Personalized re-ranking of tweets

In microblogs, the problem of information overload has troubled many users especially those with numerous followees. Users receive hundreds of tweets in chronological order and have to scan through pages of tweets to find useful information. In this paper, we propose a personalized tweet re-ranking framework for re-ranking the tweets received by a user based on his preference such that interesting tweets are ranked higher for the user. With the personalized re-ranked tweets, the user can find his interesting tweets conveniently. Modeling users’ preference in the context of tweet streams is more challenging than modeling that in the context of long documents as it is difficult to capture users’ interests with sparse short text documents like tweets. To address this challenge, we propose a media awareness tweet re-ranking model, MATR for short, to incorporate WeMedia accounts (WeMedia is a type of accounts in microblogs that only has media attributes publishing original and valuable messages), and explicitly calculate the influence of the publishers of these tweets. Experimental results demonstrate the effectiveness of our method compared to state-of-the-art baselines.