Re: UPDATE: longer version of layering document

From: Leo Obrst <lobrst@mitre.org>
Subject: Re: UPDATE: longer version of layering document
Date: Mon, 11 Feb 2002 10:59:37 -0500

> Peter and Dieter,
> 
> NIT: I don't think your example (Fig. 2) depicts what you say it should,
> in the text following the figure, i.e., "herbivore as a class, which is
> a sub-class of animal and disjunct to all carnivores."  
> But perhaps you intend it simply as a contentless template and we should
> imagine the content?
> 
> The real issue in your document is the semantic Russell-like paradox,
> no? 

Well, more or less. 

> If Peter's one suggestion below, i.e.,
>
> 1/ Move rdf:type out of the theory into the metatheory
> 
> is agreed to by the RDF/S folks, that would remove the thorniest issue,
> no?
> 
> Thanks,
> Leo

Oh yes, this would remove all of the issues!  Of course, it would be a
*drastic* change to RDF and RDFS!  Also, it would mandate a syntax
extension for OWL, as RDF would no longer have its metatheory accessible
from its theory, and thus the *syntax* of RDF could not be used to access
this metatheory.

peter

PS: As an aside, consider the way that XML Schema extends XML.  XML Schema
documents use XML syntax, but don't have the same meaning as XML documents.
This would be way of extending the syntax of RDF without extending the
syntax of RDF.  However, the RDF access to the metatheory from the theory
makes this way of extending RDF difficult.

PPS:  Of course there is nothing too wrong with RDF taken by itself.
It is just that using RDF as the basis for the semantic web is not
possible, and it is not possible because RDF appropriates all
syntax *and* all semantics.  (Well, it might be possible to have RDF as the
basis, but it would be *extraordinarily* difficult and would probably
require a complete rethink of the intended meaning of RDF.  Having RDFS as
the basis, as RDFS is now, is, however, right out of the question.)

Received on Monday, 11 February 2002 11:15:50 UTC