next up previous
Next: Conclusion Up: H3: Laying Out Large Previous: Analysis

Future Work


The choice of spanning tree has a considerable bearing on the visual impact of the drawn graph. We would like to consider additional heuristics for finding reasonable spanning trees, particularly for domains other than the Web. We are also interested in alternative arrangements of child nodes on parental hemispheres.

We are quite interested in extending the underlying visualization system to allow effortless navigation through huge graphs by automatically expanding subtrees as they approach the center of interest and contracting those that move out to the periphery.

Extremely large graphs will not be understandable if presented only at a single level of detail, even if we had an infrastructure for smooth automatic navigation through millions of nodes. We need to find appropriate abstractions for both graph theoretic structures such as subtrees and for domain-specific structures such as a Web site.

Tamara Munzner Tue Jul 22 08:42:41 PDT 1997