Improvement Of Overlapping Community Detection Based On Label Propagation Algorithm

Complex network structure exists everywhere in real life,with the development of information technology,more and more attention has been paid to complex network,Community detection can not only help people understand the function and feather of complex network in detail,but also help people know the social problem of social network well,so community detection is becoming more and more important.In real network,overlapping community detection is common,at the same time,researchers have proposed a lot of overlapping community detection algorithm,including CPM?LFM?GCE and COPRA,comparing with other overlapping community detection algorithm,label propagation is simple.This paper proposed a new overlapping community detection algorithm on the basis of the original COPRA algorithm.Our improvement includes degree threshold 0d,CPM maximal clique idea,label comprehensive influence and we also use asynchronous label propagation.Through these improvements we mean to solve the problem of algorithm randomness and improve the results accuracy.Finally,through a lot of experiments,we studied the parameters of the mixing coefficient parameter ? in LFR benchmark data set,COPRA algorithm parameter v and improved COPRA algorithm 0d on the three algorithm(CPM,COPRA,and improved COPRA)about time complexity and the efficiency of the effect.It shows that no matter the network is sparse or dense,this algorithm can find high quality community structure.Finally,in the real network dataset,we achieve the improved COPRA algorithm,the experimental results show that the algorithm has higher modularity,and it can reflect the community relationship better,it can also discover the overlapping community structure effectively.