Journal of Computer Science 9 (8): 1008-1018, 2013
ISSN: 1549-3636
© 2013 Science Publications
doi:10.3844/jcssp.2013.1008.1018 Published Online 9 (8) 2013 (http://www.thescipub.com/jcs.toc)
Corresponding Author: Norwati Mustapha, Department of Computer Science, Faculty of Computer Science and Information Technology,
University Putra Malaysia, Serdang, Malaysia
1008
Science Publications JCS
USER RECOMMENDATION ALGORITHM IN SOCIAL
TAGGING SYSTEM BASED ON HYBRID USER TRUST
Norwati Mustapha, Wong Pei Voon and Nasir Sulaiman
Department of Computer Science, Faculty of Computer Science and Information Technology,
University Putra Malaysia, Serdang, Malaysia
Received 2013-02-21, Revised 2013-06-13; Accepted 2013-07-05
ABSTRACT
With the rapid growth of web 2.0 technologies, tagging become much more important today to facilitate
personal organization and also provide a possibility for users to search information or discover new things
with Collaborative Tagging Systems. However, the simplistic and user-centered design of this kind of
systems cause the task of finding personally interesting users is becoming quite out of reach for the common
user. Collaborative Filtering (CF) seems to be the most popular technique in recommender systems to deal
with information overload issue but CF suffers from accuracy limitation. This is because CF always been at-
tack by malicious users that will make it suffers in finding the truly interesting users. With this problem in
mind, this study proposes a hybrid User Trust method to enhance CF in order to increase accuracy of user
recommendation in social tagging system. This method is a combination of developing trust network based
on user interest similarity and trust network from social network analysis. The user interest similarity is de-
rived from personalized user tagging information. The hybrid User Trust method is able to find the most
trusted users and selected as neighbours to generate recommendations. Experimental results show that the
hybrid method outperforms the traditional CF algorithm. In addition, it indicated that the hybrid method
give more accurate recommendation than the existing CF based on user trust.
Keywords: User Trust, Tag, Collaborative Filtering
1. INTRODUCTION
Collaborative Tagging Systems (Begelman et al.,
2006; Gemmell et al., 2009a; 2009b; Hotho et al., 2006;
Shepitsen et al., 2008) allow users to label digital
resources using free-form of keywords (tags). The
simplistic and the user-centered design of this kind of
systems have encouraged many Web users to annotate their
data using tags (Begelman et al., 2006; Gemmell et al.,
2009a; 2009b; Hotho et al., 2006; Shepitsen et al., 2008).
Collaborative Tagging Systems allow users to explore
other users’ bookmarks via the keywords and tracking
users who bookmarked pages that you considered
interesting (Begelman et al., 2006; Gemmell et al.,
2009a; 2009b; Hotho et al., 2006; Shepitsen et al., 2008).
Nonetheless, users will find it hard to search other users
with similar interest within a social tagging system that
have hundreds of thousands of user with users racking up
tens of thousands of bookmarks.
CF algorithm seems to be the most popular technique in
recommender systems (Konstan et al., 1997; Resnick et al.,
1994; Sarwar et al., 2000; Tso-Sutter et al., 2008) to deal
with information overload issue. However, traditional CF
algorithm focuses only on similar users’ opinions which
express in ratings and do not consider the actual content of
the items, which affected the quality of the
recommendation. To improve recommendation quality,
metadata such as content information in items and tags have
been typically used as additional knowledge.
In the past few years, the dramatic expanding of Web
2.0 Web sites and applications poses new challenges for
traditional CF recommender systems. Traditional CF