- From: Peter F. Patel-Schneider <pfps@research.bell-labs.com>
- Date: Tue, 25 Mar 2003 10:26:36 -0500 (EST)
- To: jjc@hplb.hpl.hp.com
- Cc: www-webont-wg@w3.org
From: "Jeremy Carroll" <jjc@hplb.hpl.hp.com> Subject: S&AS: rdfs compatible semantics - solipsistic! Date: Tue, 25 Mar 2003 14:14:40 +0100 > Resent-From: www-webont-wg@w3.org > Subject: S&AS: rdfs compatible semantics - solipsistic! > From: "Jeremy Carroll" <jjc@hplb.hpl.hp.com> > To: <www-webont-wg@w3.org> > Date: Tue, 25 Mar 2003 14:14:40 +0100 > X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2911.0) > Resent-Message-Id: <200303251314.h2PDEddL024817@frink.w3.org> > Importance: Normal > Resent-Sender: www-webont-wg-request@w3.org > > > > > Further reviewing my earlier notes: > > > Critical: > > 1: owl:Property is not in the vocabulary and should not appear in the > document. > Suggest that this can be substituted with IOP in most cases. Hmm. owl:Property has been in S&AS for months now. Pat introduced it as a convenience. Removing it was a bit of a pain, as IOP was defined in terms of owl:Property, but it is now done. > 2: owl:InverseFunctionalProperty, owl:TransitiveProperty, > owl:SymmetricProperty should be specified as subsets of owl:ObjectProperty. > > Otherwise we have: > > <x> rdf:type owl:DatatypeProperty . > > OWL DL entails (section 5.3.1): > > <x> rdf:type owl:TransitiveProperty . > > while this entailment violates name separation it seems unwise not to block > it. > > Suggest changing condition: > then c?CEXTI(SI(E)) iff c?IOOP?IODP and > to not include IODP (except for FunctionalProperty). Done. > Question: > > In OWL Full should there be some stronger conditions on IOP and IOOP e.g. > > IOOP = IOP > > The current document does not support "natural" OWL Full entailments like: > > <a> <p> <b> . > > does not OWL Full entail: > > <a> rdf:type _:x . > _:x rdf:type owl:Restriction . > _:x owl:onProperty <p> . > _:x owl:hasValue <b> . > > because outside owl:ObjectProperty and owl:DatatypeProperty it is > solipsistic. Yes, this needs to be changed. I believe I introduced this problem when the RDF model theory was changed to have LV <= R. I made what I thought was a benign change that made it much easier to describe OWL Full. I think that in OWL Full it suffices to say that IOOP = P<sub>I</sub>, which I have done. peter
Received on Tuesday, 25 March 2003 10:28:08 UTC