Ontology-Based Semantic Metadata Validation

Much of the Semantic Web content is generated from databases, especially the instance data based on the ontology classes used in applications. A recurring problem is that the instance data does not always semantically conform to the ontology used. It may be ambiguous, incomplete, or partly erroneous. Validating the data is necessary when it is transformed to a more semantic format. This may be a difficult and laborious task given the large and complex ontologies and databases in use. This paper discusses the problem of validating existing database instance data according to an ontology. We show how Semantic Web standards can augment the syntactic data validation schemes of relational databases and XML towards semantic validation. A metadata editor for semantic validation is being implemented. Its idea is to provide an XML-based view to a relational database and transform the data into RDF instances conforming to an RDF Schema, the ontology. The tool will be applied in tranforming museum collection data into a semantically interoperable form in the Finnish Museums on the Semantic Web project.