Computer-Based Tutors for Explaining and Managing the Process of Diagnostic Reasoning
Abstract
AI(Artificial intelligence)-based instructional programs, often called intelligent tutoring systems (ITS), use qualitative modeling techniques to represent: 1) processes in the subject domain (e.g., a steam propulsion plant, an electronic circuit), 2) problem-solving processes (e.g., diagnostic strategy, programming methods), and 3) communication processes (e.g., the Socratic method, case-method discourse, and rhetorical principles in explanation) ('Qualitative student models'). Typically, instructional programs may represent only one or two kinds of these processes. When a simulation model of problem-solving processes is incorporated in the program, a basis is provided for evaluating and assisting the student in a very general way. Such programs, which can solve the same problems given to a student, are called knowledge-based tutors (Knowledge-based Tutoring). Early in our research, we identified the importance of representing problem-solving processes in a well-structured procedural language. In a sequence of programs, we demonstrated basic Al techniques for achieving the separation of domain facts from a diagnostic procedure (NEOMYCIN), and the advantages of this separation for explanation and student modeling (IMAGE, ODYSSEUS). The generalization of our work has had a significant impact on expert systems and tutoring research.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1988
- Accession Number
- ADA211738
Entities
People
- William J. Clancey
Organizations
- Stanford University