A method for local community detection by finding maximal-degree nodes

Since obtaining complete information from large network is unrealistic nowadays, there is a growing emphasis on local community detection. However, some existing approaches are sensitive to the starting node´s position, such as the communities discovered from nodes in boundary always have lower recall rate than those from nodes in the core. Thus, in this paper, we propose a new method to detect the local community for a given node. To start, we find the local maximal-degree nodes which associate with the given node, then find the enclosing communities by calculating the local modularity of community from the local maximal-degree node, finally we optimize the communities´ structure and get the local community for the given node. Experiment results show that our method is quite effective and flexible, especially when the given node is in the community´s boundary.