This paper proposes a tool that presents the link structure visually. Access probability is derived in place of PageRank as an indicator of the importance level of a webpage. The proposed tool draws a perspective diagram on a hemisphere. The diagram shows a website as a tree structure with its Home page at the root. The tree structure is built based on the distribution of the access probabilities and the scatter diagram of their regression line. In this perspective diagram, a webpage is represented by a globe primitive with the access probability as its volume. A link is represented by a cylinder primitive with the number of transitions between webpages as its radius. Each of these can be identified by selecting it. Whether a link should be visible or invisible (number of displayed elements) can be controlled using the access probability or the number of transitions as a threshold.

View full abstract

Hierarchical data visualization is very useful for understanding, monitoring, searching, and analyzing various large-scale data. Authors are focusing on a hierarchical data visualization technique, which represents leaf-nodes as icons, and non-leaf-nodes as nested rectangles. Key technology of the visualization technique is the algorithm for packing rectangles onto limited display spaces.
This paper proposes an improved method for packing rectangles for hierarchical data visualization. The method orthogonally divides the display area by extension lines of edges of previously placed rectangles. By referring the grid-like subspaces of the display area, the technique quickly finds the adequate positions to place remaining rectangles. This paper provides numerical evaluations of results of the method, and proofs the advantages of the method against existing hierarchical data visualization techniques.

View full abstract