The Research And Implementation Of Key Technologies In Community Structure Evaluation System

With the rapid development of the Internet, the research on social network has gradually become a hot topic. The number of nodes in the social network is large and the topological structure is complex. In recent years, with the in-deep research of social network, people gradually found a unique network structure--community. At present, there is no formal and precise definition of community structure. But the structural characteristics of community can be summarized as: the nodes of the same community are closely linked with each other and the nodes of different community are sparsely linked to each other.Community structure can better describe the characteristics of social network.Therefore, it is necessary and valuable for further researches in community. In recent years, the research of community rises gradually. A series of algorithms have been proposed. And the application scope of community becomes wider and wider. In a word, community can be widely found in all aspects of our daily life.The main research contents in the field of community in this paper are as follows:The research of dynamic benchmark graph model. There are lots of complex communities in real world. At present, real-world data almost has some limitations such as the number of datasets is less and the size of datasets is small. This paper proposed a novel benchmark graph model based on LFR model. The proposed model expends the LFR model into dynamic networks.The research of dynamic community detection algorithm. Nowadays,lots of algorithms have been proposed based on different ideas.Modularity has been used widely for its unique characteristics.Unfortunately, it has a well-known problem called 'resolution limit'. This paper proposed a concept of disturbing factor and a novel dynamic community detection algorithm based on distance dynamics. The proposed algorithm has well balance between efficiency and effectiveness both in synthetic and real world networks.The research of metric in community evaluation. Different metrics of community reflect the community structure characteristics in different aspects. This paper proposed a novel metric based on network structure entropy. The proposed metric has well balance in reflect community structure.Framework of community structure evolution system. The above three contents support the evolution system in the aspect of algorithm.The results can be used as the theoretical basis of the evaluation system.The purpose of this study is to combine these three research points in the view of engineering. This paper combines the community and parallel computing framework. In the current community research field, the research on large-scale network community is less with the parallel computing framework. This study simplifies the complexity of community structure research in large-scale datasets by providing a simple and scalable programming model.