A Comparison of Knowledge Acquisition Techniques Used in the Development of Expert Systems

Abstract

The primary objective of this theses was to compare three knowledge acquisition techniques used to gather knowledge for the development of an expert system. The goal was to determine which technique produced knowledge in a form most suitable for incorporation into an expert system. The three acquisition techniques compared were interviewing, task observation, and concept mapping. Three experts were selected and randomly paired with a technique. Knowledge acquisition sessions were then conducted with each expert using the technique assigned to that expert. The knowledge extracted from these acquisition sessions was then compared. Overall, concept mapping produced more rules, in less time, and with fewer inferences than the interview or task observation techniques. Additionally, the knowledge base acquired through the concept mapping technique was more complete. Finally, concept mapping required one less translation of the knowledge to arrive at a form necessary for programming the expert system. An expert system was developed using the concept mapping technique and was validated in a field test. Results showed that the solutions provided by the expert system matched those provided by the human experts.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Sep 01, 1990
Accession Number
ADA229252

Entities

People

  • James R. Heatherton

Organizations

  • Air Force Institute of Technology

Tags

Communities of Interest

  • Air Platforms
  • Human Systems

DTIC Thesaurus Topics

  • Acquisition
  • Aircrafts
  • Airlift Operations
  • Artificial Intelligence
  • Computer Programming
  • Computer Programs
  • Computer Science
  • Computers
  • Databases
  • Engineering
  • Engineers
  • Field Tests
  • Logistics
  • Personnel Management
  • Software Development
  • Software Development Tools
  • System Software

Readers

  • Artificial Intelligence
  • Brain and Cognitive Science; Experimental Psychology; Cognitive Neuroscience
  • Systems Analysis and Design

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference