Who are More Influential Spreaders? Users with Broader Social Circle or with More Friends?

摘要

Social network plays an important role in information diffusion. Identifying the most influential spreaders can contribute to the prediction of the speed and scale of information dissemination, which is very valuable to business such as advertising delivering. In contrast to common belief that users who are highly connected are most important spreaders, this paper shows that user's social circle play a more important role in information diffusion. In this paper, we firstly define the diversity of a user's social circle, which means the dispersion extent of the user's friend distribution in the network. A method for measuring diversity and broadness of a user's social circle in network is also presented. Preliminary experiment demonstrates that user's social circle does matter in information diffusion. Moreover, a social circle weighted PageRank algorithm which takes both popularity (in-degree) and social circle broadness into account is proposed. Experiments on Susceptible-Infectious-Recovered (SIR) model show that our method in terms of identification of influential spreaders outperforms other methods such as in-degree and PageRank on two real social network datasets.