A Large Scale Social Network Community Detection Method Based on Probability Model

In this paper, we study on the problem of large scale social network community detection method, which is an important issue in Online Social Networks data mining. The proposed social network community detection system aims to divide the social network into several communities according to internal relationships among users. Particularly, this system is made up of two steps, that is, 1) detecting the community structure in a specific time period, and 2) discovering the evolution of the social network community. The main idea of this paper is to exploit LDA topic model to mine and analysis the co-occurrence relationship between different words, and then construct word relationship network. Afterwards, LDA model is used to extract community structures from the word relationship network. Finally, experimental results demonstrate the effectiveness of our method.