When Diversity Meets Speciality: Friend Recommendation in Online Social Networks

Online social networks improve social experience by con-necting users with common interests. Similar to real life, seeking good friends is much easier with recommenda-tions in online social networks. In this paper, we investi-gate a series of problems related to friendship formation in the hope of improving friend recommendation in social networks. Specially, we seek to understand whether users who contribute more are more popular among other users, whether users like to make friends with popular users and the role difference of users with different diversity of in-dividual interests in friendship formation. We propose a novel approach based on topic modeling to characterize the interest diversity degree of each user. The interest diversity features are used to help predict friend relation-ships between users. The experimental results on three large-scale datasets demonstrate the effectiveness of our method.