ENRICHING VOCABULARIES BY GENERALIZING EXPLANATION STRUCTURES

A problem with learning systems is that often the language with which the user must express information to the system is too complex. This paper discusses techniques based on Explanation-Based Learning by Observation that allow the user to enter the information required by the system in a simplified form comfortable to him or her. A user enters a description of a specific problem and its solution in simple terms. The description is then repeatedly enriched by the system to create a general solution that applies to the current problem and conceptually similar ones. The learning system first formulates, using domain-specific knowledge, an explanation of the human's solution to the specific problem. Next, the goal stated by the user is generalized with respect to any domain information about possible goals of the actions performed by the user in the solution. Finally, the explanation is extended to apply to the generalized goal. If the user's solution implicitly contains iteration or recursion, an explicitly iterative or recursive concept is produced. Each step involves a reformulation that transforms the user's simple description into a more general form, allowing the system to learn a much more general concept. A system is being developed that applies these techniques in the domain of automatic programming.