Based Triangle Ring Hub Algorithm Extension

In the real world, social network analysis has become a hot research topic in data mining such as friends network, researchers co-authored relationship network, power network widely exists in the form of multiple areas. Community Detection as an important content of social network analysis,of course, also attract a lot of experts and scholars from multiple areas pay close attention to it and had created a lot of research results.Social network tend to have high aggregation than other network, the aggregation as a measurement of intensity are mentioned in most research as an important attribute of social networks. Besides of aggregation, the center of the community reflects what the positions of individuals or orgnizations has in the network. This ideal also using in algorithms of community detection, for example.Hub algorithms, although researchers have proposed several methods of community discovery Under the guidance of the center of thought in network,but these methods alse exsit some flaws in the selection of the central node and node determing which community the node belongs to.In this paper,the algorithm extende Hub algorithm, which was used in the global community detection,and made improvements in the selection of central node and determing which community the node belongs to when judge the node,by the following:(1)Using degree of triangular loop of node in the network to measure the node of the center of the community when looking for a center node based on the ideal of Hub algorithms.(2)In order to consider the network clustering fully in the process of community discovery, this paper redefines the similarity coefficient, and it determine which community the node belongs to as well as the condition of algorithm stops.(3)In order to make the results of community Detection accurately, the paper puts forward the concept of overlap coefficient, if the community need adjust, we will according to the overlap coefficient. So the article algorithm including two steps.they are community Detection and community adjust.The experimental results show that the community discovery algorithm based on breadth-first search triangular loop guarantee the feasibility and at the same time in the quality of community can be divided the network into good results when take Zachary club network and Dolphins network, etc as experimental data.