Skip to Main content Skip to Navigation

Contraintes et représentation de connaissances par objets. Application au modèle TROPES

Abstract : This study shows that the introduction of constraints into an object-based knowledge model leads to improve both its declarativeness (constraints are known to be declarative statements about relations between object slots) and its inference capabilities (both constraint consistency and constraint satisfaction allow to complete in a consistent way a knowledge base). The repercussions of the presence of constraints in such a knowledge representation system are also discussed here. Constrained slots are referred to through the notion of access which extends the classical notion of path to handle multi-valued slots (slots whose value is a set or a list of values). The representation levels encountered (concept, class, instance) define the scope of a constraint as the set of objects upon which the constraint must hold, meanwhile they also lead to the inheritance of this descriptive feature. Similarly, maintaining a certain degree of local consistency on the slot domains requires an internal management of slot types we describe too. With regard to each of the most common inference mechanisms of an object-based knowledge representation model (instantiation, classification, methods, etc.), a suitable behaviour is attached to constraints without modifying the semantics of these mechanisms. The principles stated here have been validated in an object-based knowledge model called Tropes. A constraint programming module, called Micro, has been designed and half-weakly coupled with Tropes. Micro meets many of the requirements proper to the domain of object-based knowledge representation. Namely, it maintains and solves dynamic Constraint Satisfaction Problems (CSP), involving numerical, boolean or multi-valued variables, with both finite and infinite associated domains. Moreover, the integration performed here allows the knowledge representation system to take advantage of both the expressive and the computing power of constraints. Indeed, constraints extend and control the semantics of quite different and advanced notions like composite objects, tasks, relations, or even filters.
Document type :
Complete list of metadata

Cited literature [90 references]  Display  Hide  Download
Contributor : Thèses Imag <>
Submitted on : Tuesday, February 24, 2004 - 11:25:17 AM
Last modification on : Thursday, February 11, 2021 - 2:56:04 PM
Long-term archiving on: : Wednesday, September 12, 2012 - 4:50:35 PM


  • HAL Id : tel-00005046, version 1




Jérôme Gensel. Contraintes et représentation de connaissances par objets. Application au modèle TROPES. Interface homme-machine [cs.HC]. Université Joseph-Fourier - Grenoble I, 1995. Français. ⟨tel-00005046⟩



Record views


Files downloads