An extended validity index for identifying community structure in networks

Abstract

To find the best partition of a large and complex network into a small number of communities has been addressed in many different ways. In this paper, a new validity index for network partition is proposed, which is motivated by the construction of Xie-Beni index in Euclidean space. The simulated annealing strategy is used to minimize this extended validity index, associating with a dissimilarity-index-based k-means iterative procedure, under the framework of a random walker Markovian dynamics on the network. The proposed algorithm(SAEVI) can efficiently and automatically identify the community structure of the network and determine an appropriate number of communities without any prior knowledge about the community structure during the cooling process. The computational results on several artificial and real-world networks confirm the capability of the algorithm. © 2010 Springer-Verlag.