Research On Hierarchical Clustering Community Detection Algorithm Based On The Level Of The Betwweenness

Many systems in the real world can be represented by the Internet mode, the Internet network is made up of nodes and edges sets, in the study of network system, nodes stand for the individual, the side stand for the relationship between the individuals, composed of groups called a community. Detect the community structure is a basic task of social network analysis, it is helpful to the realization of the other social computing tasks, and can be applied to many practical problems to solve.Detect the community structure is in some measure object similarity classification of the data set. We found that:the condensed clustering algorithm type when measuring similarity does not take into account the actual background of uncontrollability, lead to inaccurate division results; Split clustering algorithms type often use global information parameters, computational complexity is higher, is not suitable for large-scale network. In view of this, in this paper, by adopting the idea of hierarchical clustering based on node betweenness and edge betweenness, the community detection algorithms are designed, and the main work is as follows:(1) Aiming at the signal missing of problems in the process of signal transmission, put forward the signal missing in the transmission of condensing clustering type community detection algorithm (SMHC). First, the algorithm think node as a source of community structure, considering the node betweenness role in the network, using the degrees centricity as probability of received signal, to quantify the signal missing value of the node in the process of receiving signals. After the process of signal transmission, the nodes of the network topology structure is transformed into the influence of the node vector matrix, then using the modularity of condensed clustering type method for community detection. Finally, on the artificial data set and the benchmark data sets, through comparing with the existing community discovery algorithm, the experimental results show that SMHC algorithm improves the accuracy of community detection.(2) The traditional split clustering type algorithms use the global information parameters, such as edge betweenness, how to speed up the split and split in when to stop are the key problems, so that the split clustering type algorithm (BQI) was proposed based on the bridge combined with the modularity incremental. The algorithm uses local information parameters "bridge" as the standard of remove edge, bridge has an important role in the connection of different communities, by removing bridge of the highest value to split the network. At the same time in order to reduce redundant steps, modularity degrees incremental method is applied to decision network split in when to stop. Finally, validated on benchmark data sets, and compared with classical splitting algorithm, which is proved that BQI community divided equally accurate algorithms with the GN algorithm, but BQI algorithm improves the execution speed.