Converting relational database to ontology in OWL

Abstract

As an extention of the current Web in representing Web data, the Semantic Web provides means to share well-defined meaning of terms using semantically annotated information. In the current Web, most of the large-scale Web applications generate Web contents dynamically at the time of user``s query requests out of underlying relational databases. To represent and process the relational data in the Semantic Web environment, the relational data require to be transformed into the ontology form. In this thesis, we propose a Semantic Web technique to automatically converting relational databases into ontology expressed in OWL by inspecting the catalog information of relational database schema and the corresponding ontology information. We also develop an automatic conversion tool, called RDB2OWL based on a set of mapping rules far not only the schema information of the source database but also its instances to create both ontology and ontological individuals. By automating the conversion of the relational schema and instances, the ontology developers can only add semantic knowledge of the application processes on top of the ontology schema frame. The proposed technique can be applied to the integration of heterogeneous database resources in the level of ontology and, therefore, provide Web users with Semantic Web services. The service architecture consists of a global ontology and local ontologies. The global ontology defines and represents the structure or the meaningful contents that is defined in the local ontologies. The local ontologies, on the other hand, describe their respective underlying database schema and instances in order to allow database schema-based services as well as semantic-based services. In the course of the service scenario, the RDB2OWL gives an automatic generation of ontology bases out of the local databases. We show an evaluation of completeness of the relational schema-to-ontology mapping principles.