Transfer of Rule-Based Expertise through a Tutorial Dialogue

摘要

This dissertation describes an intelligent, computer-aided instructional (ICAI) program, named GUIDON, with capabilities to carry on a structured case method dialogue, generate teaching material from production rules, construct and verify a model of what the student knows, and explain expert reasoning. The principle objective of this research has been to convert MYCIN, a knowledge-based consultation program, into an effective instructional tool. GUIDON combines the subject matter knowledge of the consultation system with tutorial discourse knowledge, while keeping the two distinct. MYCIN-like knowledge-based consultation programs are designed to provide expert-level advice about difficult scientific and medical problems. High performance is attained by interpreting a large, specialized set of facts and domain relations that take the form of rules about what to do in a given circumstance. Such a rule base is generally built by interviewing human experts to formulate the knowledge that they use to solve similar problems in their area of expertise. While it is generally believed that these programs have significant educational potential, little work has been done to evaluate the problems of realizing this potential.