Cooperative community detection algorithm based on random walks

Abstract

Community Detection is a significant tool for understanding the structures of real-world networks. Although many novel methods have been applied in community detection, as far as we know, cooperative method has not been applied into community detection to improve the performance of discovering community structure of social networks. In this paper, we propose a cooperative community detection algorithm, named cooperative community detection algorithm based on random walks. Firstly, it uses random walks to calculate the similarities between adjacent nodes, and then translates a given unweighted networks into weighted networks based on the similarities between adjacent nodes. Secondly, it detects community structures of networks by activating the neighbors a node whose community label is known. Thirdly, it cooperates running results of many times of our community detection algorithm to improve its accuracy and stability. Finally, we demonstrate our community detection algorithm with three real networks, and the experimental results show that our cooperative semi-supervised method has a higher accuracy and more stable results compared with other random community detection algorithms. © Springer International Publishing Switzerland 2013.