Detection of key figures in social networks by combining harmonic modularity with community structure-regulated network embedding (Q6092051)

This paper studies detection of key figures in social networks by combining harmonic modularity with community structure-regulated network embedding. It proposes an improved harmonic modularity algorithm based on Q-modularity gain. The paper defined the direct modularity increment to solve the dense intracommunity connections of internal nodes. The indirect modularity increment was defined to solve the two-step distance of a group's shared common neighbors. To improve the harmonic modularity algorithm, modularity gain is employed to measure the harmony and diversity between nodes. The paper also proposes the social rank and community structure-regulated network embedding model to detect opinion leaders. A figure detection algorithm is used to detect structural hole spanners and opinion leaders. The opinion leader is the most important influencer of intra-community nodes, and the structural hole spanner bridges the disparate nodes. Both are key figures in social network analysis. Four experiments including community detection, structural hole spanner detection, link prediction, and opinion leader detection are designed to compare the methods.