Re: bNodes wanted

From: Dan Brickley <danbri@w3.org>
Subject: Re: bNodes wanted 
Date: Sat, 25 May 2002 13:38:39 -0400 (EDT)

> On Sat, 25 May 2002, Peter F. Patel-Schneider wrote:
> 
> > From: Dan Brickley <danbri@w3.org>
> > Subject: Re: bNodes wanted
> > Date: Sat, 25 May 2002 06:09:29 -0400 (EDT)
> >
> > > On Fri, 24 May 2002, Peter F. Patel-Schneider wrote:

[...]

> > > I guess for 1/, you'd need to be careful not to describe your container
> > > with enough information (such as a daml:UnambiguousProperty, in the
> > > simplest case) to allow others to easily say things about it afterwards.
> > > I'm not sure that's always achievable. You might have some container and
> > > not want anyone to talk about it afterwards; yet others might start
> > > describing it (in RDF+WebOnt/etc) as 'the rdf:Seq that Peter mentioned in
> > > his message of 2002-05-25'. I don't think this can be avoided. Instead,
> > > we'd want strategies to avoid believing things claimed in such a matter,
> > > perhaps?
> >
> > Well, RDF has no mechanisms for creating statements in one RDF graph use a bnode
> > from another RDF, so even being able to talk about the container, does not
> > allow one to, for example, add new elements to it.
> 
> I was careful to say RDF+WebOnt/etc this time, to avoid our disagreement
> there.

True, and I missed it.

However, there is no mechanism for such in RDF, which means that, in RDF,
using a bnode for a collection is a good way of ensuring that it is not
RDF-extended in other documents.  If the use of other formalisms is
allowed, then it may be possible to undercut the mechanisms used.

> We're not, as I said to Aaron, talking about the second document _using_
> or _describing_ a bNode in the graph from the first document. The concern
> is with a second document further describing the thing-that-the-bNode-denotes.
> In this case the container that the bNode denotes. My
> understanding is that WebOnt provides facilities (like unambig-property)
> that make such descriptions possible, based on reference-by-description.
> This is easiest if the first document ascribes lots of unambiguous
> properties to the container-denoting-bNode. But mighht be possible through
> other tricks. That-seq-container-mentioned-in-doc-xyz, or subtler
> variants on this theme.

Well, if you allow RDF+X then lots of things are possible.  However, the
argument, I thought, was about RDF, and not about some unspecified
extension or modification to RDF.

[...]

> Dan

peter

Received on Sunday, 26 May 2002 17:27:44 UTC