Message Clustering based Matrix Factorization Model for Retweeting Behavior Prediction

Retweeting is an important mechanism for information diffusion in social networks. Through retweeting, message is reshared from one user to another user, forming large cascades of message forwarding. Most existing researches of predicting retweeting utilize user social relationships for modeling which leads to vast calculating amount. In this paper, we propose two message clustering based matrix factorization models for retweeting prediction. Unlike previous approaches, our models exploit the clustering relationships of messages instead of social relationships. Our models are quite general because we do not need any auxiliary information except for message content. Several experiments on real datasets show that our models are effective and outperform the state-of-the-art methods.