Community Detection Research In Complex Networks Based On Dynamic Behaviors

In the network study, the nature of the network can be shown by network topology, the topology of the network describe individual connection relationship and affect the network function of dynamic process. A quantitative description of the topological structure properties, therefore, to analyze and reveal the relationship between network structure and dynamic capabilities. Community is an important attribute of the network structure, to reveal the network of community structure has important theoretical significance to understand its function, to predict its behavior, we put forward the rationality of the model.This thesis content are as follows:(1) The problem of signed networks community detection, The definition of similarity is applied to the directed weighted network through the clustering model. Normalization is adopted to further process the similarity and the low similarity of node is refactored. signed network model is established, according to the node status of dynamics behavior testing community. Our method are applied to synthetic network and real network to verify the feasibility and through the analogy with other algorithm, it shows the advantages of our algorithm.(2) Based on the dynamics of complex network community detection, using Kuramoto oscillation model can often reveal the community structure, different from community detection and network dynamic clustering algorithm.This paper improve Kuramoto coupling model through joining the similarity to detect network community. By adjusting the parameters of the node,it evolved which network model based on the dynamic mechanism, the network into different communities and through the real network detect the effectiveness of algorithm.