[OWL] annotations and identity

[Peter F. Patel-Schneider]

From: Matthew Pocock <matthew.pocock at ncl.ac.uk>
Subject: [OWL] annotations and identity
Date: Fri, 25 Nov 2005 18:48:15 +0000

> [THY]
> 
> I am an OWL (and RDF) newbie - can I just double-check the semantics of this 
> annotation stuff?
> 
> > Class(foo partial annotation(rdfs:label "foo"))
> > 
> > does not entail
> > 
> > Class(foo partial annotation(rdfs:label "bar"))
> 
> which of the following hold from the above statements, assuming rdf semantics:

[I mean that for logical-reasoning purposes, the thing on the left can be 
treated as equivalent to the thing on the right.]

> 1:  ("foo labelled foo" == "foo labelled bar") = true
> 2:  ("foo labelled foo" == "foo labelled bar") = false
> 3:  ("foo labelled foo" == "foo labelled bar") = unknown
> 4:  ("foo labelled foo" != "foo labelled bar") = true
> 5:  ("foo labelled foo" != "foo labelled bar") = false
> 6:  ("foo labelled foo" != "foo labelled bar") = unknown

From above,

Class(foo partial annotation(rdfs:label "foo"))

does not entail (nor is entailed by)

Class(foo partial annotation(rdfs:label "bar"))

so the two are not equivalent.

> It sounds as if case 1 does not hold and that case 5 does not hold. But, 
> nobody has said anything yet that I have understand that indicates that cases 
> 2 and 4 do hold, leaving me wondering if in fact cases 3 and 6 are infact 
> good.

I don't think that this is the right way of thinking about the issue.  All you
can ask is wether a particular entailment holds.

> If it is the case that cases 3 and 6 hold under rdf semantics, then if OWL 
> strengthens the semantics to cases 1 and 5 then we are still OK - OWL has 
> meerly strenghened what was already there. Of course, if it is case 2 and 4 
> that hold then the proposed OWL semantics break RDF semantics.

The situation here is the same for simple, RDF, RDFS, and OWL entailment.

> Or have I missed the point completely?
> 
> Matthew

peter