Reference Hub3
Recommender System with Composite Social Trust Networks

Recommender System with Composite Social Trust Networks

Chaochao Chen, Xiaolin Zheng, Mengying Zhu, Litao Xiao
Copyright: © 2016 |Volume: 13 |Issue: 2 |Pages: 18
ISSN: 1545-7362|EISSN: 1546-5004|EISBN13: 9781466689053|DOI: 10.4018/IJWSR.2016040104
Cite Article Cite Article

MLA

Chen, Chaochao, et al. "Recommender System with Composite Social Trust Networks." IJWSR vol.13, no.2 2016: pp.56-73. http://doi.org/10.4018/IJWSR.2016040104

APA

Chen, C., Zheng, X., Zhu, M., & Xiao, L. (2016). Recommender System with Composite Social Trust Networks. International Journal of Web Services Research (IJWSR), 13(2), 56-73. http://doi.org/10.4018/IJWSR.2016040104

Chicago

Chen, Chaochao, et al. "Recommender System with Composite Social Trust Networks," International Journal of Web Services Research (IJWSR) 13, no.2: 56-73. http://doi.org/10.4018/IJWSR.2016040104

Export Reference

Mendeley
Favorite Full-Issue Download

Abstract

The development of online social networks has increased the importance of social recommendations. Social recommender systems are based on the idea that users who are linked in a social trust network tend to share similar interests. Thus, how to build an accurate social trust network will greatly affect recommendation performance. However, existing trust-based recommender approaches do not fully utilize social information to build rational trust networks and thus have low prediction accuracy and slow convergence speed. In this paper, the authors propose a composite trust-based probabilistic matrix factorization model, which is mainly composed of two steps: In step 1, the existing explicit trust network and the inferred implicit trust network are used to build a composite trust network. In step 2, the composite trust network is used to minimize both the rating difference and the trust difference between the true value and the inferred value. Experiments based on an Epinions dataset show that the authors' approach has significantly higher prediction accuracy and convergence speed than traditional collaborative filtering technology and the state-of-the-art trust-based recommendation approaches.

Request Access

You do not own this content. Please login to recommend this title to your institution's librarian or purchase it from the IGI Global bookstore.