AN INTELLIGENT SYSTEM FOR TASK-SPECIFIC VISUALIZATION ASSISTANCE

As the size and complexity of scientific data sets increase, scientists need help in analyzing relationships and structure within the data. Designing effective visualizations for such data sets is an extremely complex process. In this paper we describe a tool, which we call Visualization Assistant, that can provide assistance in deriving perceptually effective designs which will enhance the user's task performance. This tool considers the scientist's task and the interaction between perceptual dimensions, perceptual organization and texture segregation in generating design constraints. These constraints will limit the number of potential designs to those that will most effectively convey relationships within the data. Visualization Assistant then guides the user by ranking designs according to their perceptual effectiveness and assists the user in reading designs by offering 'attention directions'. Finally, we outline a plan for implementing this system using 'case-based reasoning'.