Making Knowledge Engineering Technology Work
Published: 1997 Juni
Buchtitel: Proceedings of the 9th International Conference on Software Engineering & Knowledge Engineering (SEKE'97), Madrid, Spain, June 18-20, 1997
The rapid-prototyping approach of the early 1980's failed to deliver high-quality knowledge-based systems. As a reaction, in the early 90's, there has been a large activity in the knowledge engineering community to define methodologies for principled knowledge-based system development. These methodologies (e.g., CommonKADS, MIKE, VITAL, TASK) succeeded in organising the development process as a set of intermediate models of the functionality of the system to be built, to end in the final implementation. Feedback from industries that use these methodologies in practice reveals, however, that building a high-quality implemented system remains a difficult and error-prone process. The main reason for this is that, although the structure of the development process is rather well understood, the transition process is less clear. In this paper, we outline an approach for assuring the transition of a conceptual model into a final implementation which satisfies particular quality criteria. The approach is based on identifying adequate transition paths and activities on models in order to achieve particular quality criteria. An essential part of the approach is formed by methodological guidelines and tool support such as change-management systems, reusable libraries, translators, decision support tools, and support for validation and verification.