaldomains such as application software, projectmanagement, system interface design and so on. Eachof the domains manages several models, components,schematics, principles, business and technologyproducts or services data, business process andworkflow guides. This in turn creates complexity in
deriving and managing users‟ preferences
andselecting right information from a tremendousinformation-base and recommending to the rightusers
roles. Thus, when the
user‟s role
is not specific,the recommendation becomes more difficult inexisting content-based filtering techniques. As aresult they do not scale well in this broader context.In order to limit the scope, this paper focuses on thescenario of EAMS and the implementation related toe-Commerce systems is left to the future work.The next section describes a detailed survey of thefiltering techniques and their limitations relevant tothe concern of this paper.
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 8, No. 7, October 20109http://sites.google.com/site/ijcsis/ISSN 1947-5500
 
2
 
RELATED WORK AND LIMITATIONSA number of content-based filtering techniques [1][2][3][4][5][10][17] have emerged that are used topersonalize information for recommender systems.These techniques are inspired from the approachesused for solving information overload problems[11][15]. As mentioned before in Section 1 that acontent-based system filters and recommends an itemto a user based upon a description of the item and a
 profile of the user‟s interests. While a user profile
may either be entered by the user, it is commonlylearned from feedback the user provides on items or
implicitly obtained from user‟s recent browsing (RB)
activities. The aforementioned techniques andsystems usually use data obtained from the RBactivities that pose significant limitations onrecommendation as summarized in the followingtable.
1.
 
There are different approaches to learning a model of 
the user‟s interest with content
-based recommendation,but no content-based recommendation system can givegood recommendations if the content does not containenough information to distinguish items the user likes
from items the user doesn‟t like in a particular context
such as if a user plays different roles in a systemsimultaneously.2.
 
The existing approaches do not scale well to filter the
information if a user‟s role is frequently changed
whichcreates a chain of roles (from current to past) for asingle user.
If the user‟s role is changed from project
manager to EA manager, this leads the users to berestricted to seeing items similar to those not relevantto the current role and preferences.
Based on the above concerns, it has been noted that anumber of filtering processing techniques exist whichhave their own limitations. However, there are nostandards to process and filter the data, so wedesigned our own technique called Domain-basedFiltering (DBF).3
 
MOTIVATIONTypically, there are three categories of filteringtechniques classified in the literature [12] including;(1) ontology based systems; (2) trust network basedsystems; and (3) context-adaptable systems thatconsider the current time and place of the user. Thescope of this paper, however, is the ontology basedsystems and we have taken the entire EnterpriseArchitecture Management (EAM) area intoconsideration for ontology-based role-orientedcontent filtering. This is because of the fact thatEnterprise Architecture (EAs) produce vast volumesof models and architecture documents that haveactually added difficulties for
organization‟s
capability to advance the properties and qualities of its information assets
with respect to the user‟s need
.The users need to consult a vast amount the currentand previous versions of the EA information assets inmany cases to comply with the standards. Though, anumber of EAMS have been developed howevermost of them focus on the content-centric aspect[6][7][8][9] but not on the personalization aspect.Therefore, at EAMS level, there is a need for filteringtechnique that can select and recommend informationwhich is personalized (relevant and understandable)for a range of enterprise users such as planners,analysts, designers, constructors, information assetowners, administrators, project managers, EAmanagers, developers and so on to serve for betterdecision making and information transparency atenterprise-wide level. In order to achieve this featureeffectively; the semantics-oriented ontology-basedfiltering and recommendation techniques can play avital role. The next section discusses the proposedapproach.4
 
PHYSICAL AND LOGICAL DOMAINSIn order to illustrate the detailed structure of DBF,it is appropriate to clarify that we have classified twotypes of domains to deal with the data at EAMS levelnamed physical domains (PDs) and logical domains(LDs). The PDs have been defined to classifyenterprise assets knowledge (EAK). The EAK is themetadata about information resources/items includingartifacts, models, processes, documents, diagramsand so on using RDFS [14] with class hierarchies(Fig 1) and RDF[13] based triple subject-predicate-object format (Table 2). Basically, the concept of PDis similar to organize the product categories in exitingontology-based ecommerce systems, such as salesand marketing, project management, datamanagement, software applications, and so on.
TABLE1: LIMITATIONS IN EXISTING APPROACHES
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 8, No. 7, October 201010http://sites.google.com/site/ijcsis/ISSN 1947-5500
 
 On the other hand, LDs deal with manipulating
the user‟s profiles organized in user model ontology
(UMO). The UMO is organized in ResourceDescription Framework [13] based triple subject-predicate-object format (Table 3). We name this asLD because it is reconfigurable according to thechanging or multiple roles of users and their interestslist. Besides, an LD can be deleted if a user leaves theorganization. On the other hand PDs are permanentinformation assets of an enterprise and all theinformation assets belong to the PDs.We discuss the DBF approach in the followingsection.
 
5
 
DOMAIN-BASED FILTERING (DBF)APPROACHAs mentioned before in Section 1 that the existingcontent-base filtering techniques attempt torecommend items similar to those a given user hasliked in the past. This mechanism does not scale wellin role-oriented settings such as in EAM systemswhere a user changes his/her role or play multipleroles simultaneously. In this scenario, the existingtechniques still bring the old items relevant to thepast roles of users which may no longer be desirableto the new role of the user. In our research weworked out to find that there are other criteria thatcould be used t
o classify the user‟s information for filtering purposes. By observing the users‟ profiles, it
has been noted that we can logically distinguish
among users‟ functional and non
-functional domainsfrom explicit data collection (when a user is asked tovoluntarily provide their valuations including pastand current roles and preferences) and implicit data
collection (where the user‟s behavior is monitored)
during browsing the system while holding currentroles or the roles he/she performed in past.DBF approach performs its filtering operations by
logically classifying the users‟ profiles based on
current and past roles and interests list. Creating LDout
of the users‟ profiles is a system generated
process which is achieved by exploring the
users‟
roles-interests
similarities as a filtering criterion.Fig 1: Physical domain (PD) hierarchyTABLE 2: RDF-BASED INFORMATIONASSETS TRIPLETABLE 3: RDF-BASED UMO
Classes andsubclasses
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 8, No. 7, October 201011http://sites.google.com/site/ijcsis/ISSN 1947-5500

You're Reading a Free Preview

Download
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->