Re: bug in RDF compatible semantics for owl:AnnotationProperty

It turns out, as we planned, that this part of the OWL semantics is
subordinate to the direct model theoretic semantics for OWL DL, so there is
actually no need for any implementor to implement this, which may explan
why no one noticed.

Test case follows:

According to the RDF-compatible semantics, the following is unsatisfiable
in OWL DL:

	ex:O rdf:type owl:Ontology .
	ex:O rdfs:comment "This is a comment"^^xsd:string .

It is, however, satisfiable in both the RDF-compatible semantics for OWL
Full and the direct semantics for OWL DL.

peter



From: Dan Connolly <connolly@w3.org>
Subject: Re: bug in RDF compatible semantics for owl:AnnotationProperty
Date: Thu, 18 Nov 2004 23:14:13 -0500

> On Nov 17, 2004, at 11:08 AM, Peter F. Patel-Schneider wrote:
> > From: mei@mi.fu-berlin.de
> > Subject: maybe it is a bug for owl:AnnotationProperty
> > Date: Sun, 14 Nov 2004 09:34:34 +0100
> >
> >> Dear Peter F. Patel-Schneider,
> >>
> >> hello,
> >>
> >> i am reading your papers OWL (RDF-Compatible) MT Semantics
> >> http://www.w3.org/TR/owl-absyn/rdfs.html, and i found a very small 
> >> bug about
> >> owl:AnnotationProperty.
> >>
> >> it was mentioned as I(rdfs:comment) is in IOAP, and the subject-value 
> >> of
> >> annotation properties is in IOT (namely owl:Thing).
> >
> > [...]
> >
> >> Now, the "wine" ontology has the rdfs:comment as one of 
> >> owl:AnnotationProperty,
> >> but "wine" is an ontology not owl:Thing which should be disjoint with
> >> owl:Ontology in OWL DL interpretation. Of course, in OWL Full 
> >> interpretation,
> >> it is OK :-)
> >>
> >> maybe, Values for annotation properties could be much less 
> >> unconstrained as:
> >>  owl:AnnotationProperty EXTi(e)<=(IOTuIX)x(IOTuLVi)
> >>
> >> maybe, about the above, i made some stupid mistakes :-) if so, i am 
> >> very very
> >> sorry for bothering you!
> >>
> >> Thank you very much!
> >>
> >> Best wishes,
> >> Sincerely yours, Jing Mei
> >
> > I believe that you are correct.  I wonder how this bug managed to 
> > escape
> > attention for so long.
> 
> I suppose the distinction is not excercised in any of our tests.
> But I'm not sure I'm following exactly. If anyone finds time to
> compose a test to demonstrate the distinction, I would like
> to take a look.
> 
> > I believe that the appropriate correction is to expand the extension of
> > annotation properties even further, to include all the appropriate
> > syntactic categories.
> >
> >    EXTi(e) <= (IOT u IOC u IDC u IOOP u IODP u IOAP u IOXP u IX) x 
> > (IOT u LVi)
> >
> > Peter F. Patel-Schneider
> > Bell Labs Research
> >
> 
> -- 
> Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Friday, 19 November 2004 12:01:30 UTC