Research On Friends Recommendation In Social Network

With development of web2.0technology, SNS become one of the most potential developed areas. For new users, how to migrate friends between networks becomes a difficult problem. Thus how to help users to expand friends is one of the difficulties.In this dissertation, we first analyze different recommendation model classified by recommended methodology and recommendation algorithm applied in social network. Then we conduct a study for two sets of issues:hot friends recommendation and personalized friends recommendation. In hot friends recommendation section, we classify user’s behavior three aspects:user profile creation, content creation and relationship building. We propose social network surfing model, and design an evaluation indicator to measure user’s contribution. Then we use this evaluation indicator to recommend hot friends. In personalized friends recommendation part, we present a two-stage recommendation model and a trust propagation based recommendation model. In two-stage recommendation model, we describe users’ similarity in two ways, and take into account the user’s friends on user’s decision-making. In trust propagation based recommendation model, we first analyze trust propagation model in social network, and then present trust-based recommendation algorithm.Then we analyze these issues in the experiment. It was found that most users in social network have relatively small contribution, and only a few users have the relatively large contribution. On the other hand, we compare the effect of different algorithm presented in personalized friends recommend part. The experimental results show that trust-based recommendation algorithm is better than the other. This is mainly because the trust-based recommendation algorithm takes into account affect from indirect friends.The major work of this dissertation is as follows:(1) Unscramble the main research of today’s recommendation algorithm and its application in social network. (2) For hot friends use social network surfing model to analyze. For personalized friends import two-state model and trust propagation based recommendation model.(3) Prove the the feasibility of above model through experiments.