Personalized Microblog Recommendation Based On Collaborative Filtering And Social Network Mining

Microblogs, such Sina Weibo, Tencent Weibo have rapidly grown in-to popular social network services recently. They provides large numbersof real-time messages for users and user can get fresh information by scan-ning their timeline. Microblog messages are presented in chronologicalorder and users scan the followees’ timelines to fnd what they are inter-ested in. However, an information overload problem has troubled manyusers, especially those with many followees and thousands of tweets ar-riving every day. In this paper, we focus on recommending useful tweetsthat users are really interested in personally to reduce the users’ efort tofnd useful information. Many kinds of information on Twitter are avail-able for helping recommendation, including the user’s own tweet history,retweet history and social relations between users. We propose a methodof making tweet recommendations based on collaborative fltering and so-cial network mining to capture personal interests. It can also convenientlyintegrate the other useful contextual information. Our fnal method con-siders three major elements on Microblogs: tweet topic level factors, usersocial relation factors and explicit features such as authority of the pub-lisher and quality of the tweet. The experiments show that all the proposedelements are important and our method greatly outperforms several base-line methods.