OBD:RDF and OWL

The OBD model is graph-based and is thus in a loose sense comparable with RDF. However, the RDF model is essentially binary predicates, which can lead to convoluted graph structures requiring lots of bnodes. The OBD schema abstracts some of these patterns allowing for more compact storage. For example, a single axiom in OWL stating that "all cell nuclei are part_of some cell" requires 4 triples in the conventional RDF encoding, but can be encoded in a single link in OBD. Thus OBD is more closely related to an OWL Model, with links corresponding to axioms.

OBD is primarily a biomedical annotation database. Note that "annotation" here means something different than in the OWL world. The OBD definition of an annotation is an information entity that posits one or more statements about the world. Examples could be an annotation derived from a literature record stating that P53 in human participates in DNA repair in certain contexts; or that a particular region of an image depicts a diseased lung. The statements can be about instances, types, or a mixture of both.

There are many ways of translating this annotation schema to RDF/OWL, depending on what kind of questions are to be asked. OBD provides a single way, in the form of a link that is "posited" by an annotation (a formal semantics for this will be provided). This can then be translated to RDF/OWL on a per application basis; or to other KR formalisms (for example, certain kinds of questions will require using of a non-monotonic logic or answer set programming)