Mining Community Framework And Hidden Community

With the development of information technology, the social networks play a more and more important role in people’s daily life. Social networks are so popular that it not only draw attentions from sociologist, but also show attraction to other scholars in different fields. Social networks consists of the participants and the relations between them. In social networks the participants often show cluster phenomenon and the cluster usually are called communities, participants within the same community have similar attributes and the connections between them are density, the connection of participations which are from different communities and have different properties are relatively sparse. The study of the community structure has important theoretical significance and practical values and it is not only beneficial to understand the function and the structure of the entire social network, to grasp its internal laws, to predict its progress direction, but also play an important role in e-commerce, computer virus transmission control, etc.After years of development, a large number of community mining strategies has been proposed, but we found that these strategies had many shortcomings:their time complexity is too high, lack of interaction with users during mining process and they are can not effectively deal with the incomplete social networks, thus we did three aspects of job just as following.1. To solve the the problems that time complexity of Community Mining Strategy is too high, lack of interaction with the users and other issues, the node centrality, power-law degree distribution and other characteristics of social networks are discussed, "critical sub-network" and "Community framework" concepts are defined, Moreover, Mining the Community Framework algorithm (MCF) and Drilling Down the Community Framework algorithm (DCF) are designed, in which MCF algorithm are used to mine the community framework of social network, and DCF are used for drilling Down Community framework in order to show the community structure from different granularity.2. To solve the problems of the community mining algorithms which can not effectively deal with the incomplete social networks, transitivity of the social network and similarity of the nodes’attribute are discussed, the concept of compensatory matrix is designed, and a method which used to build a relatively complete social network are designed,and finally, the algorithm used to mine the hidden community (MHC) are realized.3. The effectiveness of all the algorithms we proposed are verified through experiments and analysis, MCF algorithm can be used to dig out a Community framework which can show the state of community structure of complex networks, and interacting with the user, DCF algorithm can dig out high-quality community structure. MHC algorithm be able to effectively deal with incomplete social networks and dig out their hidden communities.