Research On Overlapping Communities Detection Algorithms Based On Central Communities

Nowadays, people are living in a society that filled with a variety of complex networks. Community structure is one of the basic topological properties of complex networks. In this community structure, Nodes in the same community are connected densely, while nodes from different communities are connected sparsely. Community detection can get clusters of network, it is divided into overlapping communities detection algorithms and disjoint communities detection. To lots of communities mining algorithms, many of which only focused on finding disjoint communities, but in real world of networks, overlap is often an important form of community structures. Complex networks, especial overlapping communities are significant in many fields such as public opinion controlling, information spreading and recommending, and commercial marketing.In depth research and analysis of many overlapping communities discovery algorithms, founding that some of these algorithms have not highly accuracy, some can not find a good overlapping structure, and some is only apply to the dense network, to the sparse network, the performance is poor. In order to overcome the above problem, an algorithm based on central communities are proposed and the algorithm not only have strongly adaptability to the structure of the network, but also have highly accuracy, even more can effective discover the overlapping part of the community.The main work of this paper including:1) From the central node of network to find the central community.2) Calculate the associating degree between nodes and communities, then put nodes that satisfied the given threshold condition into central communities constitute initial communities.3) Calculate the associating degree between initial communities, merging initial communities that reach the giving associating degree condition.4) Calculate the overlapping degree between communities, merging communities that fit the giving given threshold condition.In the given modularity and accuracy evaluation standard, its reasonability and efficiency have been verified through simulation experiments and the results show that the proposed algorithm is effective.