Prototyping an Expert System Shell with the Logic-Based Approach

Expert system is a computer program that is different from other conventional computer programs in that it incorporates specific knowledge, which can be human judgment, experience, and expertise, in order to provide knowledgeable advice to users. The main difference is caused from the structure of an expert system that contains the two loosely coupled parts: knowledge inference engine and knowledge base, instead of one tightly integrated structure. The loose coupling allows knowledge base contents to be dynamically added, removed, modified, or even completely changed to another subject area, whereas the inference engine remains intact and needs no modification. We refer to the knowledge inference engine as an expert system shell because it can be viewed as an outer layer program to infer knowledge advice from the inner knowledge base core. The development of expert system shell can be efficiently achieved through the support of logic-based language such as Prolog. In this paper, we propose and demonstrate a different scheme of expert system shell development using a constraint-based paradigm with the ECLiPSe constraint system. Comparisons of the two paradigms have been done in terms of computational time, memory usage, and lines of code. The experimental results reveal that the difference in lines of code of the two paradigms is insignificance, but the constraint-based paradigm uses less memory during execution and provides more concise form of knowledge representation.