Quantitative evaluation of expert systems

The rapidly increasing application of expert systems has lead to the need to assure their quality, especially in the critical areas where a fault may be serious. Metrics can provide a quantitative basis for such assurance activities. In this paper, we investigate several formal metrics that are designed for measuring the three important characteristics of expert systems: the size, search space, and complexity. The applications of these metrics to assess or predict the quality of expert systems are also addressed. In addition, a comprehensive evaluation and comparison of the metrics is conducted with the aim to assess their validity when applying them to quantify expert systems