A Random-Based Approach to Social Influence Maximization

Abstract

The problem of influence maximization is to find a small subset of nodes (seed nodes) as influential as the global optimum in a social network. Despite researchers have achieved fruitful research result which can be applied widely, a key limitation still remains, and that is this work is generally too time consuming to find seeds and difficult to be apply into the large-scale social network. In this paper, we propose a new random-based algorithm which combines “random selection” and “the optimal neighbor”, which idea both greatly reduces the computational complexity and achieves the desired effect. Our algorithm is able to avoid overlapped information and thus determine high-quality seed set for the influence maximization problem. Our empirical study with real-world social networks under independent cascade model (ICM) demonstrates that our algorithm significantly outperforms the common used algorithms in terms of efficiency and scalability, with almost no compromise of effectiveness.