W3C home > Mailing lists > Public > www-rdf-interest@w3.org > October 2001

Re: a new way of thinking about RDF and RDF Schema

From: Peter F. Patel-Schneider <pfps@research.bell-labs.com>
Date: Mon, 22 Oct 2001 22:59:58 -0400
To: tpassin@home.com
Cc: www-rdf-interest@w3.org
Message-Id: <20011022225958A.pfps@research.bell-labs.com>
From: "Thomas B. Passin" <tpassin@home.com>
Subject: Re: a new way of thinking about RDF and RDF Schema
Date: Mon, 22 Oct 2001 21:55:38 -0400

> [Peter F. Patel-Schneider]
> >
> > Ahh.  But suppose that you wanted to assign (RDF-ish) meaning to every XML
> > document?  I haven't seen any schems that can do so.
> Well, you can do so using the edge-labeling approach I posted about last
> week or so.  You get an edge labeled graph, and that ought to be "RDF-ish"
> enough.  In this approach, every element name becomes an edge label (i.e., a
> property) and there are lots of anonymous nodes - in fact they are all
> anonymous.  You can only know them by their properties, except that text
> nodes contain strings - they could be literals, in other words.

Yes your approach works.  (Or at least it works about as well as the
node-labeling approach I am trying.)

I should probably asked for schemes that assign meaning to every XML
document and do ``the right thing'' for the RDF attributes (rdf:ID,
rdf:about, ...).

> Nothing else is really possible in an automated, uniform way, because with
> no semantic information you can't know the intent of the original creator.
> Other approaches are possible but they still have to make assumptions or
> leave off information (analgous to anonymous nodes).

Agreed.  (Except that, as you stated, labeling the nodes is
graph-theoretically equivalent.)

> One of these other approaches is simply to make RDF graphs for parent-child
> relationships.  Another is to use element and attribute type resources and
> describe the XML structure with brute force.  But these seem to pretty
> unsatisfying to me.

Extremely unsatisfying, I agree.

> I'd rather produce the edge-labeled graphs.  At least
> they do look RDF-ish and seem to make a reasonable amount of sense (to me,
> at any rate).

Yes, this is another possible approach, and one that probably deserves attention.

> Cheers,
> Tom P

Received on Monday, 22 October 2001 23:00:10 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 22:44:32 UTC