0

Following this question: Difference between domain and range in rdf schema?

I have been discussing with my supervisor who told me that it is possible have several values for the range. I agree with that, but in the context of the predicate as a Literal (final value) (Not class) and the range is a datatype, it is possible various datatypes?.

I give a example: We have the predicate ".../name" so inside of name we find the property range with value xsd:String. Can i express the value "Mary" as a binary for example "1010100101" and add the datatype binary?

My argument is for final values the property range can no be associated to several values (datatypes).

Thanks in advance.

Community
  • 1
  • 1
I. Dongo
  • 29
  • 1
  • 7
  • Are you using OWL? If so, is it OWL2? – Esoteric Screen Name Jan 14 '16 at 19:24
  • Actually, we are working on RDF but we are moving to OWL. – I. Dongo Jan 15 '16 at 15:38
  • So OWL is eligible for use to answer your question? It makes a significant difference. – Esoteric Screen Name Jan 15 '16 at 16:22
  • Yes OWL, could you also explain a little bit if it's OWL2? – I. Dongo Jan 15 '16 at 16:58
  • OWL2 is a lot more robust and is what is generally meant these days. If you're moving to OWL in general, you'll be using OWL2. The only reason you'd stick to the first version is probably technical/legacy compatibility issues. [Semantic web question](http://webcache.googleusercontent.com/search?q=cache:faImQp76Q5AJ:answers.semanticweb.com/questions/14579/is-there-a-reason-to-choose-owl-over-owl2+&cd=5&hl=en&ct=clnk&gl=us). – Esoteric Screen Name Jan 15 '16 at 18:14

1 Answers1

1

It's possible to have a datatype union, or a dataOneOf restriction as range, to allow a property to have fillers from disjoint datatypes. Otherwise, it's not possible to have two axioms like:

p range xsd:int
p range xsd:string

without this implying

p subPropertyOf owl:bottomDataProperty

i.e., the property cannot appear in an assertion (the analog of an unsatisfiable class).

This is because range assertions are to be intended as AND - a reasoner infers that a filler for p belongs to all asserted ranges. This works well as long as the literal has a compatible datatype and the value fits in the value space of all range assertions; if that is not the case, it's a clash.

(This is true for OWL and OWL2 at least)

Ignazio
  • 10,504
  • 1
  • 14
  • 25