ABSTRACT
This chapter describes Graphical Instruction in LISP, an intelligent tutoring system for programming that generates explanations directly from its problem solving knowledge rather than relying on prestored text associated with that knowledge. In an intelligent tutor for programming, the problem solver must know about the variety of algorithms and the goals that each algorithm achieves. The model tracing method requires that the response manager match each student step against the action of a problem solving rule. There have been a number of recent demonstrations that knowledge representations such as rules and plans can be effectively used to track and provide assistance for a student's problem solving in mathematics, computer programming and medical reasoning. As in rule-based systems, a tutor built upon a knowledge base of plans compares each student's step in a solution to those steps suggested by its plan-driven decomposition of the problem.
