Link to the University of Pittsburgh Homepage
Link to the University Library System Homepage Link to the Contact Us Form

Personalized Recommendations Based On Users’ Information-Centered Social Networks

Lee, Danielle (2013) Personalized Recommendations Based On Users’ Information-Centered Social Networks. Doctoral Dissertation, University of Pittsburgh. (Unpublished)

Primary Text

Download (5MB) | Preview


The overwhelming amount of information available today makes it difficult for users to find useful information and as the solution to this information glut problem, recommendation technologies emerged. Among the several streams of related research, one important evolution in technology is to generate recommendations based on users’ own social networks. The idea to take advantage of users’ social networks as a foundation for their personalized recommendations evolved from an Internet trend that is too important to neglect – the explosive growth of online social networks. In spite of the widely available and diversified assortment of online social networks, most recent social network-based recommendations have concentrated on limited kinds of online sociality (i.e., trust-based networks and online friendships). Thus, this study tried to prove the expandability of social network-based recommendations to more diverse and less focused social networks. The online social networks considered in this dissertation include: 1) a watching network, 2) a group membership, and 3) an academic collaboration network. Specifically, this dissertation aims to check the value of users’ various online social connections as information sources and to explore how to include them as a foundation for personalized recommendations.
In our results, users in online social networks shared similar interests with their social partners. An in-depth analysis about the shared interests indicated that online social networks have significant value as a useful information source. Through the recommendations generated by the preferences of social connection, the feasibility of users’ social connections as a useful information source was also investigated comprehensively. The social network-based recommendations produced as good as, or sometimes better, suggestions than traditional collaborative filtering recommendations. Social network-based recommendations were also a good solution for the cold-start user problem. Therefore, in order for cold-start users to receive reasonably good recommendations, it is more effective to be socially associated with other users, rather than collecting a few more items. To conclude, this study demonstrates the viability of multiple social networks as a means for gathering useful information and addresses how different social networks of a novelty value can improve upon conventional personalization technology.


Social Networking:
Share |


Item Type: University of Pittsburgh ETD
Status: Unpublished
CreatorsEmailPitt UsernameORCID
Lee, Daniellehyl12@pitt.eduHYL12
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairBrusilovsky, Peterpeterb@pitt.eduPETERB
Committee MemberHe, Daqingdah44@pitt.eduDAH44
Committee MemberHirtle, Stephen C.hirtle@pitt.eduHIRTLE
Committee MemberSchleyer, Titustitus@pitt.eduTITUS
Committee MemberButler,
Date: 13 May 2013
Date Type: Publication
Defense Date: 9 November 2012
Approval Date: 13 May 2013
Submission Date: 15 April 2013
Access Restriction: No restriction; Release the ETD for access worldwide immediately.
Number of Pages: 317
Institution: University of Pittsburgh
Schools and Programs: School of Information Sciences > Information Science
Degree: PhD - Doctor of Philosophy
Thesis Type: Doctoral Dissertation
Refereed: Yes
Uncontrolled Keywords: Personalized Recommendations; Recommendations; Social Network-based Recommendations; Online Social Networks; Information Similarity; Watching Network; Group Membership; Collaboration Network; Citeulike; ConferenceNavigator
Date Deposited: 13 May 2013 17:58
Last Modified: 15 Nov 2016 14:12


Monthly Views for the past 3 years

Plum Analytics

Actions (login required)

View Item View Item