2

In the W3C explanation of OWL 2 QL it is mentioned that

In particular, this profile (i.e OWL 2 QL) contains the intersection of RDFS and OWL 2 DL.

This sentence is confusing me and I was wondering if somebody can clarify it. Does this mean that OWL 2 QL is a subset of RDFS? Or does this mean that it is a more restricted version of RDFS (and therefore more axioms to express these restrictions).

Assume that I have a KB that has only the following axioms used in it: rdfs:domain, rdfs:range, rdfs:subclassOF, rdf:type, owl:ObjectProperty and owl:DatatypeProperty. What can I say about the language of this KB? can I say it is expressed in RDFS, OWL 2 QL or both?

Is that the case that if I see a resource which is both an instance of something (rdf:type), and a class I would say it is RDFS, and if there is no resource which is both an instance and a class I would say it is also in OWL 2 QL?

cygri
  • 9,412
  • 1
  • 25
  • 47
Bahar
  • 770
  • 5
  • 18

1 Answers1

5

If I say, “this box contains the intersection of my t-shirts and green things”, it doesn't mean the box contains only green t-shirts. All my green t-shirts are in the box, but it may contain also blue t-shirts or green socks or whatever else.

In particular, this profile (i.e OWL 2 QL) contains the intersection of RDFS and OWL 2 DL.

The sentence just says: Any KB that meets all the requirements of RDFS and OWL 2 DL happens to also be in OWL 2 QL. But as far as that sentence goes, OWL 2 QL may also contain all sorts of other things that are neither RDFS or OWL 2 DL.

(I am also not sure that the sentence is strictly true. For example, RDFS allows anonymous individuals a.k.a. blank nodes. OWL 2 QL doesn't.)

Your example KB uses OWL constructs, so it is clearly not RDFS. The strict separation between classes and individuals is something from OWL 1 DL, and no longer present in OWL 2 DL.

The terms you mention (rdfs:domain, rdf:type, owl:DatatypeProperty etc.) are not strictly speaking axioms. They are part of the RDF vocabulary that is used to represent OWL axioms if the ontology is represented as an RDF graphs. But the corresponding axioms are DataPropertyDomain, ClassAssertion and DataProperty. When one gets serious about OWL reasoning and the various OWL profiles, it is better to forget about RDF triples for a moment, and to think in terms of actual OWL axioms. For example, the grammar for OWL 2 QL defines exactly what is and is not allowed, but uses the language of OWL axioms and not the language of RDF triples.

cygri
  • 9,412
  • 1
  • 25
  • 47