Re: semantics and RDF(S)

On Fri, 12 Jan 2001, Peter F. Patel-Schneider wrote:
> > Peter,
> >
> > On Fri, 12 Jan 2001, Peter F. Patel-Schneider wrote:
> >
> > > In the process of working on the design of DAML+OIL, I have had extensive
> > > dealings with RDF and RDF Schema.  In this message I present the biggest
> > > problem that I see with RDF and RDF Schema.
> >
> > > [...] How can I build a notion of cardinality
> > > when there is no notion of equality for much of RDF and RDF Schema?
> >
> > In the process of working on and with RDF, similar issues have bothered
> > me. Most specifically, concerns surrounding our notion of identity,
> > resource naming and the connection between URIs and the things they
> > supposedly name.
> >
> > [...]
> >
> > Dan
>
> I agree that there needs to be a notion of identity for web resources, and
> that RDF may not the ideal place to discuss it.  However, there is much
> less need for this notion in other W3C forums, so the discussion may have
> to be started here.

You've put your finger on the problem. Until pretty recently, RDF/SW was
something of a minority interest. Although we're (arguably) the group
most  concerned with seeing a clean resolution to this problem, it isn't
necessarily our job. And I fear that our proposed solutions won't be
valued by others in the Web community unless we accompany them with persuasive
material (demos, code etc) to justify any re-jigging of existing specs
and implementations. As evidence (imho), the May 2000 XML-URI debate had
an  undertone of "why should we change our important mainstream XML stuff to
fit with this minority academic RDF worldview". We can expect that in
bucketloads w.r.t. DAML/SW etc unless a persuasive case is presented. I
view www-rdf-logic as the right place to begin putting together that
case, though we do of course also have the uri@w3.org list for general
URI discussions.

> The notion of what is a web resource and the role of its name (as a URI)
> is, as expected, rather complex.  The only point that I would like to make
> now is that there is no way that we can use the ``unique name assumption''
> when working with the web.  (The unique name assumption states that
> different names are names for different objects.)   The reason for this is
> that there is a many-to-one map from URIs to web resources.  (For example,
> http://w3.org and http://www.w3.org point to the same web page.)

[nods]

> My main concerns with RDF, however, live on a different level than this.
> Suppose that we nailed down the notions of identity and equality for ``web
> resources''.

(I look forward to the day; btw 'web resources' encompass things of all
kinds, ie. we mean 'name-able on the Web' not 'downloadable via...')

but assuming this, yes...

> Then there still is a problem with identity and equality for
> RDF resources, such as bags.

Of course. With "Statements", identity conditions for rdf:Statement and so
on being an obvious candidate.  There's a lot of work still to be done,
and I don't think anyone here claims otherwise. All I'm saying is that I'm
wary of attempting a cleanup of RDF's notion of statement or container
identity without an analyis of the URI/resource infrastructure that
underpins all Web data. Running before walking, castles on shifty sands
and all that.

Dan

Received on Friday, 12 January 2001 10:22:55 UTC