Research On Community Detection Algorithm Based On Clustering Coefficient

There are many kinds of complex networks,covering all aspects of human society,and involving various disciplines.The research work of complex networks has attracted the attention of researchers in related fields,and has been paid more and more attention.Especially since the 21 st century,Internet technology has made unprecedented leap forward development,which has a great role in promoting the research of complex networks.In recent years,experts and scholars from related fields have put forward many complex network community detection algorithms,among which Label Propagation Algorithm is one of the most classic ones.Label Propagation Algorithm has the advantages of high efficiency and fast,without prior information,but also has the disadvantages of insufficient stability,easy to form huge communities and so on.In this thesis,the existing complex network community detection algorithms are studied in detail,especially the specific steps,advantages and disadvantages of the Label Propagation Algorithm are carefully studied,and the clustering coefficient property of the network is analyzed in depth.The concept of community clustering coefficient is extended from the clustering coefficient of nodes and the clustering coefficient of the network.Due to the problems of unstable results and easy to form huge community in the tag propagation algorithm,combining the concepts of point clustering coefficient and community clustering coefficient,this thesis proposes the improved label propagation algorithm based on node clustering coefficient and community clustering coefficient respectively.Among them,the improved label propagation algorithm based on the node clustering coefficient first prioritizes the nodes according to the clustering coefficient and degree of the nodes.In the initial label stage,it improves the strategy,only initializes the nodes with higher priority.In the label propagation process,it sorts the neighboring nodes according to the clustering coefficient of the nodes,selects the optimal label,avoiding the traditional label propagation algorithm's high randomness.Based on the community clustering coefficient,the improved algorithm uses the community clustering coefficient to improve the label propagation strategy,which also improves the stability and detection effect of the algorithm.Finally,both the two improved algorithms are applied to real network data set and large artificial network data set.The experimental results show that the two improved algorithms have stable detection results and high accuracy compared with the real community structure of the network.