Complex network community detection method by improved density peaks model

Abstract Searching for key nodes of complex network and clustering communities is an important and practical solution in community detection algorithms. According to above-mentioned notion, we proposed a novel complex community detection method by improved density peaks model, called IDPM. Primarily, the composite similarity is acquired by normalizing Jaccard and shortest path feature. Secondarily, threshold condition is acceded to density peaks model to obtain key nodes of network. Tertiarily, non-key nodes are assigned into groups and form base communities. The attribute of unstable nodes are corrected iteratively until the entire network is stable by neighbor estimation principle. The simulation experiments prove that the IDPM can obtain better effects of community detection on synthesize and real-world complex network.