Exploiting social circle broadness for influential spreaders identification in social networks

Influential spreaders identification in social networks contributes to optimize the use of available resources and ensure the more efficient spread of information. In contrast to common belief that highly connected or core located users are most crucial spreaders, this paper shows that both user's local and global structural properties matter in information diffusion. We propose a new metric, social circle broadness, to measure a user's information spreading influence by qualitatively combining the two above properties. Firstly, a definition of social circle diversity is introduced to measure the dispersion extent of a user's friends distribution in the network. Based on it, a method to calculate each user's local social circle broadness is presented. Preliminary experiments on a coauthor dataset demonstrate the effectiveness of social circle broadness in information diffusion. Furthermore, a social circle weighted PageRank (SCWPR) algorithm is proposed to iteratively rank each user's global social circle broadness. We conduct extensive comparison experiments against six state-of-the-art baseline methods on four real social network datasets. The results show that SCWPR outperforms all of them for influential spreaders identification in information propagation.