An Evaluation of the Circles Information Visualization Tool for Presenting Bayesian Network Output

Abstract

Bayesian Networks are complex systems models that present rich output that can be difficult to communicate to users. In this paper a novel information visualization tool is evaluated for performance on accuracy, efficiency and user comprehension criteria. The visualization is tested across a range of user tasks, including identifying important information, inferring relationships between factors and comparing model outputs. While the interpretation of model output is less accurate for the visualization tool in question, this is balanced by significant gains in efficiency and user comprehension. It is suggested that the visualization is appropriate in contexts such as operational management where users refer to the tool often for support in making uncertain decisions, and can best be defined as a casual visualization to complement existing decision making activities on a daily basis.