Re: LEAD: making an ontology about the ontology group...

From: "Daniel W. Connolly" <connolly@w3.org>
Subject: Re: LEAD: making an ontology about the ontology group...
Date: Fri, 9 Nov 2001 15:55:22 -0500 (EST)

> On Fri, 9 Nov 2001, Peter F. Patel-Schneider wrote:
> 
> > From: Dan Connolly <connolly@w3.org>
> > Subject: Re: LEAD: making an ontology about the ontology group...
> > Date: Thu, 08 Nov 2001 13:48:22 -0600
> >
> > > Peter F. Patel-Schneider wrote:
> > >
> > >
> > > > and thus can't have literals as their objects.  On the
> > > > modelling side, the relationship between social entities and mailboxes is
> > > > many-to-many, even when restricted to persons.
> > >
> > >
> > > No, there's only one person who's authorized to read
> > > mail sent to connolly@w3.org, and that's the case
> > > for many mailboxes; saying X contact:mailbox Y
> > > is saying that there's just one X that can receive
> > > mail sent to Y.
> >
> > Maybe that is true for some mailboxes, or maybe almost all, but I don't
> > believe that it is true for *all*.
> 
> Well, or course not. One must not assert X contact:mailbox Y
> for just any X and Y; in making that assertion, one claims
> that there's only one such X.
> 
> -- 
> Dan Connolly, W3C http://www.w3.org/People/Connolly/
> 

Ahh, now we have an interesting point that lives somewhere in that
no-man's-land between representation and language.  In
http://www.w3.org/2000/10/swap/pim/contact.n3 there is no information on
contact:mailbox, except that it is an unambiguous property from social
entities to mailboxes.

Now, and here is the language part, mailbox is a common name, and thus
elicits certain connotations in humans.  These connotations can (and
usually do) vary, but I claim that from the information here it is
certainly valid to assume that this relationship supposed to be *the*
relationship between social entities and people that indicates that mail
sent to the mailbox will somehow be made available for reading by the
social entity.  This (mailbox) relationship is not an unambiguous
relationship, which is in conflict with the information in
http://www.w3.org/2000/10/swap/pim/contact.n3.

Now Dan is certainly correct in stating that contact:mailbox is
unambiguous, and thus there is no representation problem in 
http://www.w3.org/2000/10/swap/pim/contact.n3.  However I claim that there
is a modelling error in http://www.w3.org/2000/10/swap/pim/contact.n3, and
that contact:mailbox should have a different name, or at least be related
to some more-general mailbox relationship, so that the expectations from
its name do not conflict with its formal properties.

By the way, Bill Wood's paper ``What's in a name?'' makes a lot of good
points about this and related issues.

Peter F. Patel-Schneider

PS:  Is it part of our job in this working group to produce something that
talks about the import of the URI's treated as natural-language strings?

Received on Monday, 12 November 2001 09:13:19 UTC