Community Structure Identification in Networks via Detecting Community Center Method

Identifying community structure in a simple network G(V, E) is essentially a clustering problem. The most important step to solve this problem is to determine the number of communities and find out community centers. Therefore, each node's local centrality CL and local dissimilarity δ are introduced. Nodes which have both high local centrality and local dissimilarity are chosen as the centers of the communities and the number of centers equals to the number of communities. After that we assign each node to its corresponding community by using Kmeans clustering method or Hierarchical clustering method. In the last part we applied our approach to a real relationship network and made a very satisfactory result.