Mediating databases and the semantic web: a methodology for building domain ontologies from databases and existing ontologies

Abstract

Ontologies play a vital role in the Semantic Web. However, ontologies are hard to construct due to high cost in the manual development process and difficulty in reaching agreement. While there are two valuable resources can be utilized to tackle the problems: relational databases and existing ontologies. Thus this PhD research proposes a methodology for semi-automated ontology development using these two resources. Firstly, an initial ontology of the knowledge embedded in a relational database will be generated through reverse engineering, mapping techniques and data mining techniques. Secondly, the initial ontology will be improved by ontology merging and alignment techniques on existing ontologies and through semantic and linguistic analysis of thesauri and dictionaries in order to gain agreement. A mapping between the evolved final ontology and the original database will also be established. This is designed for allowing external queries to be made on the database instances via the final ontology model.