W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > June 2001

cycles in subject/predicate/object? [was: call for agenda items]

From: Dan Connolly <connolly@w3.org>
Date: Thu, 07 Jun 2001 12:13:02 -0500
Message-ID: <3B1FB61E.8CC6B5D4@w3.org>
To: pat hayes <phayes@ai.uwf.edu>
CC: Jan Grant <Jan.Grant@bristol.ac.uk>, w3c-rdfcore-wg@w3.org
This business of cycles in the subject/predicate/object graph
is an issue that I have run into but never reported
"thru channels". I think cwm assumes they're acyclic.
I don't believe it's in the issues list.

Here's a test case to characterize it: is this
a legal RDF document? I'd rather it were not, but
I'm afraid that there's nothing in the text of the
specs that says it's not:

<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">

    <rdf:Description rdf:about="http://example/stuff#x">
        <rdf:subject rdf:resource="http://example/stuff#x"/>
    </rdf:Description>
</rdf:RDF>

I checked that test in at:
  http://www.w3.org/2000/10/rdf-tests/pso-cycle1.rdf

along with what I fear are the expected results:
  http://www.w3.org/2000/10/rdf-tests/pso-cycle1.n3

i.e.

    <http://example/stuff#x>    
<http://www.w3.org/1999/02/22-rdf-syntax-ns#subject>
<http://example/stuff#x> .

There's a more involved test, using a longer cycle
and all three properties at:
  http://www.w3.org/2000/10/rdf-tests/pso-cycle2.rdf
  http://www.w3.org/2000/10/rdf-tests/pso-cycle2.n3

I'd prefer that we give an axiom (perhaps in a revision
of the schema spec?) that there's a superproperty
of rdf subject, predicate, and object that is acyclic.
So from an RDF model and syntax (i.e. abstact
and concrete syntax) point of view, that's a perfectly
good RDF document. But it is inconsistent with axioms
that we state somewhere in one of our specs.


By way of welcome to Pat, the .nt files are in a format
called n-triples, for which the working specification
is a message I sent 30 May
  http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001May/0264.html
ratified in our 1 Jun telcon
  http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Jun/0008.html


Let's see if I can hack something that
spits out n-triples in KIF format... I'll follow up on that
presently if I can finish it without getting distracted...


pat hayes wrote:
> 
> >On Thu, 7 Jun 2001 jos.deroo.jd@belgium.agfa.com wrote:
> >
> > >
> > > Brian wrote:
> > > > If you'd like time scheduled at this week's telecon, please let
> >me know by
> > > > 12 noon, UK time Thursday.
> > >
> > > 1. seeing the bag and reification (I always have to retype that word)
> > > stuff coming closer and closer, I would be glad (but don't insist)
> > > to 'hear' some feedback on the proposal(s) --
> > > http://lists.w3.org/Archives/Public/www-rdf-logic/2001Jun/0125.html
> >
> >I'd like to know if S
> >
> >       S = ( jan -[said]-> S )
> >
> >is a statement?
> 
> I hope not. (What was it that jan said, again, exactly? "I say this!"
> might be about right, I guess.)
> 
> >LISP needs setcdr! to do this sort of thing* as I
> >recall; it doesn't support a side-effect-free way of doing stuff like
> >this.
> 
> Right. LISP recursions never construct pointer loops unless they are
> forced to. But you do the same: giving S a name and use assignment
> (your '=' is stipulative, right?)  pretty much is using setq.
[...]

-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/
Received on Thursday, 7 June 2001 13:13:20 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:37:04 EDT