Research On Algorithm Of Community Detection In Social Networks

The study of social networks is achieving a climax at home and abroad now recently.Many complex systems in the real life can be depicted as the form of networks, such ascommunication networks and social networks, which belong to complex networks,electricity networks and metabolic networks with the viewpoint of system science, and theInternet networks. Complex networks often have the same features-the comminitystructure. The nodes link densely in the same community but sparsely in the differentcommunities. On the community structure detection, an non-overlappde communitydetection algorithm based on the link of node and community is proposed, and animproved community on Peacock detection algorithm is also studied.Firstly, the basic characteristic of the complex network structure was deeply nalysised.The summary on the classificiation of the existing communtiy detection algorithms weremade and make some analysis on several classical algorithms,and finally make fullunderastand of them.Secondly, an algorithm on the cloness-coefficience of the node and community wasproposed. When the network initialization finished, the node should be adjusted throughthe cloness-coefficience to meet the comminity structure features.Thirdly, based on the Peacock algorithm,we use its nodes replication, extending thenetwork by copy some nodes. To divided the community,we using the introduction of thecandidate maximum module method. Newman’s module degree function is used todetermine whether need to continue separating. The division of the nodes map back to theoriginal network. If the replicated nodes belong to different communities, thecorresponding to the original is the overlapping nodes in the network.Finally, the non-overlapped community detection was researched on the realisticnetworks, and the results of the experiments show its availability. And also the overlappedcommunity detecting algorithm was researched on the one-dimensional network and thetwo-dimensional network.