User recommendation with tensor factorization in social networks

Abstract

The rapid growth of population in social networks has posed a challenge to existing systems for recommending to a user new friends having similar interests. In this paper, we address this user recommendation problem in social networks by proposing a novel framework which utilizes users' tagging information with tensor factorization. This work brings two major contributions: (1) A tensor model is proposed to capture the potential association among user, user's interests and friends in social tagging systems; (2) A novel approach is proposed to recommend new friends based on this model. The experiments on a real-world dataset crawled from Last.fm show that the proposed method outperforms other state-of-the-art approaches.