KSL-92-12

Diagrammatic Reasoning in Problem Solving

Reference: Iwasaki, Y.; Tessler, S.; & Law, K. Diagrammatic Reasoning in Problem Solving. Knowledge Systems Laboratory, February, 1991.

Abstract: Diagrammatic reasoning is a type of reasoning process in which the primary means of inference is the direct manipulation and inspection of a diagram. This type of reasoning, which we also call visual reasoning or pictorial reasoning, is an integral part of human problem solving behavior, especially for problems involving spatial relationships among physical or abstract objects. Our research goal is to investigate the role that visual reasoning plays in the overall problem solving process. In this report we present our ideas about the human visual reasoning process, and describe our proposal to emulate this capability with a computer program. Our approach is to develop a computer program with both symbolic and diagrammatic reasoning components that will be able to reason qualitatively using pictures in the same way that poeple do. Our conjecture is that the familiarity of the mixed problem solving approach taken by the program will be more understandable to a user and more useful as an aid to teaching visualization skills than a purely symbolic reasoning program would be. Computatuional efficiency may also be enhanced because the program would be able to focus more effectively on the critical features of the problem. Our program will solve a class of structural analysis problems within the domain of civil engineering, in part because their solution process requires domain-specific knowledge as well as visualization skills. We postulate that domain knowledge may influence the way visual reaoning is performed. Since we have previously implemented the same set of problems in a traditional symbolic approach, we will have the opportunity to analyze the relatioship between symbolic and diagrammatic problem solving and to examine the relative advantages of each.

Full paper available as ps.


Jump to... [KSL] [SMI] [Reports by Author] [Reports by KSL Number] [Reports by Year]
Send mail to: ksl-info@ksl.stanford.edu to send a message to the maintainer of the KSL Reports.