Research On Algorithm Based On Bipartite Networks Community Division

The rapid development of network science has brought us a lot of new ways of thinking,ignited our enthusiasm for intensive study,and shorten the distance between people.This article will help the growth of network science by exploring the bipartite network.Research community structure plays an important role in understanding the structure and function of the entire bipartite network.At present,many researchers have proposed different algorithms for the bipartite network community division.This paper proposes two different algorithms from a new perspective.The results are as follows:(1)An Intimacy and Attraction Algorithm(IAA)is proposed,which aims to improve the accuracy of community partition.First selecting each node in the U-type to be divided into different communities,a creative formula for intimacy and attraction is proposed,and merging communities,then dividing the V-type nodes into existing communities and calculating the modularity Q.Until Q is no longer increased or the merging conditions are not satisfied.In this way,a complete community structure is created.The mutual information and modularity were used for experimental analysis on the artificial bipartite network dataset generated by computer simulation and the real bipartite network dataset.It is concluded from the experimental results that the IAA algorithm does not input any parameters and obtains higher accuracy than other algorithms.(2)A Similarity Clustering Algorithm(SCA)is proposed.The algorithm first selects all the nodes in a certain type,calculates the similarity between the nodes to obtain the core node set,and sequentially expands the nodes of the core node and its neighbors to the community,thus obtaining the type.Finally,another type of node is clustered into existing communities to get the final result.Through experimental analysis,SCA can get good community classification results.Then analyzed at different values of parameter data sets.