Research On Network Community Division Algorithm Based On Spectral Clustering

Complex network is made up of some relatively independent and overlapping community structure.Community structure of the nature of the research can help us to understand and analyze the network.The first and most basic problems in the research of community structure is a community detection.In recent years,the community structure detection have become the focus in the network analysis.Therefore,scholars have put forward a lot about community structure detection algorithm.In these algorithms,spectral classification algorithm will eventually problem to matrix eigenvalue and eigenvector,reduce the complexity of algorithm and obtained satisfactory effect.Based on this,this paper presents a network community structure discovery algorithm based on node clustering coefficient and a spectral clustering community detection algorithm based on bridgeness.In this paper,the main work is as follows:(1)A network community structure discovery algorithm based on node clustering coefficient is proposed.The algorithm studies the relationship between node clustering coefficient and the network community structure.First of all,based on the given CCE rules which leads to the similarity matrix,namely the network density matrix.Then,the Laplace matrix is constructed by using the network density matrix,and a network community structure discovery algorithm based on node clustering coefficient is derived.Finally,the experiment is carried out in real network data sets,results show that the algorithm in obtaining satisfactory results at the same time also can improve the time efficiency.(2)A spectral clustering community detection algorithm based on bridgeness is proposed.The algorithm studies the relationship between the bridgeness and the network community structure.First of all,the coefficient matrix is derived from the maximum group bridgeness.Then,the Laplace matrix is constructed by using the coefficient matrix,and a spectral clustering community detection algorithm based on bridgeness is derived(BRI algorithm).Finally,the experiments are carried out in real network data sets and artificial network data sets,It is verified that the bridgeness introduced into the community detection algorithm can effectively divide the network into meaningful community structures,and the partition result is close to the real network partition.