Computer-aided knowledge engineering, a framework for building and maintaining expert systems
- Washington State Univ., Pullman, WA (United States)
Conventional software has been around for a long time. Methodologies and tools have been developed to ensure the validity and maintainability of its applications during both development and productive life cycles. Knowledge-based expert systems, on the other hand, are a newly emerging field. They are equipped with knowledge representation schemes and inferencing capabilities that represent and manipulate complex data and knowledge types. They are more complex, solve different types of problems, and demand different types of methodologies to ensure the validity, soundness, and completeness of their knowledge bases. The knowledge engineering team has to make decisions on the knowledge representation schemes (taxonomic, decomposition, descriptive, and declarative knowledge), inferencing mechanisms, and validation and verification mechanisms. Historically, during the development life cycle, knowledge-based validation and verification have been carried out manually. It is the process by which knowledge engineers and their experts review the knowledge base and look for syntactic and semantic errors, build decision trees, check for the logical connectivity of the rules, run as many test cases as possible, and watch for mismatches between the system's and experts' output. This is a time-consuming, error-prone process; it does not guarantee finding all errors. Manually built decision trees are tedious and need to be redrawn after each change in the knowledge base. Finally, running test cases is a hit-or-miss activity; it only tells if the system has errors; it does not tell where an error occurred or what caused it.
- OSTI ID:
- 6946013
- Report Number(s):
- CONF-931160-; CODEN: TANSAO
- Journal Information:
- Transactions of the American Nuclear Society; (United States), Vol. 69; Conference: American Nuclear Society (ANS) winter meeting, San Francisco, CA (United States), 14-18 Nov 1993; ISSN 0003-018X
- Country of Publication:
- United States
- Language:
- English
Similar Records
Guidelines for the verification and validation of expert system software and conventional software: Volume 5, Rationale and description of verification and validation guideline packages and procedures. Final report
Guidelines for the verification and validation of expert system software and conventional software: Rationale and description of V&V guideline packages and procedures. Volume 5