Implementation of heuristic problem solving process including analogical reasoning

Abstract

This paper discribes a heuristic problem solver named HPSA. HPSA is constructed to explore human problem solving process with hypotheses formation including analogical inference. HPSA simulates two general types of human problem solving process; the first is deductive problem solving using domain-specific knowledge of the target domain and common knowledge, and the second is analogical reasoning executed between the target and source domains which are selected on the basis of some similarities. This system has the following advantages which most of precedent studies lack; that is, (1)HPSA enables simulation of a whole process of heuristic problem solving, besides either deductive problem solving or analogical reasoning, (2)problem solving with analogical reasoning can be executed from pragmatic aspects, i.e. goal-oriented problem solving and modification of pragmatic aspects can be simulated, (3)all phases of analogical reasoning are realized, and (4)multiple analogy is also realized. This problem solver is partly based on the observations on actual human problem solving processes with hypotheses formation. Hence, HPSA also tries to provide a cognitive simulation tool.