Research On Improved Node Coverage Methon In Community Detection

Complex networks exist in every corner of the world.It is of great application prospect to further study the structural characteristics of complex networks.Community structure is one of the important characteristics of complex networks.In recent years,community detection in complex networks has attracted wide attention from scholars in various fields and has quickly become one of the most popular research directions.This thesis addresses the shortcomings of community detection algorithm based on node coverage(NCA),the main focuses of this thesis are as follows:(1)Firstly,the background and significance of complex networks were introduced,the importance,of studying complex networks was explained.Then the research status of complex networks and theory related to complex netwrks were introduced.Some classical community detection algorithms were analyzed about their advantages and disadvantages.(2)The algorithm of community detection based on hierarchical coverage was studied in this thesis.For the NCA algorithm that needed to know the global topology of the network in advance,this algorithm adopted the method of randomly selecting initial nodes,which reduced the complexity of the algorithm.For the similarity problem of many nodes that were not connected in the complex network.The RA algorithm was adopted in this algorithm.RA algorithm is an algorithm based on resource allocation.The algorithm can more accurately calculate the similarity between nodes and predict the hidden rules between nodes.(3)For the problem of abnormal nodes would be encounter,ed in the NCA algorithm,the judgment condition was increased in this algorithm.By comparing the topology information between the nodes could be used to determine whether the node could increase the closeness of the community.In the process of detection,the abnormal nodes would be identified and accurately judged.Experiment results indicated that,Compared to classic algorithms and NCA algorithms,The algorithm of community detection based on hierarchical coverage did not need to know the clustering core in the network in advance,and the algorithm could be applied to large-scale networks.In addition,due to the addition of the RA algorithm and the judgment conditions,the accuracy of community classification was improved and the community structure was more compact.