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: Wed, 01 Jan 2003 22:36:44 -0500 (EST)
Message-Id: <20030101.223644.35008324.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: 01 Jan 2003 16:43:20 -0600

> On Wed, 2003-01-01 at 15:59, Peter F. Patel-Schneider wrote:
> > From: Dan Connolly <connolly@w3.org>
> > Subject: ISSUE: owl:Class name misleading; try owl:Set?
> > Date: 01 Jan 2003 14:25:21 -0600
> > 
> > > 
> > > Based on some comments about sameClassAs
> > > http://lists.w3.org/Archives/Public/public-webont-comments/2002Dec/0004.html
> > > 
> > > and chatting with a few people about the relationship
> > > between rdfs:Class and owl:Class, I have come to
> > > the conclusion that owl:Set would be a better name
> > > for what we currently call owl:Class.
> > 
> > OWL classes do not act like sets.  They have an identity distinct from
> > being a set of their instances.
> 
> I don't understand what you mean by that.

An OWL class is a resource.  It has a class extension, which is a set, but
the OWL class is not (necessarily) the set.  OWL classes thus have two
pieces, or identities.

> > > In RDFS, one can have two classes whose
> > > members are {a,b,c}, but they can be distinct;
> > > the one class might be rdfs:label'ed "first
> > > three letters" and the other might e rdfs:label'ed
> > > "my three favorite letters". RDFS classes have
> > > properties other than their extension.
> > > 
> > > In DAML+OIL, if they have the same members, they're
> > > the same class.
> > 
> > Incorrect.  Two distinct classes in DAML+OIL can have the same members.
> 
> For example?

<daml:Class rdf:ID="Foo">
  <daml:sameClassAs rdf:about="daml:Thing">
</daml:Class>

Foo and Thing are distinct DAML+OIL classes that necessarily have the same
members. 

> I thought the DAML+OIL semantics said that classes that have
> the same members are identical. In
>   http://www.daml.org/2001/03/model-theoretic-semantics
> IC maps names directly to sets of members. So two names
> that denote the same set of members denote the same thing,
> no?

No.  A DAML+OIL class is a graph node.  This is stated explicitly in the 
DAML+OIL semantics.  IC thus maps a class into its extension.  It is true
that DAML+OIL classes are not resources.

> > Two distinct classes in DAML+OIL can even necessarily have the same
> > members.
> > 
> > > Gee, to me, that sounds like a set.
> > 
> > If it was correct, then DAML+OIL classes would be more like sets.
> > 
> > > In fact, I asked for terms for sets as soon as
> > > I realized how rdfs:Class works:
> > > 
> > > vocabulary for traditional sets
> > > From: Dan Connolly (connolly@w3.org)
> > > Date: Tue, Nov 06 2001
> > > http://lists.w3.org/Archives/Public/www-rdf-comments/2001OctDec/0210.html
> > > 
> > > We already use set-theoretic terms like
> > > intersectionOf, unionOf...
> > > so I think owl:Set is a better name for what
> > > we call owl:Class. And owl:sameMembersAs
> > > is probably better than owl:sameClassAs.
> > 
> > OWL classes are not sets, so owl:Set is not a good name for them.
> > 
> > owl:sameMember[s]As might be a better name for owl:sameClassAs.
> > 
> > > These should be theorems in full owl:
> > > 
> > > 	?X owl:sameMembersAs ?Y
> > > 	<==>
> > > 	?X rdfs:subClassOf ?Y.
> > > 	?Y rdfs:subClassOf ?X.
> > 
> > This is now true for owl:sameClassAs
> > 
> > > and
> > > 
> > > 	?X rdf:type owl:Set.
> > > 	?Y rdf:type owl:Set.
> > > 	?X owl:sameMembersAs ?Y.
> > > 	==>
> > > 	?X owl:sameAs ?Y.
> > 
> > This is not true in OWL.
> 
> No? Hmm... I guess I read the semantics document
> too fast.
> 
> Jeremy/Jos, please add that to the test suite, regardless
> of whether it turns out to be an entailment test
> or a non-entailment test.
> 
> > > (we did rename equivalentTo to sameAs, didn't we?
> > > I should look that up...)
> > > 
> > > I suppose this is a new issue; I don't think I could
> > > argue that it's an editorial fix. Sorry I didn't
> > > get it in sooner.
> > 
> > I'm not sure what you are proposing here.  It seems to me that you are
> > either saying that OWL classes already act like sets, which is incorrect,
> 
> I don't understand how it's incorrect. Please explain.

See above.

> > and thus that various names should be changed, or that OWL classes should
> > act like sets, which I would oppose.
> 
> Why?

Because it would be a last-minute change to OWL.  I don't think that the
working group should be making changes to OWL from now on, except as
required to fix problems.


peter
Received on Wednesday, 1 January 2003 22:36:54 GMT

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