Research On The Complex Network Community Detection Algorithm Base On Local Expansion

Community detection is one of the most important techniques for the analysis of complex networks.It plays a very important role in controlling complex networks,understanding network functions and predicting individual behavior in the network.With the continuous expansion of network scale,some classic community detection algorithms are difficult to adapt to large-scale complex networks due to high time complexity and difficulty to grasp the global information of network.Therefore,a community detection method based on the local information of network is proposed.The local community algorithm to detect community through local information of network,and is more suitable for large scale of complex network than other methods.This thesis studies the problems related to local community detection algorithms in community detection.Through analysis,there are some problems in local community detection algorithm: It is found that the initial position of the seed has an important influence on the final result of the local community detection algorithm;Most local community detection algorithms require repeated judgments on nodes in the network during community expansion and slow in the process of community expansion and are difficult to adapt to large-scale networks.The main work of the thesis is as follows:1.For the problem that the initial position of the seed has an important influence on the final result of the local community detection algorithm,the thesis proposes a node centrality based local community detection algorithm(NCLCD).The algorithm selects the node with high locality as the seed node to join the set,and continuously optimizes the seed node set during the community expansion process to ensure the centrality of the initial seed node.The experimental results show that the algorithm has certain advantages for the community detection in complex networks.2.In order to solve the problem that most local community detection algorithms are slow in the process of community expansion.The thesis proposes a graph traversal based local community discovery algorithm(GTLCD).The algorithm uses the node with the lowest degree as a starting point to form the initial community detection by influence function and threshold.Then use fitness function to get the final cover.The experimental results show that the algorithm can effectively dig out community structure in the network and have faster speed.