Local Detection of Communities by Neural-Network Dynamics

Community structure is a hallmark of a variety of real-world networks. Here we propose a local method for detecting communities in networks. The method is described as 'local' because it is intended to find the community to which a given source node belongs without knowing all the communities in the network. We have devised this method inspired by possible mechanisms for stable propagation of neuronal activities in neural networks. To demonstrate the effectiveness of our method, local detection of communities in synthetic benchmark networks and real social networks is examined. The community structure detected by our method is perfectly consistent with the correct community structure of these networks.