Influence Maximization on Signed Social Networks with Integrated PageRank

Online social networks (OSNs) have received a lot of attentions recently since they provide a new platform for product promotion and online viral marketing. Influence maximization problem has been extensively studied on some existing influence diffusion models in number of domains. However, most of the existing studies consider OSNs as friendly networks only containing friendship relationships, whereas the hostile relations do exist in many OSNs in real life, e.g., Epinions and Slashdot. In this paper, we integrate the PageRank on signed social networks and use the integrated PageRank to study influence maximization in OSNs with both friend and hostile relations which are respected as positive edges and negative edges on signed networks. In addition, we use the extended vote model to study the influence diffusion on signed networks. We then conducted comprehensive experiments on real social networks to select initial k seeds for influence maximization, and results indicate that our integrated PageRank method performs better than other heuristic algorithms.