Research On Important Nodes Mining And Evolution Models Analyzing In Complex Networks

Complex networks analyzing becomes a research direction with the rapid development of network theory and computer technology. In the real world, many fields can be abstracted as complex networks. Deep studying the theory of complex networks model and the process of evolution, we can explain the common law of complex systems that hide in the nature world, society world and organic sphere. And in the process of complex networks research, the discovery of important nodes and its evaluation method which have very important significance to improve the robustness and reliability of complex networks. Due to the diversity of complex network structure, the study of important nodes evolution and evolution model established is particularly important. Therefore, how reasonable reproduce the evolution process of important nodes, establishing the better practical complex networks evolution model that make it more in line with the network characteristics of the real world, which becomes the core issue of the complex networks research.Therefore, important nodes mining has important theoretical significance to improve anti-destroying ability of network.This dissertation introduces important nodes mining, important nodes evolution and evolution model analyzing in the complex networks, The work is as follows:(1) Some common Important nodes mining indexes are analyzed, such as degree centrality, betweenness centrality, Katz centrality, and so on. Studies have shown that these single indexes have advantages and disadvantages; Also some comprehensive evaluation methods are studied, these methods combine the advantages of single indexes and make up for the defects of them. With the study of complex networks evolution model, analyzing the evolution of BA model and the connection mechanism of BA model, discussing its existing problems, also some improved evolution model based on BA evolution model are studied.(2) Based on the existed importance nodes mining methods, this dissertation proposes a new node important evaluation algorithm (the BKC), this method combines betweenness centrality and Katz centrality, combining global characteristic of betweenness centrality and local characteristic of Katz centrality, effectively avoid the betweenness centrality which is given priority to the shortest path alone. The feasibility of the algorithm is verified in small networks and public data sets, and compared with other four important nodes evaluation methods on the public data sets, finally the proposed algorithm is verified effective significantly superior to other algorithms on the important nodes mining.(3) Micro structure researches on complex networks based largely on their own property to the influence of the network evolution, but less researches on opinion leaders and the structure holes. During the process of network dynamic evolution, analyzing the change rule of the important nodes and the structure holes in the network. Finally according to the problems existed in the BA model, this dissertation puts forward an improved network evolution model, the model is a evolution model based on BKC algorithm, it adopts the BKC algorithm to replace with degree in BA model as node priority connection mechanism. The computer simulation results show that, the improved evolution model not only conforms to the power-law distribution form, but is better than the BA evolution model on the average path length and aggregation coefficient, therefore the improved network evolution model proved conforms to the characteristics of the real world, and more rationality.