PRNSGA-II: A Novel Approach for Influence Maximization and Cost Minimization Based on NSGA-II

Abstract

Influence maximization aims to extract a k-size seed node set to get a maximum influence spread under a specific propagation model which is a popular research topic in viral marketing these years. Companies want to select influential people to help them increase the sales of productions. With the limited budget, companies are unable to afford the huge cost of finding influential people. Thus, how to solve the multi-objective problem i.e. influence maximization problem and cost minimization problem (IM-CM) attracts more researchers attentions. In this paper, we propose a novel framework called PRNSGA-II to solve IM-CM. As an important index in complex networks, PageRank describes the importance of nodes. So we add PageRank to our first objective function to improve the quality of seed nodes. Then to calculate the cost of nodes, we use the degree centrality as our second objective function. Finally, we adopt NSGA-II which is a classical and effective multi-objective framework to solve IM-CM. We use three public datasets to verify our algorithm. The results of experiments demonstrate the effectiveness of our algorithm.