W3C home > Mailing lists > Public > www-rdf-logic@w3.org > March 2003

Re: SUMMARY: Treating a class as both an instance and as a class, i.e., a metaclass

From: Thomas B. Passin <tpassin@comcast.net>
Date: Wed, 12 Mar 2003 19:05:20 -0500
To: www-rdf-logic@w3.org
Message-id: <000701c2e8f4$3c578c70$6401a8c0@tbp1>

[Roger L. Costello]
>
> Thanks a lot for all of the excellent messages to my question regarding
> the nature of a class being used as both an instance and as a class
> (i.e., a metaclass).  I believe that I now have a firm understanding.
>
> I have created a few slides that summarizes the discussion:
>
>     http://www.xfront.com/metaclasses/
>
> Please let me know if I have made any mistakes.

Roger, I really appreciate the work you do to clarify and pull together
difficult subjects (like XML Schemas).  But  do not think you have quite got
there yet in this case.

When you talk about a class, there are two different kinds of things you
might want to say.  One of them amounts to __restrictions on the  properties
of its members__ - what criteria discriminate between members and
non-members.  In your Aircraft example, the wingspan of an F-16 is an
example.  OWL but not plain RDF (because the predicate is not intrinsically
defined) can describe such restrictions on properties.

Now a class is a collection, a mental construction, and as such does not
have a wingspan.  It is a kind of shorthand to say that class Aircraft has a
wingspan of 10 meters.

The class itself, however, may have been first described at such-and-such a
date by so-and-so, and it may be an member of some meta-class.  It may have
4305 instance members.  It may be defined by its intension rather than its
extension.  Thus, a class considered as an instance can have properties, but
they have a rather different nature from restrictions on the properties of
its members.  That is because the use of a class as an instance is in a
different context from its use as a collection.

In the threads of the last week or two, clear examples of OWL asserting
property restrictions to instances were presented.  I think your summary
should make clear these distinctions and illustrate with syntax appropriate
to each.  After this, I think it should discuss whether the constructions
you have shown in the current version of the presentation are proper ways to
express the shorthand for property restrictions, and whether there are any
other important shorthands (eg, are restrictions on relationships, e.g.,
transitivity, statements about the sets as collections, or are they
shorthand for statements about instances?).

Cheers,

Tom P
Received on Wednesday, 12 March 2003 19:21:23 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:52:43 GMT