Research On Dynamic Networks Community Structure Discovery Methods And Its Applications

Complex networks community discovery is crucial for understanding the structure and function of the network. Most of the real world networks are evolving over time, which requires the community discovery algorithm for dynamic networks. In this paper, we study the dynamic networks community structure discovery methods and its applications. This paper proposes an algorithm for detecting evolutionary community structure in dynamic weighted networks. Our approach, which considers the historic community structure, does not need to give the number of communities in advance, and it can automatically discover the evolutionary community structure in weighted networks whose number of nodes and communities is changing in different timesteps. For identifying community evolution paths, we propose a two-stage approach for tracking community evolution. Finally, the evolutionary community structure discovery algorithm and the community evolution paths method, which are proposed in this paper, are applied in scientific collaboration networks in project985universities. The results show that both methods have practical value. The main work is as follows:(1) This paper puts forward an algorithm for dynamic weighted networks whose number of nodes, weight of edges and number of communities is changing in different timesteps. This method, which considers the historic community structure, can find the community structure which is temporal smoothness. The main strategies are to build an evolutionary matrix as the input matrix which consider the previous community structure; then find an initial community from a node with maximal node strength; expand the community by adding nodes that can improve the quality of the community; lastly, finding the communities whose number id smaller than the threshold, and merging the communities which can improve the total quality. In addition, synthetic data and real world data are used to test the algorithm to demonstrate the feasibility and effectiveness of this method.(2) This paper puts forward a method for analyzing the network community evolution。 This paper puts forward an improved match matrix. According to the community members between adjacent timesteps, we calculate the matching metrics then analysis the community predecessor and successor. Lastly, analysis of the main phenomena occurring during the lifetime of a community:birth, disappear, surviving, merging and splitting. In addition, synthetic data and real world data are used to test the algorithm to demonstrate the feasibility and effectiveness of this method.(3) The community structure discovery method and the community evolution analysis method are applied to scientific collaboration networks in project985universities. Firstly, using community structure discovery algorithm find the evolution community structure of the scientific collaboration networks. Then, using the community evolution analysis method find the evolution path of the scientific collaboration networks. The experimental results show that the proposed methods can find community structure and community evolution paths effectively. It has a practical value for real data analysis.