A Novel Approach for Discovering Local Community Structure in Networks

The algorithms for discovering global community structure require the knowledge about entire network structures, which are still difficult and unrealistic to obtain from nowadays extremely large network. Several local algorithms that use local knowledge of networks to find the community for a given source node were proposed. However, these algorithms either require predefined thresholds which are hard to set manually or have lower precision rate. In this paper, we propose a novel method to discover local community for a given node. Firstly, we find the most similar node which is adjacent to the given node, and form the initial local community D together with the given node. Then, we calculate the connection degree of nodes belonging to D’s neighbors, and add the node whose connection degree is maximum to D if the local modularity measure will be increased. We evaluate our proposed method on well-known synthetic and real-world networks whose community structures are already given. The results of the experiment demonstrate that our algorithm is highly effective at discovering local community structure.