An approach to the evaluation of a diagnostic system with incomplete information

The task of evaluating diagnostic systems is receiving an increased attention, particularly for those systems liable to suffer disturbances or malfunctions that must be detected in a short period of time. An efficient diagnostic process should be capable of guaranteeing an optimal trade-off between the response time and the quality of the results. An approach to diagnosis systems with incomplete information is based on the following steps: the first step consists of obtaining the most accurate extended representation, that is all the complete information contained in the rule base. From the analysis of this information, the characteristics that an optimal rule base exhibit must be deduced. Finally, the quality of the rule base (RB) is calculated by comparing the characteristics of RB to those of an optimal rule base.