Research On Visualization Techeniques For Massive Social Network Graph

With the rapid development of Social Network Services, there appears more and more social network data which is based on the internet. However, people's ability of analyzing and understanding the massive data is far behind the ability of collecting data. How to analyze data and display the result more effectively and how to improve people's understanding about the discovered feature of data is gaining more and more attention from researchers.Graph visualization helps users gain insight into data by turning the data elements and their internal relationships into graphs and its usefulness depends on the readability of resulting layouts. Fast and clearly display of the relation value between vertices is the critical technology of huge social network graph's high performance drawing. However, the traditional social network graph drawing algorithms can't reflect the relation value between vertices clearly, or can't perform well when meet the massive data set.In this paper, we introduce the existing massive social network graph visualization technique in details, especially the visual clutter reduction technique and layout algorithm. We also enumerate some important graph interaction techniques. Based on these, we propose a new method called Multi-level Weighted Fruchterman Reingold Layout (MWFR) which can both reflect the relation value between vertices clearly and can deal with massive social network graph, and we prove the advantage of WFR method through sufficient experiments. Our main contributions are summarized as below.First of all, we define a representation form of edge weight in relation network graph. We then propose a new layout method called WFR which can deal with weighted social network graph, based on the Fruchterman Reingold (FR) layout algorithm. We introduce the edge weight to the force-model of FR layout algorithm and turn the parameter so that it can fit to the visualization of the social network graph. Besides, we also propose two evaluation measures about the quality of the weighted graph's drawing. Through sufficient experiments, we prove that the WFR layout algorithm is better than most traditional layout algorithms both on the layout quality and the time complexity.Furthermore, we design and implement a fast massive social network graph drawing algorithm called Multi-level Weighted Fruchterman Reingold (MWFR) layout which can clearly reflect the relation value between vertices. MWFR layout algorithm applies the modified WFR layout method to the coarsening and refinement mode based multi-level massive social network graph drawing algorithm. The experiments shown that MWFR layout algorithm can reduce the running time of massive social network graph drawing remarkably while only sacrifices a little layout quality.Based on the theoretical research above, we design and implement a relation network graph visualization prototype system. The prototype system is integrated with several social network graph layout method and some generally visualization technique. Users can reveal some the hiding characteristic of data set or remove the noise data by using this visualization tool. With the help of this prototype system, users even can predict the possible analysis result intuitively, and choose the most suitable social network analytical solutions to the data.