Knowledge Discovery in Rule-Bases
Résumé
In rule based, automated management systems, knowledge is represented explicitly as long as it is necessary for the functioning of the system. However, some knowledge which might be quite useful for maintenance purposes, remains implicit and disseminated in the rule base. We present an original approach for the discovery of such implicit knowledge, based on machine learning techniques. We will illustrate the use of our approach in the book-keeping domain, where it has proven its interest within the scope of an industrial project.