Knowledge Acquisition for Expert Systems in Construction

Abstract

The authors have investigated knowledge acquisition both by developing eight systems themselves and by involvement in a survey of 70 systems developed by others. They conclude that the key to successful expert system development lies in the organizational environment in which the application is undertaken. Hence development should be client led and problem oriented with clearly defined end users. Simple applications are more likely to be successful. Several knowledge elicitation techniques will be required for any given application. The authors have investigated unstructured and structured interviews, case histories, rule induction, observation, paper models (intermediate representations), iterative prototyping, use of published material and the problems of eliciting uncertain knowledge. A general approach to knowledge elicitation is suggested. Systems developed by the authors and described in the report include BREDAMP (building dampness diagnosis), CRANES (tower crane selection), BID/NO BID (bidding decision), MATSEL (boiler tube steel selection), RRES (aeroengine fault diagnosis) and GECES (alternator out-of-balance correction).

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 01, 1988
Accession Number
ADA203458

Entities

People

  • Chris Cooper
  • Geoffrey Trimble

Organizations

  • Loughborough University

Tags

Communities of Interest

  • Biomedical
  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Artificial Intelligence
  • Civil Engineering
  • Computer Programming
  • Computer Programs
  • Computer Science
  • Computers
  • Construction
  • Contracts
  • Databases
  • Engineering
  • Engineers
  • Expert Systems
  • Graphics
  • Knowledge Based Systems
  • Materials
  • Structural Engineering
  • Word Processors

Readers

  • Government Contracting/Procurement.
  • Software Engineering.
  • Theoretical Analysis.