Re: RDFS closures, polite discourse

Dan,

I find Peter's comments useful and directly to the point.  RDFS not 
decidable.  I don't believe I've ever heard Pat claim otherwise. 

-Chris

Dr. Christopher A. Welty, Knowledge Structures Group
IBM Watson Research Center, 19 Skyline Dr., Hawthorne, NY  10532     USA   
 
Voice: +1 914.784.7055,  IBM T/L: 863.7055, Fax: +1 914.784.6912
Email: welty@us.ibm.com, Web: http://www.research.ibm.com/people/w/welty/




Dan Connolly <connolly@w3.org>
Sent by: www-webont-wg-request@w3.org
06/06/2003 12:48 PM
 
        To:     "Peter F.  "Patel-Schneider <pfps@research.bell-labs.com>
        cc:     Christopher Welty/Watson/IBM@IBMUS, www-webont-wg@w3.org
        Subject:        Re: RDFS closures, polite discourse



On Thu, 2003-06-05 at 10:38, Peter F. Patel-Schneider wrote:
> From: Dan Connolly <connolly@w3.org>
> Subject: Re: WOWG: Report from WWW 2003 - OWL presentation/issues
> Date: 05 Jun 2003 10:27:02 -0500
> 
> > 
> > On Thu, 2003-06-05 at 09:58, Christopher Welty wrote:
[...]
> > > More to the point, I believe it to be the case that RDFS is
> > > undecidable (has this been proven?)
> > 
> > on the contrary; that RDFS is decideable is so clear that
> > nobody has bothered to prove it.
> > 
> > The deductive closure of an RDFS KB is finite. You can
> > work it out with a pencil.
> 
> [...]
> 
> The deductive closure of an RDFS KB is decidedly *not* finite!

For practical purposes, such as reasonable discourse
between cooperative WG members, it is.

"it is only necessary, in practice, to add the triples which use those
container properties which actually occur in any particular graph or set
of graphs in order to check the rdfs-entailment relation between those
graphs."
  -- http://www.w3.org/TR/rdf-mt/#rdfs_entail

I stand by my claim: "that RDFS is decideable is so clear that
nobody has bothered to prove it."

Must you be hyperbolically disagreeable at every possible
opportunity, Peter? I grow weary of it.



> For example, the RDF closure of the empty RDF graph includes
> 
>                rdf:nil rdf:type _:a1 .
>                rdf:nil rdf:type _:a2 .
>                rdf:nil rdf:type _:a3 .
>                ....

I don't believe so. I checked the spec and I don't find
any justification for your claim.

> there are several other closure rules that result in infinite closures.

I don't believe so.

> In fact, the RDFS closure rules have an infinite set of axioms, 
including
> 
>                rdfs:_1 rdf:domain rdfs:Resource .
>                rdfs:_2 rdf:domain rdfs:Resource .
>                rdfs:_3 rdf:domain rdfs:Resource .
>                ...

I don't believe so. Perhaps you're thinking of

[[[
rdf:_1 rdf:type rdfs:ContainerMembershipProperty .
rdf:_2 rdf:type rdfs:ContainerMembershipProperty .
...
]]]
--  3.3 RDFS interpretations
  http://www.w3.org/TR/rdf-mt/

but you don't cite a source; I presume that you didn't actually
check a source.

Picking nits can be constructive if one actually reviews
the relevant text of specs in development in the process,
cites sources, and suggests improvements.

But I don't see anything constructive in your message
at all, Peter.

-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/