Research And Implementation Of A Community Petection Algorithm Based On Closed Walks

Research on complex networks can help us understand the structure of network.Furthermore, we can understand the behavior and function of networks. Right now,the theory and methods of complex networks which play an important role ininterdisciplinary, are widely used in many subjects and change our daily life.In this paper, we firstly interpret some basic concepts and theories of complexnetworks, aftermath several classic community detection algorithms are introduced.After we analyze problems of the three classic algorithms, a concept called “closedwalks” is introduced. By counting the number of closed walks in complex networks,we try to analyze complex network from a new point of view and solve the problemthat current algorithms cannot balance the time complexity and precision. This papermainly proposes a new measure that integrates the concept of closed walks andclustering coefficients to replace the edge betweenness, and gives a hierarchicalclustering method that removes edges iteratively with the lowest value forcommunity detection. We tested our method on computer generated networks andreal-world networks. The results showed that our method is a better tradeoff oflow-accurate and time-consuming measures.According to experimental results and our analysis, closed walks of order threeand nontrivial closed walks of order four can be considered as basic elements inconstructing community structure. Trivial closed walks are not as important asnontrivial closed walks in analyzing the structure of networks. Lastly, this papermainly mentions the platform of experiment, the design of key algorithm and theapplication of our algorithm.