Research Of Community Detection Based On Constant Structure In Complex Networks

All kinds of relationship in the society could be abstracted into network topologies,so in complex networks,research on the network topology has been a hot topic.The main direction about it involves to detect community and to learn some key attributes in the network.Recently,researchers have proposed many different methods: About research direction consist of the static and dynamic community research;About the research of community discovery algorithms divide global perspective and local perspective;About community discovery evaluation mechanism,like calculating modularity and community mutual information are generally,then another new research is community stability,as we all know the goodness of community finding result is related to the quality of community discovery method directly.In this paper,the community discovery algorithms contain two sides: the global and local.And the content is about how to improve its stability,and how to evaluate the quality of the community.The main research include the following aspects:(1)How to use the local expansion method to detect the stable community structure in complex networks.Local optimization does not need to know the size of the network or the relevant some prior knowledge,especially the problem of calculating large complex networks.Firstly,we can take advantage of it adequately.Secondly,the stability of a network has a great influence on the community structure.Some research find that different local optimization algorithms lead to different results,but analysis the same part to find some permanent nodes.In addition,defines the stability of nodes in the network,introduce stability calculation method.Then proposes a local community detection algorithm to find the stable community.Finally,this algorithm has proved in the LFR benchmark networks and real-word networks,which effectively find the stable community and have a high accuracy results.(2)How to use the global method to detect the stable community structure in complex networks.Global optimization method provides an earliest way to do a community detection,including graph partition,hierarchical clustering et al.Because the number of nodes and connected edges are so large,also part of them are sparse which could not play a role in the network,so we should do some preprocess on the network.We can detect some core stable communities of the network through random perturbations.And seem core communities as super nodes,combining with the remaining nodes to form a new network,and then using maximize modularity algorithm,find a high modularity and high stability of the community.At last,the algorithm is tested on the real network and the integrated network.The result indicates that the stability and the modularity of the network are higher after preprocess the original network,and the robustness of the network is also improved.In summary,this paper applied two ideas of global optimization and local optimization to study the stable community structure of complex networks.By comparing with other algorithms,we could prove the effectiveness of algorithm and the goodness of the community structure.Besides,the community discovery algorithm provides reference for further research.