EXTENDING CONCEPTUAL DEFINITIONS WITH DEFAULT KNOWLEDGE

In description logics, default knowledge is exclusively treated as inc idental rules. However, as few concepts are definable using only stric t knowledge, imposing strict definitions leads to terminological knowl edge bases that mostly contain partially defined concepts. This is a r eal problem because such concepts can only be inserted as leaves of th e terminology. Moreover, instance recognition is biased as these conce pts must be explicitly mentioned as properties of these instances. It follows that partially defined concepts are described with necessary b ut not sufficient conditions. As a solution to these problems, we prop ose to integrate defaults in concept definitions and we argue that thi s is essential for our diagnosis application. We introduce a descripti on language AL(delta epsilon) with default(delta) and exception(epsilo n) connectives. The cornerstone of our approach is the introduction of a definitional point of view where a default can be part of a concept definition, whereas in the classical inheritance one it is only viewe d as a weak implication. we go on to describe a map between the defini tion of a concept and its inherited properties, and we show that the c ombination of these definitional and inheritance levels considerably i mproves the capabilities of classification processes. In particular th is allows us to distinguish sure from probable instances and typical f rom exceptional instances. Finally we provide a specific operation, ob ject refinement, which consists in enlarging object descriptions with exceptions in order to find additional concepts the object is an insta nce of. This operation is useful for our diagnosis application.