Overlapping Community Discovery Algorithm Based On Interaction

In the past few years, through the study of complex networks, researchers introduce the concept of community, and found that almost all real-world complex networks have community structure. In order to find the community structure in complex networks, scholars proposed many algorithms for community detection. Community detection algorithm has important applications in different disciplines and fields, such as biology, Internet, sociology, computer science and so on. In community detection areas, the aim of overlapping community detection is to discover the overlap between the communities, the overlap, due to its characteristics, is of special significance and value in a complex network. For example, it can be inferred from the overlap that whether the community have a similar connection, for dynamic community, we can predict the future changes in communities based on the overlap parts, or make the current communities reverted to previous form, if the remove overlap, then the communities is completely independent.This thesis presents an overlap community detection algorithm based on interactive degree. The main research contents and the innovation of this paper are as follows:(1) Based on the theory of interactive degree, propose a new definition of overlap. This new definition makes interpretation of overlap both more realistic and more reasonable. This algorithm regards the node with the max degree as the initial node. And this algorithm divides the community agglomerated nodes measured by the interactive degree. The final results of overlap are more reasonable.(2) Many overlap community algorithms only apply to the no weighted network, this algorithm can be used to weighted networks. Undirected weighted or directed weighted networks using this algorithm can get more desirable results.(3)This paper uses dolphins networks, American college football networks and rhesus monkey network as experimental data. With comparing the results of other algorithm, we will know that this algorithm is valid and reasonable.