A New Information Diffusion Model in Weibo: UA-SCIR

Abstract

To illustrate how information spreads in Weibo, this paper proposed a user attributes based diffusion model (UA-SCIR) which defined four node states: susceptible, contacted, infected and refractory. The contacted state was first introduced to describe the nodes who had read the message but had not decided whether reposting or ignoring. Moreover, we recalculated the transfer probability between various states based on user attributes. Experiments on real-world dataset show that our UA-SCIR model can demonstrate the information diffusion progress more accurately than the existing models.