Research On Community Detection Algorithm In Complex Network

With the progress of science and technology,many complex networks appear in the real world.The quantitative data analysis and data mining about complex network help people understand the hidden information and rules behind the huge data.Community structure is a very important feature of the complex network,so the community detection has become a hotspot for many scholars at home and abroad.The Greed optimization method based on the objective function is an important method in community detection field.And as the definition defects of objective function,this kind of methods have the resolution problems.In addition,the community has three structural characteristics.They are non-overlapping,hierarchy and overlapping.How to fully dig up the three structure is an important work in community detection.This paper studies the community detection problem of complex network.Based on the thought of greed optimization about an objective function,this paper carries out the following two aspects of research.First,we propose a two-phase algorithm for detecting hierarchical community structure in complex network.The greed optimization methods based on the modularity function has the problem of resolution.The optimization method based on the permanence function can solve the problem of resolution,but there are many small communities generated.The number of community detected by the method is always greater than the actual one.Thus,combing the advantages and the disadvantages of the above two methods,we propose a two-phase algorithm for detecting hierarchical community structure in complex network.Compared with the major state-of-the-art methods,the results indicate that the community structure detected by the proposed method match better with the ground truth network.Second,we propose an overlapping community detection algorithm based on the history label sequence of node.The two-phase algorithm only find the non-overlapping and hierarchical community structure and can't find the overlapping structure.Some classical overlapping community detection algorithm has the problem of poor stability.This paper propose an overlapping community detection algorithm based on the history label sequence of node.Compared with the classic overlapping community detection algorithm,the propose algorithm is stable.Due to the fewer iteration times,the proposed method saves the storage space.