W3C home > Mailing lists > Public > public-owl-wg@w3.org > November 2007

Re: Question on RDFS3.0 (was Re: proposal - Fragments redux )

From: Alan Wu <alan.wu@oracle.com>
Date: Fri, 30 Nov 2007 18:21:59 -0500
Message-ID: <47509B17.5020503@oracle.com>
To: Ivan Herman <ivan@w3.org>
CC: Jim Hendler <hendler@cs.rpi.edu>, OWL Working Group WG <public-owl-wg@w3.org>

Ivan,

I am speaking for myself.

Ivan Herman wrote:

>Alan, Jim
>
>(I changed the subject line and removed a direct reference to the ISSUE
>to untangle the thread...)
>
>I have a technical question/clarification on
>
>http://www.w3.org/2007/OWL/wiki/Fragments
>
>As far as I know (but people who are much more familiar with the
>reasoning algorithms than I am can prove me wrong) one of the issues
>with OWL-Full is that OWL-Full allows statements on 'itself', so to say.
>Ie, I can, in OWL-Full, make a statements on the core vocabulary of the
>language itself (I can say that rdf:type is functional, for example).
>Among other aspects, OWL-DL makes a strict separation of the core terms
>and does not allow any statement changing their semantics (well, they do
>not make sense in DL terminology).
>
>I was wondering whether RDFS3.0 cannot include the same sort of
>restrictions. Such a variant of OWL might be even easier to handle and
>implement without loosing real functionality. Have you guys met any
>applications in your practice that would exploit this self-definition
>facilities?
>  
>
I personally haven't encountered any application that redefines/restates 
the core vocabulary.
Oracle does not encourage this kind of practice to customers because it 
could be confusing.
On the other hand, Oracle does not prohibit it because we want our 
inference system be
applicable to as many applications as possible.

There are cases that people want to use owl:sameAs on classes so that 
assertions
regarding those classes (defined from separate domains) can be merged. 
The goal in that
particular case is to simplify query writing. This is different from 
what you asked though.

>Again, more savy people might prove me wrong in that this is not an
>essential problems for, say, an RDFS3.0 reasoner; in which case this
>mail is just an extra noise on the mailing list:-)
>  
>
This is a very interesting point you brought up. Essentially do we want 
to put
some restrictions on RDFS 3.0 similar to those on OWL DL compared to OWL 
FULL?
Right now, OWLPrime does not enforce such a restriction during forward 
chaining
so that more flexibility is provided for customers. Another motivation 
for Oracle to
take such an approach was:
more restrictions => more checking at run time => slower performance.

I am leaning towards no (or less) restrictions.

>Thanks
>
>Ivan
>  
>
Thanks,

Zhe
Received on Friday, 30 November 2007 23:25:28 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 January 2008 14:13:27 GMT