A Three-Step Approach for Influence Maximization in Social Networks

Scale-free property shows that complex networks tend to have a set of most influential key nodes. Influence maximization (IM) is a problem of identifying a set of most influential key nodes within a given network propagation model. However, most of the traditional IM algorithms cannot well balance the propagation range and computational efficiency. This paper proposes a three-step approach integrating the greedy, community detection and K-shell decomposition in order to solve the IM. The experimental results indicate that the proposed three-step approach is close to the greedy algorithm and better than several centrality-based approaches including degree, closeness, and the K-shell in terms of the propagation range, but runs several orders of magnitude faster than the greedy.