Re: SEM: Layering bug

From: Dan Connolly <connolly@w3.org>
Subject: Re: SEM: Layering bug
Date: 01 Jan 2003 14:09:40 -0600

> On Mon, 2002-12-30 at 12:00, Jeremy Carroll wrote:
> > >> 
> > >> OWL Full:
> > >> 
> > >> Empty graph
> > >> 
> > >> owl-full entails
> > >> 
> > >> rdfs:Literal rdfs:subClassOf owl:Thing
> > 
> > DanC:
> > > How do you get that? I thought owl:Thing was
> > > disjoint from rdfs:Literal.
> > 
> > In OWL DL but not OWL Full.
> 
> That's not my understanding.
> 
> > At
> > http://www-db.research.bell-labs.com/user/pfps/owl/semantics/rdfs.html#5.4
> > 
> > we read IOT=R1
> > in other words
> > 
> > owl:Thing owl:sameClassAs rdfs:Resource .
> 
> That's broken.
> 
> owl:Thing is disjoint from rdfs:Literal and rdf:Property,
> in my understanding.
> 
> Peter/Pat, please fix/confirm/explain-why-not.

Looking at OWL as an extension to RDF we currently have the following
situation.

There are two vocabulary entailment relationships defined in OWL - OWL/DL
entailment and OWL/Full entailment.  Both of these definitions are
relationships between arbitrary RDF Graphs.  

Empty graph OWL/Full entails
rdfs:Literal rdfs:subClassOf owl:Thing .
owl:Thing owl:sameClassAs rdfs:Resource.

Empty graph OWL/DL entails
rdfs:Literal owl:disjointWith owl:Thing .

There are two subsets of RDF Graphs that are of interest for OWL - OWL/DL
ontologies in graph form (OWL/DL graphs) and OWL/Lite graphs.  OWL/DL
graphs are defined as RDF graphs that can result from the translation of
OWL abstract ontologies.  OWL/Lite graphs are not defined in the
semantics document (because they are uninteresting from its perspective).
The obvious definition of OWL/Lite graphs is graphs that can result from
the translation of OWL/Lite abstract ontologies.  All OWL/Lite graphs are
OWL/DL graphs.

For RDF graphs G1 and G2, where G1, G2, and G1uG2 are all OWL/DL graphs, G1
OWL/Full entails G2 if G1 OWL/DL entails G2.

In the above example, 
rdfs:Literal owl:disjointWith owl:Thing .
is not an OWL/DL graph, so the above theorem does not apply.

Peter F. Patel-Schneider
Bell Labs Research
Lucent Technologies

Received on Wednesday, 1 January 2003 17:23:06 UTC