An Efficient Influence Maximization Technique Based on Betweenness Centrality Measure and Clustering Coefficient (Bet-Clus)

Abstract

Graph theory has found its rigorous applicability in defining the social network phenomenas. Influence maximization in real world graphical networks have been an area of keen interest to researcher. Determining the initial seed set population for an influence spread model is found to be an N-P complete problem. Many optimization and heuristic algorithms have been proposed to solve the problem in the past. However, all the proposed solutions are constrained in terms of the scalability, their correctness etc. In this article, a graph theoretical approach based metric has been defined to effectively identify the initial seed set population for the network. The seed nodes are selected based on their clustering coefficient and betweenness centrality scores on a ranking basis. The top ’k’ ranked nodes are selected as seed nodes (information carriers) in the network. Using Independent Cascade (IC) diffusion model, the network coverage is calculated. The proposed method attains higher network coverage in comparison to the base IC model. Also, the method’s performance is compared and found to be superior when the seed nodes are selected based on other centrality measures and graph measures. The results have been evaluated over four different benchmark networks- CORA, Citeseer, PubMed (Citation Networks) and Amazon Computers Network (Product Network).