A community-based entropic method to identify influential nodes across multiple social networks

Abstract Identifying influential nodes in social networks − by determining the most effective and efficient set of primary users − is a crucial task for maximizing the spread of influence. The competitive world of social networks and the variety of services they provide have persuaded users to apply for the membership of several networks. This situation makes involvement of the shared individuals across multiple social networks essential for the effective dissemination of information. In this paper, we introduce a community-based entropic method to identify influential nodes in multiple social networks, taking into account the effect of shared users. The proposed method, titled IM2-SE , focuses on calculating a total structural entropy for each node. This structural entropy is defined as the weighted sum of the structural entropies of the node in the networks and in the communities, it belongs to. The structural entropy of a node is based on Tsallis entropy considering both local and global structures of the network, and it utilizes the degree of centrality and betweenness centrality. Then, the nodes with the highest total structural entropy are identified as the influential nodes. The empirical evaluation of the proposed method on the real-world datasets including homogeneous and heterogeneous networks, demonstrates its superior performance to the baseline methods. Furthermore, the impact of the shared nodes between networks in identifying the influential nodes has been examined, revealing that up to 50% of the identified influential nodes are the shared nodes between specific networks.