Skip to Main content Skip to Navigation
Theses

De l'optimisation à la décomposition de l'ontologique dans la logique de description

Anh Le Pham 1
1 Laboratoire d'Informatique, Signaux, et Systèmes de Sophia-Antipolis (I3S) / Equipe KEWI
Laboratoire I3S - SPARKS - Scalable and Pervasive softwARe and Knowledge Systems
Abstract : Reasoning with a large knowledge base of description logics (DLs) is a real challenge because of intractable inferences even with relatively inexpressive logic description languages. Indeed, the presence of axioms in the terminology (TBox) is one of the main reasons causing an exponential increase in the size of the search space explored by the inference algorithms. The reasoning in the description logics is essentially to test subsumption relations between concepts. Therefore, one always seeks to optimize this reasoning. The techniques for improving the performance of a DL reasoner therefore divide naturally into three levels. The first is the conceptual level considering techniques for optimizing the structure of axioms in the TBox. The second examines the algorithmic level techniques to reduce storage requirements of the tableaux algorithms and the number of subsumption tests. The third finds the strategies for optimizing the queries given to a knowledge base.In this thesis, we study an approach to ontology decomposition called the "overlap decomposition", which aims at optimizing the reasoning and methodology for ontology design. Optimization of reasoning is to find a method that sub-divides the axiom set of an ontology into a set of sub-ontologies. Each sub-ontology contains a subset of axioms of the original ontology. This allows a relative reduction of the reasoning time; methodology of the ontology design allows an ontology to be replaced by a set of ontologies in a more or less "optimal" organisation. For the first objective, the overlap decomposition of ontology results in the decomposing ontology (decomposing TBox) represented by the distributed description logic. Intuitively, implementing the reasoning algorithms simultaneously on these sub-ontologies - each having a reduced search space -could lead to a relative reduction of time. An important property of this ontology is that it is interpreted in the same domain as the original ontology. This is a basis which suggests two reasoning algorithms for the decomposing ontology. Regarding the goal of design methodology, we introduce two methods of ontology decomposition based on the heuristic graph decomposition. One is based on the minimal separators of graphs and the other is based on the normalised cut of the graph region.
Document type :
Theses
Complete list of metadatas

Cited literature [85 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00507431
Contributor : Estelle Nivault <>
Submitted on : Friday, July 30, 2010 - 12:17:51 PM
Last modification on : Wednesday, October 14, 2020 - 4:23:57 AM
Long-term archiving on: : Thursday, November 4, 2010 - 10:25:49 AM

Identifiers

  • HAL Id : tel-00507431, version 1

Collections

Citation

Anh Le Pham. De l'optimisation à la décomposition de l'ontologique dans la logique de description. Interface homme-machine [cs.HC]. Université Nice Sophia Antipolis, 2008. Français. ⟨tel-00507431⟩

Share

Metrics

Record views

488

Files downloads

1153