Visualization of assertion confidence in fuzzy rules

A system is presented which is oriented at aiding medical professionals in the diagnosis of neuromuscular disease using a fuzzy rule-based classification system. Visualization of the fuzzy rules, which are contributors to the overall classification, allows the user to determine their level of confidence with the classification of the system. During the development of this system, the choice between two alternative components arose, which required that an evaluation of the two contrasting visualization techniques be performed. An emphasis on communication of information to the operator resulted in a decrease in the effectiveness of traditional HCI evaluation techniques. Instead, a methodology is introduced which attempts to quantify the level of information which is absorbed by the observer. Comparison of user performance with each style of visualization on the same dataset allows the experimenter to determine which alternative is most effective. It is revealed that this methodology is generic enough that it can be applied in other visualization heavy applications where traditional techniques have failed to produce an adequate evaluation.