Research And Application On Dynamics Of Community Structure Based On Complex Network Theory

Complex networks as a new cross science, in this decade, attracted great interest and attention from different fields. For sociology, computer science, biology, neuroscience,economics, and many other fields, with the help of the study of the theory of the complex network method for quantitative analysis and qualitative interpretation, and helps to reveal the complex network behind represents complex system of common law that has important scientific significance. In the complex network of research topic, dynamics,related with the human health and social security, becomes the research focus.The methods that combined complex network theory and traditional epidemiology knowledge have become the main trend of epidemic spreading study. But current dynamic model researches have ignored the impacts of community structure in the transmission process, as well as the complex characteristics of human contact models on the spread of epidemic, so it is difficult to recreate the process of the epidemic spreading. On the other hand, the scale of actual network are very huge, while it is impossible to get the structure at the same time, all these things have limited the effectiveness and applicability of traditional immunization strategy, and effected the order of the core spreading nodes. Therefore, this paper are based on actual networks and model to solve the above problems, and we have studied the mesoscale structure characteristics and the impact of human contact models on the epidemic and rumor, to looking for the key mechanism and explore the prediction of the epidemic spread on community network. Deep mining the core node in the spread of the epidemic on the community network, and put forward effective immune algorithm, improving the control ability, also provide the theoretical basis and scientific reference for the epidemic prevention and control on the community network.FIRST,To predict the epidemic spreading on complex networks, based on the different contact models in the small-world network and scale-free network, this paper has used the variance of the infection density as the feature to represent predictive. In addition, considering the community structure of the network, we have imported the distance of transmission source and bridge node as a random factor, analyzed the changes of the variance. Through the experimental results it is not hard to find: the mesoscale community structure has great randomness in the spread at the early stage. The reason is that this feature has a positive correlation with the distance, meaning that a long distance has a poor result, which proves that it is impossible to accurately predict the spread process.SECONDLY, As the fact of unknown structure of the community network, this paper proposed an immune algorithm that based on the detection of center bridge node within the local area. The algorithm used the self-avoided random walk process to find and immune the bridge node and its neighbors who have more weak-ties. On the one hand,we can use the overlap degree of the node’s friends and its neighbors to identify whether it is a bridge node. On the other hand, we can also use the heterogeneity of the distribution of weak connection on community network, and found that a random neighbor of a bridge node was likely to be a center bridge node. The experimental results have shown that the self-avoided random walk immune algorithm can effectively improve the immune effect of the epidemic spread on community network, and also show the robustness.NEXT, in the article,we discuss how to mining the core nodes of the community network during the transmission, and propose a sorting algorithm which combined with the internal and external location importance of the local community. The algorithm references the idea of K-Shell decomposition algorithm, and takes advantage of the degree distribution heterogeneity of weak connection in the community network. At the same time, we find the inside and outside location of nodes in the local community, and obtain the sorting result of core nodes location in the community network by calculation. The experimental results show that the algorithm using the heterogeneity characteristics of weak-ties efficiently, improves the sorting result accuracy of transmission core nodes of the local community, and has higher robustness.In the END, based on the foregoing questions about forecast, immune and sorting,for deepening and understanding the space and time influence of mesoscale structure on transmission dynamics, we use typical public opinion transmission events as the research background, using the statistical physics and other methods, analyze the influence and interaction between the transmission dynamics and the area represented by mesoscale quantitatively. Through the empirical analysis results, we find that:(i) regional public opinion transmission has heterogeneity on the distribution, however, the ZIPF distribution of velocity change is small, which reveals that the regional concern for public opinion has convergence trend during the transmission process.(ii) social and economic level has significant effect on public opinion in the transmission process.(iii) controlling public opinion targeted the area of the economic development or high education, can effectively guide the trend of transmission and enhance the positive control of public opinion.