Influence Maximization in Online Social Networks Using Community Structures

Abstract

line social networks (OSNs) have  dominated modern life on a global scale. The immense  popularity of online social networks increases day by day  as they help us in modelling various types of processes like  viral marketing, rumor controlling, collaborative filtering,  market prediction and controlling diseases spread. In the  realm of complex networks research, influence  maximisation in social networks has long been a  challenging task. Influence maximisation is the method of  identifying k-seed nodes or influential nodes in order to  increase overall influence in a network. Ranking the nodes  using network node-centrality metrics is one of the  conventional techniques for finding prominent nodes in a  social network. However, estimating global centrality  metrics like betweenness centrality is computationally  exhaustive and typically not scalable for very large size  networks such as a country's whole population. In this  paper, we provide a novel approach for extracting communities from the underlying social network to  identify prominent nodes aka "influential nodes”.  Experimental results indicate that the seed nodes identified  by the proposed approach have high betweenness  centrality in the social network thus rendering the proposed  approach significant.