Knowledge Discovery by Network Visualization

Hyperlinks among webpages are very important information and are widely used for webpage clustering and webpage ranking. With the explosive growth in the number of webpages available online, the exploration of hyperlinks among webpages becomes a very challenging problem. Information visualization provides an effective way of visualizing hyperlinks and can help users gain insights into the relationships of webpages. In this paper, we present some novel computer graphics techniques to visualize the hyperlinks among webpages. We propose a visual encoding scheme for five dimensional hyperlinks data and two constrained 3D layout techniques for incoming and outgoing links of a single webpage. To reveal the hierarchical structure of webpages as well as the hyperlinks information, we extend the treemap representation. Our representations are visually appealing and can effectively reveal linkage patterns among webpages. Experimental results and a user study demonstrate the effectiveness of our system. Our system can facilitate E-learning and help students understand the complex structures and hidden patterns in network datasets.