Re: mapping rule for InverseFunctionalProperty etc

On Tue, 2003-09-16 at 10:48, Peter F. Patel-Schneider wrote:
> From: Jeremy Carroll <jjc@hpl.hp.com>
> Subject: mapping rule for InverseFunctionalProperty etc
> Date: Mon, 15 Sep 2003 15:59:12 +0300
> 
> > The mappjng rule for
> > ObjectProperty( ID InverseFunctional )
> > 
> > =>
> > 
> > ID rdf:type owl:ObjectProperty . 
> >     [opt if one of the last three triples is included] 
> > ID rdf:type owl:InverseFunctionalProperty .
> > 
> > Does not appear to achieve the apparent desired effect.
> > 
> > Because while
> > 
> > eg:foo rdf:type owl:InverseFunctionalProperty .
> > 
> > is hence a legal OWL Lite ontology, the following is not
> > 
> > _:a rdf:type owl:Thing .
> > _:a eg:foo _:b .
> > _:b rdf:type owl:Thing .
> > eg:foo rdf:type owl:InverseFunctionalProperty .
> 
> Agreed.  The translation rule for individualvaluedPropertyID should be
> changed to make its first triple be optional if there is a triple in the
> translation of the ontology that give the ID one of the three
> more-specific OWL property types.
> 
> Again, this should be agreed on by the WG.
> 
> I believe that OWLP already acts in this modified manner.

Let's see if I understand this... the test-case-sketch above
shows a difference between how OWLP is implemented and how
are specs are written?

If that's the case, then yes, please, add it to the test suite
and let's get a WG decision to make the change (the alternative--
for the WG to adopt the test in such a way that confirms
the existing design and says OWLP and Jeremy are wrong--seems
silly.)

Anybody care to guestimate what changes to the text of
the other docs are needed?

-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/