RE: proposal for working on the ontology language

If one has a modifiable meta level then one must be able to declare the
semantics of newly defined primitives. UML has an extensible 
meta level but is ambiguous due to it's poorly defined (natural
language) 
semantics, which hinders interoperability.

MMF tries to clarify that (it is a UML 2.0 proposal backed by IBM and
Rational) using a model-theoretic semantics, which states what models
are valid instances of any particular meta-model expressed in MML.
See http://www.cs.york.ac.uk/puml/mmf/index.html for details.

Regards,

Raphael



-----Original Message-----
From:	Peter F. Patel-Schneider
Sent:	Wed 12/12/2001 4:50 PM
To:	lobrst@mitre.org
Cc:	www-webont-wg@w3.org
Subject:	Re: proposal for working on the ontology language

Precisely.  Systems like Protege have a meta-level in the same sense
that
RDF has.  The way you create classes is to create instances of a
metaclass.
However, this is not a modifiable meta level.

A modifiable meta level would have something like what
rdfs:ConstraintProperty was supposed to do.  That is, it would be
possible
to either 1/ change the way that existing class constructs worked or 2/
add
fundamentally different class constructs by creating metaclasses.  
One (very hard) example would be to add defaults to a formalism without
them.  Another (simpler) example would be to add range constraints to a
formalism without them.

There have been a number of proposals for this sort of meta level, going
back to the 1970s.  However, I'm not aware of any (continuing) use of
these
facilities, except in object extensions to LISP-like programming
languages (e.g., CLOS).  I should have ruled this out from the start.

peter



From: Leo Obrst <lobrst@mitre.org>
Subject: Re: proposal for working on the ontology language
Date: Tue, 11 Dec 2001 18:30:48 -0500

> Peter,
> 
> Although not formalized, I am thinking of Protege (not that it's
> modifiable, you can only add consistently defined new meta-classes). I
> also think of CLOS, again as mostly an unformalized language.
> 
> "Peter F. Patel-Schneider" wrote:
> > 
> > From: Leo Obrst <lobrst@mitre.org>
> > Subject: Re: proposal for working on the ontology language
> > Date: Tue, 11 Dec 2001 18:14:42 -0500
> > 
> > [...]
> > 
> > > I also note that there is no meta level to DAML+OIL and I think
that was
> > > a conscious choice, no?, though I don't know the history of that
> > > decision. Sometimes having a modifiable meta level is a very good
thing
> > > (future language extensions, e.g.)
> > >
> > > Leo
> > 
> > I would be very interested in hearing of cases where a modifiable
meta
> > level was actually used.  I wouild be even more interested in
hearing of
> > such cases that also involved a logical formalism.
> > 
> > peter
> 
> -- 
> _____________________________________________
> Dr. Leo Obrst		The MITRE Corporation
> mailto:lobrst@mitre.org Intelligent Information
Management/Exploitation
> Voice: 703-883-6770	7515 Colshire Drive, M/S W640
> Fax: 703-883-1379       McLean, VA 22102-7508, USA