Re: on media types for OWL (5.13)

On Wed, 2002-10-30 at 20:44, Peter F. Patel-Schneider wrote:
> 
> From: Jim Hendler <hendler@cs.umd.edu>
> Subject: Re: on media types for OWL (5.13)
> Date: Wed, 30 Oct 2002 20:22:02 -0500
> 
> > At 6:09 PM -0500 10/30/02, Peter F. Patel-Schneider wrote:
> 
> [...]
> 
> > >Is an agent that is validly reading the following OWL document
> > >
> > >   <rdf:RDF xmlns:rdf="...the usual...">
> > >    <rdf:Description rdf:about="http://foo.ex/bar#john">
> > >      <rdf:type>
> > >         <rdf:Description rdf:about="http://foo.ex/bar#Student">
> > >      </rdf:type>
> > >    </rdf:Description>
> > >   </rdf:RDF>
> > >
> > >allowed to respond that it does *not* entail
> > >
> > >   <rdf:RDF xmlns:rdf="...the usual..."
> > >    <rdf:list>
> > >     <rdf:first rdf:resource="http://foo.ex/bar#john" />
> > >     <rdf:rest rdf:resource="...the usual...#nil />
> > >    </rdf:list>
> > >   </rdf:RDF>
> > >
> > >
> > >[...]
> > >
> > >My belief is that there needs to be several media types to keep things like
> > >this straight.

I didn't get your point until you replied to JimH; it looked like
a trick question about the lists issue.

Now I think I understand...

> > Peter -
> >   Sorry, but this one went right over my head.  Why does the first one 
> > entail the second one?  I don't understand why it entails the list 
> > (john)? This makes it hard for me to figure out where the problem is 
> > that needs a special Mimetype.
> 
> Because, as has been mentioned multiple times, the OWL syntax has to exist
> in all interpretations, or else the entailments don't come out right, and
> a list like the one above is a part of OWL syntax (namely a part of a oneOf
> whose sole element is John).
> 
> >   Also, if the first one was not an OWL document (i.e. was app/RDF) 
> > how would it change the point you're trying to make?
> 
> Well, the whole point is that you can't tell what kind of a document it is
> just by its contents.  If you consider the first document as RDF, then the
> entailment would not follow.

I don't know what you mean by "consider a document as RDF",
nor have you been explicit about which entailment relationship
you're referring to.

If the question is re-phrased in terms we use in the specs,
it be comes clear, and it becomes orthogonal to mime types:

Does the first document simply-entail[1] the second? no.
Does it fast-owl entail the second? yes.
And so on.

[1] http://www.w3.org/TR/2002/WD-rdf-mt-20020429/#entail

> >   Afraid I need this is words of one syllable - this conversation has 
> > gotten above my limited level of competence on mime types
> 
> Well, my understanding of MIME types is also quite limited, but I believe
> that they are supposed to tell you how to interpret the bits of a
> document.  My point is that there is no way to distinguish between RDF/XML and
> OWL/RDF/XML documents, and the difference matters.

I don't see any need to distinguish between RDF/XML and
OWL/RDF/XML documents; I see only a need to distinguish
between simple-entailment, rdfs-entailment, fast-owl
entailment, large-owl, entailment, etc.


I'm not really sure we're getting anywhere beyond
my "take your pick of the 3 mime types" proposal;
you haven't convinced me that app/owl is
necessary, nor do I have any compelling argument
that app/rdf is sufficient.

In case it matters, I'd be happy to put the "take
your pick of the 3" proposal in our next WD,
but *not* close the issue until we have
a chance for community review... feedback
might eliminate or endorse some of the options.


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

Received on Wednesday, 30 October 2002 23:06:05 UTC