User behavior prediction: A combined model of topic level influence and contagion interaction

People post, share and adopt short text/multimedia messages in OSNs every day. Understanding and being able to predict user behaviors in OSNs can be helpful for several areas such as viral marketing and advertisement. In this paper we propose a probabilistic model which combines the impacts from message interactions and topic level social influence to predict the user behavior of adopting contagions. Using two datasets: a collected Weibo data and a DBLP citation network, we testify that the combined model could predict user behavior more accurately.