A Visual Analytical Framework for Exploring Large-Scale Social Networks Based on Overlapping

• 1、Department of Computer Science, Beijing University of Posts and Telecommunications

Abstract：Social network analysis provides us an effective way to characterize the relationships among people in massive social networks. How¬ever, scalability is still a key issue for large-scale social network visualization. To get an overview of massive social networks, we propose a visual interactive clustering framework for exploring sta¬tistically significant overlapping communities. The framework im¬plements the Visual Analytics Mantra: “Analyze first—Show the important areas—Zoom, Filter and Analyze Further—Details on Demand”. To visualize statistically significant overlapping com¬munities in large-scale networks interactively, we study the prob¬lem of enumerating maximal cliques in real-world social networks. By filtering vertices with low degrees by the (k-1)-core decompo¬sition algorithm, we develop several maximal clique algorithms to enumerate all the k-maximal-cliques in large-scale social networks. As our algorithms are extremely fast, we can explore the structure of large-scale social networks through visual interactive clustering techniques. Our primary contribution is the design of the visual an¬alytical framework and k-maximal-clique algorithms for exploring massive social networks. To evaluate the effect of our framework in real applications, we develop a visual analytical prototype tool called CliqueVis and present empirical studies of exploring real-world call graphs which contain several millions of customers. After getting an overview of call graphs, we can zoom in the commu¬nities which we are interested in to obtain more details about their communication patterns.

Keywords： Visualization framework large-scale networks visualization ocial networks analyze Maximal Clique Detecting Algorithm