Retweet prediction using social-aware probabilistic matrix factorization

Abstract

Retweet prediction is a fundamental and crucial task in social networking websites as it may influence the process of information diffusion. Existing prediction approaches simply ignore social contextual information or don’t take full advantage of these potential factors, damaging the performance of prediction. Besides, the sparsity of retweet data also severely disturb the performance of these models. In this paper, we propose a novel retweet prediction model based on probabilistic matrix factorization method by integrating the observed retweet data, social influence and message semantic to improve the accuracy of prediction. Finally, we incorporate these social contextual regularization terms into the objective function. Comprehensive experiments on the real-world dataset clearly validate both the effectiveness and efficiency of our model compared with several state-of the-art baselines.