• home
• current students
• graduate
• events

Presenter: Dr. Aaron Quigley - Post-Doctoral Researcher, Mitsubishi Electric Research Labs, Cambridge, Massachusetts - Faculty Applicant
Supervisor: Dr. Nigel Horspool - Chair, Department of Computer Science

Date: Mon, January 14, 2002
Time: 13:30:00 - 14:30:00
Place: Engineering Office Wing (EOW) Room #430

ABSTRACT

ABSTRACT:

Force-directed graph drawing algorithms produce effective visualizations of abstract relational information, which tend to reveal the natural clusters in the data. However, these methods tend not to scale well due to the high computational cost of calculating the forces between every pair of nodes. This talk will present a series of efficient force directed drawing algorithms, for the two and three dimensional layout of large graphs with thousands of nodes. These algorithms are based on a hierarchical clustering of the nodes, codified in a graph model. This model is inspired by space decomposition approaches from N-body particle physics and plasma flow modeling.

The FADE paradigm presented in this talk also addresses the problems of screen space use, the cognitive load on the user, the time to render the picture, and creating pictures of measurably high quality. The graph model used in this paradigm is based on a geometric clustering which allows for the multilevel viewing of large undirected graphs along with the efficient layout of the graph.

This talk will outline the results of two case studies which demonstrate the models, measures and methods of the FADE paradigm, which are required to produce and evaluate the drawing of large amounts of relational information. The effectiveness and efficiency of the methods are evaluated with rigorous quality measures. The large data sets used in these case studies are from the Matrix Market at NIST and resource flow graphs from the Bauhaus Project produced by partial RIGI analysis.