W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > May 2003

Re: restrictions bnodes

From: Dan Brickley <danbri@w3.org>
Date: Fri, 9 May 2003 07:22:21 -0400
To: Jan Grant <Jan.Grant@bristol.ac.uk>
Cc: Brian McBride <bwm@hplb.hpl.hp.com>, graham Klyne <gk@ninebynine.org>, RDF Core <w3c-rdfcore-wg@w3.org>
Message-ID: <20030509112220.GC21436@tux.w3.org>

* Jan Grant <Jan.Grant@bristol.ac.uk> [2003-05-09 11:54+0100]
> 
> On Fri, 9 May 2003, Brian McBride wrote:
> 
> >
> > Jan, Graham,
> >
> > A couple of questions:
> >
> > a) Were you aware of the restrictions on bnodes (individuals can't be
> > objects of > 1 triple, and no cycles of individuals) from your review of AS+S?
> 
> Yes, this is a consequence of the wording in section 4.2
> 
> 
> > b) Do you consider such restrictions significant?
> 
> They're significant, but not showstoppers, I think - at least, that's
> what I _thought_, because I was focussing on the use of RDF to express
> OWL DL. Looking at the reverse process (going from RDF graphs to the
> abstract syntax), it does appear that the idiomatic use of bnodes that
> characterise the RDFS applications I've seen thus far (I'm talking about
> grassroots stuff like FOAF) doesn't fit well here. That is, I'm coming
> around to the point of view that this is maybe more of an issue than it
> might at first appear.

If bNodes can only be used as the object of a single triple, they lose most 
of their value as a construct in the language. As does rdf:nodeID for that 
matter. 

<Image>
 <depicts>
  <Person>
    <mbox rdf:resource="mailto:danbri@w3.org"/>
  </Person>
</Image>

<Group>
 <member>
  <Person>
   <homepage rdf:resource="http://www.w3.org/People/DanBri/"/>
  </Person>
 </member>
</Group>

...is OK in OWL, but if we add in an rdf:nodeID on the two Person elements
to express that they serialize descriptions of the same (un-named) resource, 
we're in trouble? Ouch. That breaks most of my uses of RDF, and a lot of 
deployed FOAF documents.

What would an OWL friendly re-serialization of the above look like? I guess they
would just invent some URI to name the Person? (yuck...)

If the restriction is that individuals can't be the _object_ of multiple 
triples, we could use an inverse of the 'depicts' of 'member' property instead,
but that's an ugly hack. Is this restriction across all of OWL, or just the 
DL & Lite dialects?

Dan
Received on Friday, 9 May 2003 07:22:34 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:57:29 EDT