RDF Validation

The RDF schemas specification allows the authors of new RDF vocabularies to set some constraints on underlying data model. Our service takes one physical RDF/XML encoded document and checks its validity with respect to the RDF schemas it refers to.

The RDF validation service at W3C is based on the following procedure:

1. The RDF description is compiled to the corresponding data model i.e. to 3-tuples. Let us call this set of 3-tuples B.
2. All references to RDF schemas i.e. namespace declarations result in the fetching of these schemas and their compilation to the corresponding data model i.e. to 3-tuples. Let us call each set of 3-tuples from each schema as Si.
3. The SiLRI - an evaluation environment of F-Logic programs - is initialized with
   • facts i.e. the whole data model that consists of the union of B and Si.
   • rules which check the constraints
4. The query result will present all constraints that have been violated. The current supported constraints include
   • domain
   • range

People have suggested other criteria that should/could be verified as well. The following list presents the ideas without formal F-Logic description. If someone has the time to formalize them into programs, I will be happy to integrate them into the online service.

• non-looped inheritance chains with subPropertyOf and subClassOf
• missing predicate definitions
• 'type' and 'subClassOf' cannot exist for the same node

Thanks for advice on putting an RDF validation service up go to Ralph Swick, Eric Miller, Bert Bos, Stefan Decker, and Dan Brickley.