Analogical Explanations

Abstract

Analogical problem solving has been an area of considerable research in Al for some time, but the use of analogies in language generation as a means of explanation has attracted much less attention. Analogies are very interesting linguistic devices, because they convey a large amount of information in a relatively concise fashion. While using analogies is very common in language, most generations systems are not able to use analogies when providing explanations. In this paper, we discuss analogies in general, and outline a framework which will lay the foundations for knowledge representation suitable for generating inter-domain analogies. We argue that language should help in knowledge representation, and employ the Upper-Model, an ontology of abstractions used in natural language generation, to help organize representation of domain terminology. Explanations, Analogy, Natural language generation.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 01, 1990
Accession Number
ADA269585

Entities

People

  • Cecile L. Paris
  • Vibhu O. Mittal

Organizations

  • University of Southern California

Tags

Communities of Interest

  • Advanced Electronics
  • Biomedical

DTIC Thesaurus Topics

  • Abstracts
  • Analogs
  • Artificial Intelligence
  • Blood Coagulation
  • Causal Reasoning
  • Classification
  • Cognitive Science
  • Computers
  • Demographic Cohorts
  • Expert Systems
  • Health Services
  • Information Science
  • Language
  • Models
  • Natural Languages
  • Ontologies
  • Reasoning

Readers

  • Artificial Intelligence
  • Systems Analysis and Design