Skip to Main content Skip to Navigation

AROMA : une méthode pour la découverte d'alignements orientés entre ontologies à partir de règles d'association

Abstract : This thesis deals with Knowledge Engineering and Knowledge Discovery in Databases (KDD). More precisely, by using the association rule model, we propose a new matching method designed to match ontologies provided with textual data (i.e. thesaurus, web directories, catalogues etc.).

In the literature, most ontology or schema matching approaches rely on similarity measures and, consequently their vast majority is restricted to finding equivalence relations only. In this context, we propose to use the asymmetric nature of the association rule model, of interestingness measures, and of the implicative statistical analysis in order to overcome the restrictions of only-similarity based approaches. The main contribution of this thesis is the introduction of an extensional and asymmetric matching method based on the discovery of significant implication rules between two textual hierarchies.

Our method follows a three-step KDD process: First, the pre-processing step reindexes ontologies on a common set of terms extracted from textual data; Next, the association rule discovery aims at finding a set of implications between hierarchies; And finally, the post-processing step allows to provide consistant and minimal (non-redundant) alignments.

The other four contributions of this thesis are : (1) an extended model of alignment dealing with implication. We define the notions of the closure and the minimal cover of an alignment so as formalize its redundancy and consistancy. We also discuss the symmetricity and cardinality of alignements. (2) the implementations of AROMA and AROMAViz supporting the validation of alignements. (3) an extension of a semantic evaluation model taking the implications into account. (4) the study of the efficiency and the behaviour of AROMA obtained on several benchmarks (web directories, catalogues and OWL ontologies) with the use of a selection of six interestingness measures.

The obtained results are promising because they underly the complementarity of our approach with existing ones.
Document type :
Complete list of metadatas

Cited literature [91 references]  Display  Hide  Download
Contributor : Jérôme David <>
Submitted on : Thursday, December 20, 2007 - 11:07:48 AM
Last modification on : Monday, October 19, 2020 - 10:54:33 AM
Long-term archiving on: : Thursday, September 27, 2012 - 12:00:43 PM


  • HAL Id : tel-00200040, version 1



Jérôme David. AROMA : une méthode pour la découverte d'alignements orientés entre ontologies à partir de règles d'association. Interface homme-machine [cs.HC]. Université de Nantes, 2007. Français. ⟨tel-00200040⟩



Record views


Files downloads