Social Recommendation With Multiple Influence From Direct User Interactions

With the sustained development of social networks, increasing attention has been paid to social recommender systems. Current studies usually focus on indirect factors such as the similarity between users, but multiple direct interactions, such as mentions, reposts, and comments, are seldom considered. This paper addresses direct connections between users in social recommender systems. We analyze direct interactions to investigate the connection strength between users, and then, user preferences and item characteristics can be better described. Based on the analysis of social influence between users and users' influence over the whole social network, we propose a recommendation method with social influence, which makes full use of information among users in social networks and introduces the mechanisms of macroscopic and microscopic influences. Direct interactions between users are incorporated into a matrix factorization objective function. Real-world microblog data are applied to verify our model, and the results show that the proposed recommendation method outperforms other state-of-the-art recommendation algorithms.