Review of Rule Quality Measurement: Metrics and Rule Evaluation Models

Abstract

A rule-based system is a system based on the set of rules used to make inference knowledge. The system gathers knowledge into the representation of knowledge in the form of a rule. However, the knowledge in the form of the rule is inductive, meaning that the algorithm can construct the rule by studying a limited number of cases and then the induced rule of a limited number of cases and then generalize it to the general reality from time to time. This, of course, has the degree of inaccuracy in expressing reality into knowledge, or an experienced expert builds it but it is not absolute that the knowledge it possesses is 100% accurate or always consistently true from one time-space location to another time-space location. Therefore, the need for a formula that can measure the quality of the resulting rule and assess the consistency of the rule. In this study, we did a review of the ideas of people trying to measure knowledge built inductively by either the algorithm or the experts. These measurements are based on several parameters defined by them according to the underlying assumptions. This review seeks to partially present how ideas to measure the rule as knowledge representation from a varied viewpoint and how people construct evaluation models to assess the resulting regulations either from the experts or human experts as well as those resulting from the induction rule algorithm much developed.