A Product Recommendation Approach Based on the Latent Social Trust Network Model for Collaborative Filtering

Recommender systems take advantage of dynamic and collective knowledge to make personalized recommendations to each user. Collaborative filtering as a well-known technique in recommender systems often encounters some challenges such as spare rating data and malicious attacks. Trust-based collaborative filtering employs the social trust network to make recommendations in order to alleviate the above problems. Unfortunately, explicit trust information is quite deficient, which leads to the limited recommendation capability. Therefore, a latent social trust network model is proposed to improve the recommendation performance. The latent social trust comes from the coupling trust and the co-citation trust as well as the similar interests between users. Based on the latent trust information, a new social trust network can be built and then be used to predict the target user's taste. The experimental results demonstrate that our approach can rationally infer the trust relationships between users and highly improve the recommendation performance.