W3C home > Mailing lists > Public > www-webont-wg@w3.org > January 2003

Re: ISSUE: owl:Class name misleading; try owl:Set?

From: Peter F. Patel-Schneider <pfps@research.bell-labs.com>
Date: Fri, 03 Jan 2003 10:58:11 -0500 (EST)
Message-Id: <20030103.105811.104771860.pfps@research.bell-labs.com>
To: connolly@w3.org
Cc: www-webont-wg@w3.org

From: Dan Connolly <connolly@w3.org>
Subject: Re: ISSUE: owl:Class name misleading; try owl:Set?
Date: 03 Jan 2003 09:12:14 -0600

> On Fri, 2003-01-03 at 05:49, Peter F. Patel-Schneider wrote:
> 
> > > In fact, I asked that it be added to the owl schema:
> > > 	owl:Thing owl:disjointFrom owl:Class.
> > > 	owl:Thing owl:disjointFrom rdfs:Literal.
> > > 	owl:Class owl:disjointFrom rdfs:Literal.
> > > and I thought Mike or Pat said they'd do it... or at
> > > least think about it.
> > 
> > This prevents classes from being instances in OWL/Full.
> 
> It prevents owl:Classes from being owl:Things;
> It doesn't prevent owl:Classes from being rdf:Resources
> nor does it prevent rdfs:Classes from being rdf:Resources.

Correct, I was being sloppy.  

> This is by design, no? Perhaps that's not the way other
> folks understood the design, but that's what I had in mind when we
> closed the layering issue.

OWL/Full does not have this situation.  In fact, it is not possible in
OWL/Full, as 
1/ OWL/Full identifies the class extensions of owl:Thing and rdfs:Resource
   (see Section 5.4 of AS&S);
2/ rdfs:Literal is a subset of the class extension of rdfs:Resource
   (because the class extension of rdfs:Resource is the entire universe,
   from Section 3.3 of RDF Semantics);
3/ rdfs:Literal is non-empty because it contains all the untyped literals,
   including all strings.

OOPS: Point 3 is not (no longer?) true!!!  There are NO semantic
constraints on rdfs:Literal except that it denote a class (and be the range
of rdfs:comment and rdfs:label).  In fact rdfs:Literal appears to be
completely broken.  I'm composing a message to www-rdf-comments on this
that I will copy to www-webont-wg.

> In any case, that's what I'm proposing now.

This would go against the design of OWL/Full.  

[...]

[Now talking about making OWL classes much more similar to sets.]

> That it's a reasonable thing to do doesn't follow as a logical
> necessity from the fact that it's technically possible, no.
> 
> But I think it is a reasonable and useful thing to do, for
> other reasons. OWL introduces terminology such as
> intersectionOf and unionOf. As was pointed out when
> I presented some early model theory proposals, the
> traditional usage of "intersection" has exactly
> *one* intersection of any classes C1, C2, C3, ... CN.
>
> i.e.
> 	?I1 owl:intersectionOf (:C1 :C2 :C3).
> 	?I2 owl:intersectionOf (:C1 :C2 :C3).
> 	==>
> 	?I1 owl:sameAs ?I2.

Well, it is the case that intersection is an functional constructor of
sets, but this doesn't mean that the intersection of classes need work the
same way.

[...]

peter
Received on Friday, 3 January 2003 10:58:23 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:57:56 GMT