On the Role of Ontology in Modern Expert Systems Development

Ontologies, in the strict artificial intelligence sense, are used to define the allowable objects1 belonging to a universe of discourse or knowledge domain. An expert system is an artificial intelligence application that performs symbolic reasoning over a knowledge domain. When designing an expert system, the choice of a knowledge representation schema and its accompanying programmatic encapsulation largely determine the system\u27s present and future reasoning capabilities. Therefore, for engineering and other business reasons, it is reasonable to ask if an expert system\u27s knowledge base should be constructed from a formally specified ontology rather than by relying on implicit ontologies inherent in the chosen expert system\u27s stereotype. This paper examines the general question of whether or not the creation of a formal domain ontology is a prerequisite to designing and building an expert system. A set of decision metrics is derived from survey data, which is then augmented by feedback from the academics and professionals who participated in the survey. The relative merits of each metric are discussed as well as topics for further investigation