Research And Application On Local Community Detection Algorithm For Bipartite Network

As a burgeoning field of study,network science has received more and more attention in recent years.Because of the important meaning of community structure in complex network,it has been a key direction of the research.Along with the research,many community detection methods for different problems have been proposed,and the discovery of local community is one of them.Local community detection is to solve the problem of obtaining the whole community which the focal node belongs to,based only on local information in large-scale,dynamic network.The research on this problem has a wide range of application scenarios.As a kind of typical complex network,bipartite network already had many specialized research on it,including its global community detection algorithm.But the problem of local community detection also exists for bipartite network,which has not yet been systematical researched.This is the research topic of this thesis.From the background and significance of the subject,this paper introduces the research status of elated fields,describes the definition and characteristics of bipartite networks,along with a variety of classic community detection algorithms for one-mode network,and typical community detection algorithms for bipartite network,etc.Based on these,three key issues of local community detection in bipartite network are analyzed,presents a similarity-based bipartite local community detection(SBLCD)algorithm,solves the problem of local community detection mainly based on one certain class of nodes in bipartite network.Then,in order to solve the problem of dividing two classes of nodes equivalently in bipartite network,a centrality-based bipartite local community detection(CBLCD)algorithm is presented,which can guarantee the consistency of local community detection result from any kind of node.In the experimental stage,a series of experiments are designed by using actual bipartite network datasets.Starting from the two classes of nodes in different order,produce a comparison of local community detection results,which verifies the stability of algorithm.The global partitioning results are obtained by the iteration of algorithm.Through those results,the modularity of bipartite network can be used to measure the accuracy of algorithm.The experimental results show that two algorithms both have good outcomes and achieved the expected target.