Brigitte Safar

Alignment-based Partitioning of Large-scale Ontologies

Ontology alignment is an important task for information integration systems that can make different resources, described by various and heterogeneous ontologies, interoperate. However very large ontologies have been built in some domains such as medicine or agronomy and the challenge now lays in scaling up alignment techniques that often perform complex tasks. In this paper, we propose two partitioning methods which have been designed to take the alignment objective into account in the partitioning process as soon as possible. These methods transform the two ontologies to be aligned into two sets of blocks of a limited size. Furthermore, the elements of the two ontologies that might be aligned are grouped in a minimal set of blocks and the comparison is then enacted upon these blocks. Results of experiments performed by the two methods on various pairs of ontologies are promising.

Pattern-based mapping refinement

Semantic alignment between ontologies is a crucial task for information integration. There are many ongoing efforts to develop matching systems implementing various alignment techniques but it is impossible to predict what strategy is most successful for an application domain or a given pair of ontologies. Very often the quality of the results could be improved by considering the specificities of the ontologies to be aligned. In this paper, we propose a pattern-based approach implemented in the TaxoMap Framework helping an engineer to refine mappings to take into account specific conventions used in ontologies. Experiments in the topographic field within the ANR (The French National Research Agency) project GeOnto show the usefulness of such an environment both for a domain expert and an engineer, especially when the number of mappings is very large.

Domain ontology and Galois lattice structure for query refinement

We present a system for interactively querying a topical resources repository and for navigating the set of responses obtained from the repository. Our system makes use of a domain ontology for dynamically constructing a Gallois lattice. Nodes in this structure are resources clusters which are interactively examined by the user during the construction process.

Construction automatique d'adaptateurs guidée par une ontologie pour l'intégration de sources et de données XML

This paper deals with integration of XML heterogeneous information sources into an information server according to an approach combining mediation and data warehousing. A schema, or ontology, is used to access to the external sources and also to the local data. We propose a unified method based on such an ontology able to achieve the two kinds of integration. Our contribution is twofold. First, we propose techniques to automate the generation of mappings between the ontology and a new source. Second, we present an approach to automate the construction of wrappers starting from the description of the abstract content of a source and ending by data extraction. Experiments on real data in the tourism domain have been achieved. Analysis and comments of the results are given.

Representation of Ontologies for Information Integration

An information integration system provides a uniform query interface to a collection of autonomous and distributed sources, connected to each other thanks to a global mediated schema, called domain ontology. The problem addressed in the paper is how to represent such an ontology into CARIN-ALN, a formalism combining classes and rules. We focus on the choices for representing classes, properties and constraints using the characteristics of the formalism. We also propose a method in two steps for representing a domain ontology in the framework of a mediator. The first step is directed by the formalism and the functionalities of the mediator. The second step is an optimization phase guided by the way functionalities of the mediator are implemented.

N2R-part: identity link discovery using partially aligned ontologies

Thanks to the initiative of Linked Open Data, the RDF datasets that are published on the Web are more and more numerous. One active research field currently concerns the problem of finding links between entities. We focus in this paper on ontology-based data linking approaches which use linking rules based on the available schemas (or ontologies). This kind of systems assume to have beforehand a set of mappings between ontology elements. However, this set of mappings could be incomplete. We propose in this paper a data linking approach called N2R-Part. It is based on the computation of similarity scores by exploiting at the same time properties for which a mapping exists and those for which there is no mapping. We illustrate throughout an example how the exploitation of the unmapped properties improves the data linking results.

Finding Successful Queries in a Mediator Context

In this paper, we study failing queries posed to a mediator in an information integration system and expressed in the logical formalism of the information integration system PICSEL1. First, we present the notion of concept generalisation in a concept hierarchy that is used to repair failing queries. Then, we address two problems arising while rewriting a query using views. The first problem concerns queries that cannot be rewritten due to a lack of sources, the second one concerns queries that have only unsatisfiable rewritings.

Automatic Ontology Population from Product Catalogs

In this paper we present an approach for ontology population based on heterogeneous documents describing commercial products with various descriptions and diverse styles. The originality is the generation and progressive refinement of semantic annotations leading to identify the types of the products and their features whereas the initial information is very poor quality. Documents are annotated using an ontology. The annotation process is based on an initial set of known instances, this set being built from terminological elements added in the ontology. Our approach first uses semi-automated annotation techniques on a small dataset and then applies machine learning techniques in order to fully annotate the entire dataset. This work was motivated by specific application needs. Experimentations were conducted on real-world datasets in the toys domain.