A Versioning Management Model for Ontology-Based Data Warehouses

More and more integration systems use ontologies to solve the problem of semantic heterogeneities between autonomous databases. To automate the integration process, a number of these systems suppose the existence of a shared domain ontology a priori referenced by the local ontologies embedded in the various sources. When the shared ontology evolves over the time, the evolution may concern (i) the ontology level, (2) the local schema level, and/or (3) the contents of sources. Since sources are autonomous and may evolve independently, managing the evolution of the integrated system turns to an asynchronous versioning problem. In this paper, we propose an approach and a model to deal with this problem in the context of a materialized integration system. To manage the changes of contents and schemas of sources, we adapt the existing solutions proposed in traditional databases. To support ontology changes, we propose the principle of ontological continuity. It supposes that an evolution of an ontology should not make false an axiom that was previously true. This principle allows the management of each old instance using the new version of ontology. With this assumption, we propose an approach, called the floating version model, that fully automate the whole integration process. Our work is motivated by the automatic integration of catalogs of industrial components in engineering databases. It has been validated by a prototype using ECCO environment and the EXPRESS language.