RE: RDF for metadata

> -----Original Message-----
> From: ext Peter F. Patel-Schneider 
> [mailto:pfps@research.bell-labs.com]
> Sent: 15 October, 2001 19:32
> To: drew.mcdermott@yale.edu
> Cc: www-rdf-logic@w3.org
> Subject: Re: RDF for metadata
> 
> 
> From: Drew McDermott <drew.mcdermott@yale.edu>
> Subject: RDF for metadata
> Date: Mon, 15 Oct 2001 11:04:20 -0400 (EDT)
> 
> > 
> > One suggested role for RDF is as metadata for XML.  (The "semantics"
> > of XML as some have imprecisely said.)  The idea is that RDF (or
> > RDF+rules, or RDF+DAML+OIL) would be used to make assertions that
> > constrain the meaning of XML tags.  E.g., if there is a 'family' tag
> > in some XML vocabulary, so we could write
> > 
> >    <family>
> >       <father><person> ...</person></father>
> >       <mother><person> ...</person></mother>
> >       <children>
> >           <person> ...</person>
> >           <person> ...</person>
> >       </children>
> >    </family>
> > 
> > then we could give constraints on these symbols.  (No child has more
> > than one mother, or the like.)
> > 
> > Has anyone actully done this for any existing XML vocabulary?  Are
> > there pointers to such efforts?  I'm having trouble visualizing
> > exactly how RDF would do this job.
> > 
> >                                              -- Drew McDermott
> 
> I'm not aware of any.  The problem is that (most) XML is not 
> valid RDF.

There also comes to mind the fact that the scope of an XML content
model is the instance, yet the scope of an RDF "content" model is
a knowledge base built from possibly multiple sources (XML instances).

If one is able to encapsulate all knowledge in a single XML
instance, then it may make sense to be able to define constraints
such as "no child has more than one mother" since you have
a "closed world" defined by the single XML instance.

However, in the case of a knowledge base built from the syndication
of multiple sources, each defined by a separate instance, it doesn't
make sense to define such constraints at the XML instance scope
(irregardless of whether one uses a DTD, XML Schema, RELAX, RDF, etc.)
since two instances could each define only one mother, but a 
different person, and such a conflict could never be caught until
one syndicated those separate instances into a common knowledge base
and applied schema constraints on the syndicated knowledge base.

So, if you are looking to define in RDF structural constraints on 
individual XML instances (i.e. DTDs in RDF), cool, but your example 
suggests that you are looking to define constraints on knowledge which 
transcends the individual XML instance scope.

Cheers,

Patrick

--
Patrick Stickler                      Phone:  +358 3 356 0209
Senior Research Scientist             Mobile: +358 50 483 9453
Nokia Research Center                 Fax:    +358 7180 35409
Visiokatu 1, 33720 Tampere, Finland   Email:  patrick.stickler@nokia.com

Received on Wednesday, 17 October 2001 15:52:17 UTC