W3C home > Mailing lists > Public > www-rdf-interest@w3.org > December 2000

Re: Statements/Stating: a proposition

From: Bill de hÓra <dehora@acm.org>
Date: Thu, 21 Dec 2000 23:21:46 -0000
Message-ID: <008701c06ba4$d17ccec0$0986883e@dehora>
To: "ML RDF-interest" <www-rdf-interest@w3c.org>
Hash: SHA1


My second go at this (I was a bit curt the last time in hindsight,

: > : Consider a statement S which occurs in two documents,
: > : http://foo and http://bar.
: > :
: > : Let RS be a reified statement representing both S and
: > : its occurrence in http://foo.  Thus:
: > :
: > :    (occursIn, RS, http://foo)
: > :
: > : is true.

Unless I missed something, (occursIn, RS, http://foo) is false. There
are an infinite number of resources that might model S, one of which
is RS, according to my understanding of the RDFm. But I don't see how
the presence of a representation of S in a given document
automatically implies the presence of a represention of RS in that
document. That has to be an inference of sorts, or some kind of
normative behaviour allowed to processors. Otherwise, how can one be
sure which of these very many potentially modelling resources of S RS
is the representation of, or should be? If anything, should we have
an infinitely sized alt container hanging off this document
containing all the possible representations of the resources that
model S, where RS is the default? That might be mathematically sound
(I don't know), but it's not useful surely for an implementation. I
have the feeling that the set/math model of RDF such as it is, is
being intertwined with machine dependent implementations.

: > If S then your ensuing statement might
: > be true iff RS is present with S in http://foo.
: I don't see why.

It might be, if there is syntactically present an RS that models S in
the document (possibly we need a predicate noReallyOccursIn).
Normally I'd assume a closed world, and say that (occursIn, RS,
http://foo) is false unless that model of S, RS, is
syntactically/literally present. Now, it might be the case that an
RDF processor finds it convenient to automaticaly infer the presence
of RS. But I don't see that in the RDFm (I shall go and check again
though :).

: > Also, I'm not altogether sure that RS can represent both S and
: > occurence.
: Neither am I.  The message you are responding to was an attempt
: to explain why I don't think it works.

Ah. My bad.

: > Again this is ambiguous. Do you mean an occurence of S
: > within http://foo, or do you mean that RS stands for the
statement S
: > and any occurence/instance of S?
: The text I wrote does say "and its occurrence in http://foo".  Does
: that not distinguish it from "any occurrence/instance of S"?

Yes now: I didn't know what "its" was referrring to.

: I was feeling today that I at least, made some progress in
: understanding Pierre-Antoine's proposal.  The language we used
: was sufficiently precise to clarify differences in our
: conceptions of what is going on here.
: However, I feel sufficiently chastised that I'll have a go
: at a more formal approach, something I should have done a
: while ago.

I feel the same way, but too lazy to get formal (yet). I don't know
whether Pierre's proposal actively causes contradications, but I
don't see how it's generally useful either in processing terms.
Pierre, help?

- -Bill de hÓra

Version: PGP 7.0

Received on Thursday, 21 December 2000 18:24:03 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:47 GMT