Semi-automatically building ontologies from relational databases

Semantic Web applications meet a serious problem in practice: the shortage of semantic data (ontologies). Since the vast majority of data are still stored in relational databases, they are unavailable for the most next generation Web applications, such as Semantic Digital Libraries. Therefore, it becomes one of core challenges of Semantic Web whether applications can automatically retrieve semantic information from existed relational databases. This paper proposes a middle graph-based formal model language, W-graph, to help retrieve an ontology from existed relational database instances. The main idea is to execute SQL procedures to retrieve the semantic information of the database instances, transform semi-automatically the result sets into the medium model, then transform the model into the ontology complete automatically. This method not only maps schemata to the middle model, but also populates the model with data stored in databases.