Title: Towards a Solution of Large Graph Visualization: Layout, Rendering, Interaction
Technical Area: Data Visualization
Graph visualization is an increasingly important area of research within Information Visualization, both because of its everyday applications and the need in many fields to view large and complex graphs. With the growth of data on www, graph visualization and navigation techniques appear in numerous analysis topics such as Social Networks, Business Intelligence, and etc. While, there have been increasing number of challenges in the ability to visualize and to navigate in these potentially large graphs.
- Large graph is a term for massive data sets having large, more varied and complex graph tructure, it creates values for business and research, but pose significant challenges in terms of networking, storage, management, analytics, and etc. When we layout large graphs, memory quickly become limited and layout cost lots of time since most layout algorithms can not be executed in parallel.
- Large graphs in limited screen size will definitely result in visual clutter and confusion. Users do not know where to go to find what they need. On the other hand, the results imply in visual clutter not only increases errors, but also increases the confidence with which Users make decisions.
- When we present large graph to users, we need make it clear how to abstract valuable nformation for users and how to allow user to interactively navigate the large graph to get more information.
- Most visualization applications are run on web, the performance of rendering and interaction is also a big challenge. Both memory of web browers and drawing engine performance is limited.
The theme target is to visualize large graph on web browers, including key info abstraction, large graph ayout, rendering and interaction. The measurable outputs of the research may include but not limited to:
- Work out a solution of large graph visualization.
- Provide layout algorithm and effective interactions for large graph.
- Resolve potentially problems in large graph visualization, such as visual clutter.
Related Research Topics
- Node-link graphs are most popular for visualizing inherent relations within data. This epresentation of graphs includes trees and more general networks. In addition to node-link layout, a few new types of visualization models have been proposed, such as space Nested layout.
- With the proliferation of data, parallel graph analytics has become a difficult task because it nvolves executing iterative algorithms on very large graphs. This has led researchers to explore cceleration strategies, some of which produce approximate results. There are two main strategies: algorithmic and code-centric. The algorithmic approach is application specific and thus the ideas for one application may not transfer to others. The code-centric approach transforms the pplication so that at runtime it switches between code versions or skips computations to save time, albeit acrificing accuracy. However, for applications whose behavior is input sensitive, intelligent skipping is difficult as the program lacks global view of input characteristics.
- Visual clutter is one of the primary problems caused by large graphs. A good layout should otentially minimize visual clutter. However, trying to find a good layout for reducing visual clutter is not practical, since this kind of problem usually involves optimization, which is usually difficult for arge graphs. Researchers have made many different attempts to address or at least minimize this problem. Here are some widely used ways: edge displacement, node clustering, and sampling.