Structuring diagnostic knowledge for multiple usage

✍ Scribed by R.J. Allwood; A. Goodier

• Publisher: Elsevier Science
• Year: 1996
• Tongue: English
• Weight: 488 KB
• Volume: 26
• Category: Article
• ISSN: 0965-9978
• DOI: 10.1016/0965-9978(95)00116-6

✦ Synopsis

Large and reliable knowledge-bases for expert systems are as difficult and expensive to construct as large conventional programs. The discipline of structured programming has brought great dividends to procedural programming and this paper attempts to show that similar dividends can be obtained from careful structuring of knowledge-bases. Whilst developing a system with 276 goals and over 1000 rules to diagnose faults in flat roofs, the authors identified a structure into which any defect diagnosis system could place its goals and its knowledge. The benefits come not only in assisting experts in the early and critical phase of identifying all possible defects and in the equally difficult task of eliciting knowledge about those defects, but also from the possibilities of using the completed knowledge-base for tasks other than diagnosis, such as design checking and surveying. These ideas are illustrated by examples taken from the flat roof diagnosis system.