Content-based visualization for intelligent problem-solving environments

Abstract

Mechanisms are proposed to allow visualization to become an active agent in the problem-solving environment. Two main problems are addressed: dealing with too much data and allowing simulation steering. The agent solves these problems by extracting semantically interesting data for spatial and temporal understanding, enabling a dynamic and flexible behavior for simulation-integrated control, and intelligently creating visualizations that intuitively display the selected data. A significant event language is proposed to capture the semantically interesting data through event expressions which can then be parsed and monitored. Behavior is achieved through programmable, hierarchical finite-state machines with events mapped to the arcs and interactions with the visualization and simulation mapped to the nodes. A knowledge-based data mapping process has been designed which uses visual perceptual knowledge and problem-specific knowledge represented in cognitive maps to create visualizations. This architecture has been applied within a watershed simulation application and a prototype has been developed.