Using social network information to enhance collaborative filtering performance

摘要

Although recommender systems have been comprehensively analysed in the past decade, the study of social-based recommender have not been studied fully. Many social networks capture the relationships among the nodes by using trust scores to label the edges. The bias of a node denotes its propensity to trust/mistrust its neighbours and is closely related to truthfulness. It is based on the idea that the recommendation of a highly biased node should be removed. In this paper, we compute the bias and prestige of nodes in networks where the edge weight denotes the trust score. We propose a model-based approach for recommendation employing matrix factorisation after removing the bias nodes from each link, which naturally fuses the users' tastes and their trusted friends' favours together. Through experiments on publicly available data, we demonstrate that the proposed recommendation models can better utilise user's social trust information, resulting in increased recommendation accuracy.