Machine Transformation of Advice Into a Heuristic Search Procedure

A key problem in learning by being told is operationalization: the development of procedures to implement advice that is not directly executable by the learner, such as the advice “avoid taking points” in the card game hearts. One way to operationalize such advice is to reformulate it in terms of a general “weak method”, such as heuristic search. This chapter is a case study in the mechanical mapping of domain-specific problems onto general methods, using as a detailed example the derivation of a heuristic search procedure for the advice “avoid taking points.” The derivation consists of a series of problem transformations leading from the advice statement to an executable procedure. The operators used to perform these transformations are implemented in a program called FOO as domain-independent transformation rules that access a knowledge base of task domain concepts. Some of the rules construct a crude generate-andtest procedure; others improve it by deriving new heuristics based on domain knowledge and problem analysis. To test its generality, FOO was also used to operationalize a music composition task; many of the same rules proved applicable.