Re: Still no paradox (was: Re: The Peter paradox isn't.)

From: Pat Hayes <phayes@ai.uwf.edu>
Subject: Re: Still no paradox (was: Re: The Peter paradox isn't.)
Date: Mon, 25 Feb 2002 19:00:27 -0600

[...]

> >Given an OWL KB, I need to be able to determine if an object in that KB
> >satisfies a restriction that is not necessarily mentioned in the KB.
> >I would be prepared to do this somewhat indirectly, as in
> >
> >     <John, child, Joe>, <Joe, rdf:type, Person>
> >     |= <John, rdf:type, :_2>, <:2, owl:sameAs?, :_1>,
> >        <:_1, owl:onProperty, child>, <:_1, owl:hasClass, Person>
> >
> >However, I view any approach that does not come up with some way of doing
> >the above as fundamentally broken.
> 
> Then you have broken it before you even begin.
> 
> Is that turnstile supposed to be in RDF?? Then obviously that is not 
> valid. What could possibly support such a conclusion in RDF?

No, obviously it is not RDF entailment.  It is whatever entailment will
come out of a same-syntax extension.

> If it is supposed to be in OWL, then we need to be clear about which 
> parts of the conclusion are OWL syntax and which are OWL assertions. 
> Seems to me to be insane to require that OWL syntax be 
> self-describing *in OWL*, so I wouldn't expect it to be valid in OWL 
> either.

Well one of my points is that in RDF syntax everything is a triple, and has
assertional import, so in a same-syntax extension the more-complex syntax
required for OWL has to be encoded in triples in some way, which then need
to be handled correctly.

> What might be reasonable is that it is valid in what might be called 
> OWL-in-RDF, ie the special language gotten by imposing the special 
> OWL-in-RDF semantic conditions on the RDF rendering of OWL syntax; 
> but then it would simply be the inference of a piece of OWL from a 
> synonymous piece of RDF.
> 
> It seems to me that you are asking the 'layering' of OWL onto RDF to 
> be impossibly ambitious: you want it to be an *implementation* of the 
> syntax of OWL in RDF. If you want to impose that kind of condition, 
> then ALL layering is broken, because RDF (or even FOL, for that 
> matter) doesn't have the semantic moxie to support recursive 
> evaluation processes. Moral: don't impose impossible conditions.

Well, there are lots of layering possibilities, but it seems to me that
there were lots of people in the working group that wanted OWL to layer on
top of RDF in *exactly* the same way that RDFS layers on top of RDF.  I
tried to do just that and ran into paradoxes.

> Pat

By the way, I'm more-than-willing to use another layering relationship
between RDF(S) and OWL, as I believe that I have indicated many times in
the recent past.  

peter

Received on Tuesday, 26 February 2002 13:57:45 UTC