Research On Algorithms Of Finding Complex Network Community

A complex network is a simplified representation of a complex system in which theentities of the system are represented by the nodes in the network and the interrelationsbetween entities are represented by the links joining pairs of nodes. Recently, the detec-tion and analysis of community structures in complex networks has attracted a great dealof attention. Network clustering algorithms can be used to analyze the topological struc-tures, understand the functions, recognize the hidden patterns, and predict the behaviorsof complex networks.Since many algorithms work based on the whole network, and the shortcoming ofthe algorithms is that the time complexity is relatively high, and its relevance is relativelyweak. Here, we consider that the "local community" is more relevance for some networksresearch based on the local information of the networks. This structure is of specialsignificance for the local information given. In this paper, two algorithms are proposed:(1) Bipartite graphs are relatively simple. We represented complex network asbipartite graph, and transformed the nodes of original network to the "bottom set" ofbipartite graph. In the conditions of two given nodes, we achieve the "relational com-munity". This algorithm combines the Clauset's idea of finding local community andconcept with betweenness to find the "relational community" through the network of lo-cal information. Since this algorithm do not need to be calculated base on the entirenetwork, it demonstrates excellent detection when randomly given nodes. This algorithmis for finding the "relational community" of the network.(2) We combining degree centrality with flow betweenness centrality, a methodto find central community is proposed. Firstly, the node's degree centrality and flowbetweenness centrality will be calculated for determining the geometric center of thenetwork and the node connecting most paths while information, material or energy trans-mitted through the network. At the same time, these two indicators will be taken intowhole consideration and the node with relatively high indicators will be found out. Then,CPM community discovery algorithm will be used on this node and its neighbor-nodesto find the central community of the network. This method mentioned above can help tofind the relatively"important" community of the network, which has certain significancefor the analysis of the spreading mechanism on the complex network, successive failureand so on. Finally the central community of the public transport network of LanZhou isanalyzed by the method we proposed, and our results indicate that the central commu-nity plays a central role in the whole network. This algorithm is for finding the central community of the network.