Concept Formation by Incremental Analogical Reasoning and Debugging
Mark H. Burstein · 1986
This chapter presents a model of learning by analogical reasoning. The model is based on two main ideas, namely, (1) that the analogies used in learning about an unfamiliar domain depend heavily on the use of previously formed causal abstrac-tions in a familiar or base domain; (2) that these analogies are extended incremenrally to handle related situations. CARL is a computer program that learns about the semantics of assignment statements for the BASIC programming language. It is described as an illustration of causally driven analogical reasoning and learning. The model maps and debugs inferences drawn from several commonly used analogies to assignment in response to presented examples. It has often been said among A1 researchers that learning something new requires knowing a lot about it already. This is certainly true for learning by analogy. This chapter shows how prior knowledge can be applied in one specific kind of learning by analogy, namely, the formation of new concepts in an unfamiliar domain