Research And Implementation Of Big Graph Visualization For The Process Of Iterative Computing

With the rise of Facebook,Twitter and other sites,the size of social networks becomes complex and huge increasingly.In general,how to dig out the pattern and structure with potential value from these complex and huge network information is one of the challenging hot topics in the area of data mining.To handle such a large graph data,some large graph processing platform arises at the historic moment,for example,Pregel,Giraph,Graphlab etc.These systems are BSP model,and BSP programs have many supersteps.But in the process of iterative calculation,the structure of the graph and the change of the vertex aboult value and label is unvisible to the user.In order to make users have a better understanding of a larger graph with process of iterative calculation,aid the users in the analysis of large data,dig out the hidden information in big data,so this thesis puts forward the visualization for a larger graph iterative calculation process.In addition,this thesis divides graph data into static data and dynamic data,in order to implement the visualization of iterative calculation process.First of all,we need to implement the visualization of static data.However,the current graph visualization system can display limited scale of graph and often with the problems of long running time,poor visualization and low space utilization.So we design and implement a large graph visualization system and visualization model for the iterative calculation of big graph,the main contribution of this thesis is as follows:(1)Designing and implementing graph visualization for big graph based on layer.Firstly,by analyzing the traditional label propagation algorithm,we design a fast and efficient large graph clustering algorithm,and then based on the size of the combined cluster,we propose a small cluster combination strategy;taking the two algorithms as a preprocessing basis of big graph visualization system,and finally implementing the big graph visualization system based on layer.(2)Based on the analysis of the characteristics of traditional force directed layout algorithms and the dispaly of visualization system based on layer,this thesis proposes the force directed layout algorithm based on the weight of edges.And using Bary-Centralizing algorithm to optimize the initial layout,and then speeding up the layout convergence.Finally,this thesis proposes the evaluation criteria for the layout algorithm.(3)Based on large graph visualization system,researching and implementating the visualization model for iterative process of big graph that can show every step or every few iterative calculation process according to the demand of users.Users can have an intuitive understanding for the process of algorithm iteration.Experiments show that this thesis can effectively visualize the information of graph data,and the layout algorithm of this thesis shows efficiency in the large scale social network.Finally,this thesis implements the big graph visualization.