Community Detection Method Based On Local Information

With the proposed of the small world network model and the scale-free network model,the study on complex networks is achieving a climax at home and abroad now.Many real-world systems in nature and society can be seen as complex networks.For example,traffic networks,large power grids,social networks and so on.The community structure is an important feature of many network models,which plays an important role in understanding the internal information of the community and solving practical problems.This paper focuses on the community detection method based on local information.The detection method based on local similarity and local centrality and the method based on interaction quality optimization are proposed in this paper.The community detection algorithm based on local similarity and local centrality is a kind of aggregation algorithm.There are two main steps in the algorithm.First,the local similarity is used to determine whether each pair of neighbors is located in the same community,and then each node is connected to the neighbor who satisfied the local centrality.This novel local centrality definition reduces the white noise of the center standard defined only by node’s degree.Experiments on a variety of networks and results of comparing with the classic of community detection methods proved the effectiveness of the proposed method.The community detection algorithm based on interaction quality optimization has two main local indexes,interaction cohesion and interaction density.These two measurements respectively represent the cohesiveness of the community to the neighboring nodes and the density of connection within the community.The joining of interaction density makes the interactive optimization algorithm can adapt to the network with different topological structures.The process of community detection adopts the idea of "Adjacency Frame".The performance testing of the proposed algorithm in classic generating networks and real networks improved that it can detect the overlapping communities efficiently.