Knowledge validation in the building of a knowledge-based system

摘要

In the field of knowledge engineering, knowledge acquisition and knowledge-based system (KBS) validation have been two separate fields until now. We explain in this article how the recent trends of knowledge acquisition, based on lifecycle and knowledge modeling, may trigger a new approach in knowledge validation. The KBS lifecycle makes it possible to classify different types of validation in the different phases of the lifecycle, and to perform a complete validation in quite a systematic way, validating one phase after another. However, knowledge validation essentially deals with both modeling and implementation phases in a KBS lifecycle. During the knowledge modeling phase, an expertise model can be intrinsically validated. It may also be a guide to structure the KBS. And, it is perhaps the main important point, an expertise model may be considered as specifications for the KBS. We precisely illustrate these points with the knowledge modeling used by the KADS methodology. Conversely, validation may influence knowledge acquisition. Indeed, our approach of the knowledge validation makes it possible to call into question the expertise model. More generally, a result of our study is that one only can mention the lifecycle of a KBS if an integrated view of knowledge acquisition and validation is used.