Research On Mapping Object-Oriented Data Model To OWL DL Ontology

As an extension of the current Web, the Semantic Web is designed to make Web contain much information, so that computers can easily understand the information to interact and collaborate between human and computers conveniently. Ontology is the formal description of a domain, and the realization of the Semantic Web relies largely on ontology construction. Therefore, ontology learning technology has become a hot topic in the Semantic Web. Currently, many researches have focused on how to constructe ontology by extracting knowledge from different knowledge sources, such as text, dictionary, ER model, and realtional data model. However, less research on ontology construction from Obeject-Oriented data models have been done. With the wide utilization of Object-Oriented data models (e.g. multimedia, CAD/CAM, database, etc.), and lots of domain knowledge and data are stored in Object-Oriented databases, constructing ontology from Object-Oriented data models will facilitate the development of ontologies.The paper focuses on how to construct OWL DL ontology from Object-Oriented data models, i.e. how to transform Object-Oriented data models into OWL DL ontology. Based on the formal definitions of Object-Oriented data models and OWL DL ontologies, the paper proposes a formal approach and an automated translation tool that can transform an Object-Oriented data model and its corresponding database instances into OWL DL ontology structure and OWL DL ontology instances, respectively.First of all, we propose the formal definitions of Object-Oriented data models and OWL DL ontologies. At the same time, semantic information that implied by the definitions is in-depth analysed. Furthermore, the feasibility that the object-oriented data model can be converted to the OWL DL ontology is proved by analysing and comparing differences and similarities between them. What’s more, a formal translation from Object-Oriented data models to OWL DL ontologies by a semantics-preserving translation algorithm is proposed, which realize the transformations from the Object-Oriented data model and its corresponding database instances to the OWL DL ontology structure and instances, where the translation of instance level is based on the translation of schema level. Of course, the translation algorithm is proved to be feasibility and correctness. Finally, based on the proposed methods, the paper designs and implements an automated translation tool, which can automatically construct ontologies by extracting the schema and instance information stored in the Object-Oriented database db4o. Case studies show that automatic ontology learning from Object-oOriented data models is feasible, and that the proposed method is semantic maintained and effective.