Alignement d'ontologies basé sur des ressources complémentaires Illustration sur le système TaxoMap

Abstract

Pour identifier des mappings entre les concepts de deux ontologies, de nombreux travaux récents portent sur l'utilisation de connaissances complémentaires dites de "background" ou de support, représentées sous la forme d'une 3 ème ontologie. Leur objectif commun est de compléter les techniques classiques d'appariement qui exploitent la structure ou la richesse du langage de représentation des ontologies, et qui ne s'appliquent plus quand les ontologies à apparier sont faiblement structurées ou se limitent à de simples taxonomies. Cet article comporte deux parties. La première présente une étude de différents travaux utilisant des connaissances de support, en commençant par leur schéma général commun, suivi par une analyse des travaux en fonction du type de connaissance de support utilisée. Une seconde partie est consacrée au système d'alignement TaxoMap. Nous présentons le système et son contexte d'utilisation. Nous décrivons ensuite l'utilisation de WordNet comme connaissance de support ainsi que les résultats d'expérimentation obtenus. ABSTRACT. A lot of alignment systems providing mappings between the concepts of two ontologies rely on the use of background knowledge, represented most of the time by a third ontology. The common objective is to complement current matching techniques which exploit structure or features represented in ontology representation languages and which fail when ontologies are only hierarchies or weakly structured models. This paper has two parts. First, we present a state-of-the-art of research work using background knowledge. A common general scheme is first introduced followed by an analysis of works that differ by the kind of background knowledge they use. The second part is dedicated to TaxoMap. We present the use context and the general architecture of the system. Then, we describe the way WordNet is exploited in TaxoMap as support knowledge together with experimental results.

Semantic Integration Process of Business Components to Support Information System Designers

“…Several works on the alignment of ontologies have emerged over recent years; most of them are based on an external resource that can be either a general ontology or domain ontology [16], [17]. Similar experiments with similar results are described in [18].…”

“…The use of textual and lexical resources, in particular the case of WordNet as knowledge support or background, what was proposed by many researchers [15]. [19] [20], [16]. An original proposal comes from [21] which analyzes the semantic resource available online.…”

“…This step consists in aligning ontologies obtained from the transformation of BC. We can use any alignment method based on targeted complementary resources, also called background ontologies or support ontologies [21,] [16], [23] and [6]. The domain ontology plays the role of targeted complementary resource and thus will be the support of ontologies alignment.…”

“…We assume that the design of an IS intended generally a business domain and that business components model fragments of this domain. Otherwise, semantic integration systems are mostly based on the alignment of ontologies; this issue has given rise to several works [16], [5] and [6]. We relied on results of these works to support semantic integration process and have proposed semantic integration process based on the alignment of ontologies using domain ontology and a method of measuring semantic similarity.…”

An Ontology-Based Method for Semantic Integration of Business Components

“…Aligning ontologies represents a great interest in application domains that manipulate heterogeneous knowledge, such as semantic web, communication in Multi-Agent Systems, data Waterhouse, schemas/ ontologies integration [14], etc. Several works on the alignment of ontologies have emerged over recent years; most of them are based on an external resource that can be either a general ontology or domain ontology [3], [14].…”