An efficient algorithm for mining a set of influential spreaders in complex networks

Abstract Identifying the influential nodes in social network is of significance for information spreading, virus control and contagious disease detection. In this paper, the problem of influential spreaders selection is transferred into a problem to find groups with dense connections. Inspired by the fact that the network clustering coefficient would increase with the removal of peripheral nodes by the k-shell decomposition method, we select nodes with the highest k-shell value and interconnected with each other as the core to form an initial group. Then the neighbour nodes closely connected to the group are gradually added into it. The most influential node identified by degree centrality in each dense group would finally be selected as the initial spreaders and the k-shell value of all nodes in the group are set to 0 before searching for the next group. Therefore, the proposed method can guarantee not only the spreaders themselves are influential, but also the distance among them is relatively scattered. The experimental results in six real networks indicate that the spreaders identified by the method are more influential than several benchmark algorithms, including the discount degree method, VoteRank, LIR, k-shell and degree centrality.