A collective synchronous behavior model on social media

Collective synchronous behavior is a pervasive phenomenon that we can discover in nature and virtual social media. Traditional data mining methods, however, mainly concentrate on analysis of individual behavior. In sociology, many well-known models are not suitable for the social media environment as well, in which huge amounts of data are generated everyday. In this paper, we proposed an innovative model that consists of multiple hidden Markov chains. By learning from the observations from a group of people, our model can not only predict the steady future state of a collective, but also measure the dependency property, reactive factor, of individuals. Experiment result shows that our model has ability to distinguish the behaviors of different persons.