Accession Number : ADA184345

Title :   Analogical Inference and Analogical Access.

Descriptive Note : Technical rept.,

Corporate Author : ILLINOIS UNIV AT URBANA DEPT OF COMPUTER SCIENCE

Personal Author(s) : Gentner,Dedre

PDF Url : ADA184345

Report Date : 04 Aug 1987

Pagination or Media Count : 56

Abstract : Analogy is a powerful technique in commonsense learning and reasoning. People use analogies in problem solving, in developing mental models of a new domain, and in communicating knowledge. To model these natural uses of analogy, we need to understand the whole process of analogizing from the first, starting with access and ending with drawing inferences or extracting a principle from an analogy. In this paper, I first review the structure-mapping theory of analogical processing and describe a simulation of the theory. I then extend this framework to accessing analogy. I discuss some recent research in our lab that suggests that the accessibility of an analogical match is governed by different factors from its inferential soundness. Finally, I consider some competing theoretical approaches to analogy and suggest an integrated architecture for analogical processing.

Descriptors :   *LEARNING, *ANALOGIES, *ARTIFICIAL INTELLIGENCE, REASONING, MAPPING(TRANSFORMATIONS), STATISTICAL INFERENCE, PROBLEM SOLVING

Subject Categories : Humanities and History
      Cybernetics

Distribution Statement : APPROVED FOR PUBLIC RELEASE