QUEST: Quality of Expert Systems (QUEST: Kwaliteit van Expertsystemen)
Abstract
This study contains a summary of the results of the technology project QUEST: Quality of Expert Systems, carried out under commission of the Dutch Ministry of Defence, Directorate Defense Research and Development. Participants in the project are TNO Physics and Electronics Laboratory (TNO-FEL) , University of Limburg (RL) and the Research Institute for Knowledge Systems (RIKS). After an analysis of the problems encountered in expert systems development, a quality framework is developed which takes a view at the quality problem from three perspectives: the quality of the development process, the quality of the specifications and the quality of the expert system viewed as a product. In order to get a better grasp of the problem a number of methods and techniques, derived from conventional and artificial intelligence systems development, are reviewed. Secondly, the conceptual similarities between databases and knowledgebases are stressed. The use of conventional specification methods, in particular Nijssens Information Analysis Methodology (NIAM), is considered. In addition to this algorithms for preserving consistency and integrity of the knowledgebase are compared. Thirdly the modularity and structure of knowledgebases is examined, together with the applicability of conventional testing methodologies in expert systems. Lastly it is demonstrated that the integration of database theory and artificial intelligence signifies a step in the direction of a better quality control of expert systems.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 1990
- Accession Number
- ADA245620
Entities
People
- Jacques H. Lenting
- Michael Perre
Organizations
- Netherlands Organisation for Applied Scientific Research