Structuration Conceptuelle et Physique de l'Espace des Objets dans un Système de Gestion des Données et des Connaissances

Knowledge Base Management Systems are not designed to manage large amounts of data in an efficient manner, contrary to DBMS of which it is the primary purpose. We present an approach that unifies KBMS and DBMS, by providing an efficient management of wide volumes of (shared) data and reasoning capabilities (instance classification). We propose a conceptual structuring of the object space that relies on logical properties used in the definition of classes, namely the Classification Space, which supports instance classification, integrity checking and object indexing through an appropriate physical representation. This representation also supports semantic query optimization and integration of heterogeneous data.