Issue No. 06 - November-December (1997 vol. 12)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/64.642963
Many enterprises use several expert system shells to develop production-rule-based expert system applications. Each shell has its own production-rule language (and inferencing capabilities). Knowledge engineers are often burdened with understanding several production-rule languages to support the enterprise's expert system applications. This article examines the syntactic similarities and differences in the production-rule languages from five expert system shells. The authors have developed a composite, or canonical, production-rule syntax for the languages, which provides knowledge engineers with a common language for defining production rules. Similarities and differences in the capabilities provided by each production-rule language show when information can, or cannot, be reused in different expert system shells. The reuse of individual syntax components is tracked as a measure of the amount of similarity between the languages. The similarities and differences constrain the features of the composite syntax that can be used for a target shell, and limit sharing of production-rule information.
V. M. Johnson and J. V. Carlis, "Building a Composite Syntax for Expert System Shells," in IEEE Intelligent Systems, vol. 12, no. , pp. 60-66, 1997.