The Nature of Heuristics

Abstract

The bottleneck of building expert systems is knowledge acquisition, and one long-range solution is for the program to learn via discovery. New domains of knowledge can be developed by using heuristics, yet as they emerge new heuristics are needed. They in turn can be discovered by using a body of heuristics for guidance. How exactly does this process work? Must there be a separate body of 'meta-heuristics'? How intertwined are heuristics with Representation of knowledge? In trying to find new heuristics, is it cost- effective to try to improve the existing representation of knowledge, and if so how can this be automated? What is the nature of heuristics, their 'first-order theory'? What are the implications of such a theory upon the design of a program which discovers new heuristics? These questions are among those that our research -- and this paper -- address.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 29, 1980
Accession Number
ADA096511

Entities

People

  • Douglas B. Lenat

Organizations

  • Stanford University

Tags

Communities of Interest

  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Acquisition
  • Algorithms
  • Artificial Intelligence
  • Computer Programs
  • Computer Science
  • Computers
  • Expert Systems
  • Geometry
  • Language
  • Mathematics
  • Number Theory
  • Numbers
  • Plane Geometry
  • Scientific Theories
  • Set Theory
  • Theorems
  • Uncertainty Principle

Fields of Study

  • Computer science

Readers

  • Artificial Intelligence