Towards Understanding the Challenges Facing Effective Trust-Aware Recommendation

TitleTowards Understanding the Challenges Facing Effective Trust-Aware Recommendation
Publication TypeConference Paper
Year of Publication2010
AuthorsShi, Y, Larson, MA, Hanjalic, A
Conference NameRSWeb2010: Workshop on Recommender systems and the Social Web
PublisherACM
Conference LocationBarcelona, Spain
Abstract

We introduce a method for generating semi-synthetic social data collections, which we use to study trust-aware recommendation. Specifically, we examine the effects of social graph degree distribution on user-based collaborative filtering that substitutes trusted users for conventional neighbors. Our semi-synthetic data collec-tions are created via a naïve pruning process that maps a user-item matrix onto various social graphs with the degree distributions of real-world Web-based social systems. Our goal is to extend our understanding of the challenges facing effective trust-aware rec-ommendation beyond the current possibilities, which are limited by data set availability. The improvement offered by trust-aware recommendation is shown to have substantial dependence on the degree distribution of the social graph.

AttachmentSize
Towards Understanding the Challenges Facing Effective Trust-Aware Recommendation_RSWeb2010.pdf169.67 KB