Method Of Dynamic Pattern Mining Based On Complex Network

With the vigorous development of computer and Internet technology, great attention is given to dynamic network, and dynamic pattern not only can reflect structure features and functional information of complex system, but also reflect its evolution and behavioral characteristics in dynamic network. But previous research on complex network has focused on static network that does not change over time, such as snapshot in a certain moment or integrated network obtained over a period of time. Hence these researches ignore dynamic of complex network. But almost all complex networks have dynamic characteristics, so dynamic model research has important theoretical significance and application value. Therefore, in this paper we make analysis and research on regularity behavior of complex dynamic network.Firstly, method of frequent periodic dynamic pattern and frequent jump pattern mining based on complex network is proposed. PDP and FJP have special properties in dynamic pattern. They not only represents frequent pattern but also can show regular behavior in evolving social network. In order to reduce the computing complexity of algorithm, we construct matrix of directed graph to present the evolution of complex network over time. Moreover, a regular edge pattern searching algorithm based on MDG is designed, it is used to select frequent and regular edge existence sequence(EES) in MDG with progressive scan. Then, depth-first search method is used to mine PDP and FJP.Secondly, we propose early clustering and evolutionart clustering algorithm based on improved k-means algorithm. It can solve the limitation of traditional k-means clustering method and the dependence on the initial value, to get better clustering effect. We select clustering opration for next moment snapshots by getting initial cluster centers through the selection of seed node and according to changes of complex dynamic network. It achieves smoothness of clustering results, and the evolving graph of network is obtained.Again, we propose critical path mining method for complex software network with complex network characteristecs. The previously mentioned dynamic pattern minging method to software network for mining critical execution paths with law behavior characteristics. We build SN-DW model to describe the relationship between elements of software systems, and found networked software systems rules call relations, and better study its internal law.Finally, in this article, the proposed algorithms for analysis and validation, we applied the algorithm with real social network and synthetic data.