ABSTRACT
The most fundamental step in the knowledge acquisition phase of the development of an expert system is the elicitation of knowledge from a skilled individual. The knowledge acquisition phase has typically involved the knowledge engineer’s working closely with a specialist to elicit relevant knowledge from the latter’s domain. This is typically a tedious and ad hoc cycle that consists of extensive verbal interviews followed by the construction of prototypes, testing, and more interviews. This approach has two significant drawbacks—it has been extremely laborious, and domain experts often have difficulty articulating their knowledge in forms useful to the knowledge engineer. Indeed, it has been suggested (Feigenbaum & McCorduck, 1983) that “the problem of knowledge acquisition is the critical bottleneck in artificial intelligence” (p. 80).
