Re: comments on issue 5.19 (classes as instances) and 4.6 (equivalentTo)

From: pat hayes <phayes@ai.uwf.edu>
Subject: Re: comments on issue 5.19 (classes as instances) and 4.6   (equivalentTo)
Date: Thu, 15 Aug 2002 13:01:39 -0700

> >From: "Jeremy Carroll" <jjc@hplb.hpl.hp.com>
> >Subject: Re: comments on issue 5.19 (classes as instances) and 4.6 
> >(equivalentTo)
> >Date: Tue, 30 Jul 2002 10:01:06 +0200
> >
> >[...]
> >
> >>  Is there anything fundamentally wrong with the approach taken in the RDF
> >  > model theory?
> >
> >You mean aside from paradoxes?
> 
> There are no paradoxes in RDFS, and such paradoxes that have arisen 
> in DAML+OWL layering are due to inappropriate layering strategies, 
> not the RDFS model theory. 

Agreed, but these inappropriate layering strategies are still on the table.

> Please stop implying that there is 
> something paradoxical about nonwellfoundedness; that is a 
> foundational issue which was resolved about 15 years ago. There is a 
> relative consistency proof for NWFST against ZF.

Agreed.  I misspoke in my message.  

The problems occur when the RDF vision, by which I mean
	KBs are finite collections of n-triples.
	All finite collection of n-triples are KBs.
	Every resource in a KB denotes an element of the domain of discourse.
	Every n-triple in a KB requires the presence of a relationship in
	an interpretretation
is applied to formalisms with sufficient expressive power. 

> Pat

peter