Extending Problem Solver Capabilities through Case-Based Inference

Abstract

This document reviews work done on case-based reasoning. In this sort of reasoning, the problem solver makes inferences based directly on previous cases rather than using the more traditional method of reliance on general knowledge. Case-based reasoning results in several enhancements to problem- solving behavior over time. First, recall of previous failures warns the problem solver of the potential for failure, and allows it to avoid the repetition of past mistakes. Second, the previous decisions that have been made are suggested to the problem solver so that its decisions do not all have to be made starting from scratch. This lessons the search space, and also serves as a way of shortcutting the constraint satisfaction process. Third, if abstract schemata can be derived from cases that have been seen previously, generalized knowledge can be augmented. This allows real shortcuts in problem solving. Decisions that previously took several steps in reasoning to make may become easier through the application of a generalized schema. Keywords: Cognitive psychology.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 01, 1987
Accession Number
ADA191332

Entities

People

  • Janet L. Kolodner

Organizations

  • Georgia Tech

Tags

Communities of Interest

  • Autonomy
  • Human Systems

DTIC Thesaurus Topics

  • Abstracts
  • Acquisition
  • Classification
  • Computer Science
  • Computers
  • Language
  • Learning
  • Linguistics
  • Machine Learning
  • Military Research
  • Natural Languages
  • Psychology
  • Reasoning
  • Social Sciences
  • Technical Information Centers

Readers

  • Applied Combinatorial Optimization and Logic Circuit Design.
  • Educational Psychology
  • Team-Based Human-Centered Cognitive Task Decision Making and Information Performance.

Technology Areas

  • AI & ML
  • AI & ML - Machine Learning Algorithms
  • Space