The Research Of Overlapping Community Structure In Complex Networks

The research starts with the background and meaning of the complex networks, introducing the current research situation and recommending the community structure of comlex networks which is a research focus through introducing the basic theoretical knowledge. It is challenging but promising to explore the community structure which has theoretical significance and practical value to understand the structure and functional properties of the network.At present, some achievements has been made about community structure of the network, and proposed many community discovery algorithms. However, the majority algorithms belong to the rigid division of the network. Supposed that the discovered community is non-overlapping with each other, a node can only belong to one community. This is very effective for the network which has a clear structure. However, community structure is not so obvious in the actual network. In the network composed by a large number of nodes whose connection is confusing, the transition between them is excessively smooth, and has the overlap community structure.In this paper, the research results about the overlapping community structure of the complex networks are reviewed in detail. Some latest representative algorithms are introduced, such as clique percolation algorithm, the optimized algorithm based on overlapping modularity, the algorithms which can detect overlapping and hierarchical community structure in the network. All of these can provide help and reference for the future development.Finally, this paper makes an empirical analysis on an enterprise network, an artificial network, and BBS network for users' interest. The results show that:the overlapping community structure exists in these networks to some extent. Depending on the analysis to the enterprise network, we discover that increasing the properties of overlapping appropriately and intentionally may promote the communication among the employees, and is helpful for the development of enterprises healthily. By analyzing the man-made network, the results shows that:the network has a very good properties of overlapping and hierarchical, which gave us a kind of enlightenment that we can get more potential information effectively by dividing the network which has a good property of hierarchical layer by layer. The empirical research of BBS network for users' interest shows that this network is a scale-free network whose distribution of degree and community size can be described by power-law distribution. The community size is extremely uneven, so there are some large-scale communities which are called maximal community. In addition, there is some overlapping department between communities.