Define 'RDF' (was RE: MISC: Internet ...)

> From: Dan Connolly [mailto:connolly@w3.org] 
> On Fri, 2002-05-24 at 19:09, Peter F. Patel-Schneider wrote:
[...]
> > Yes, and this is no longer RDF.
> 
> I accept that as your opinion. I disagree.
> Argument by assertion is really no fun.

It may be no fun, but it's what tends to happen all too frequently when two
parties cannot agree on their definitions and are thus using terms in
different ways.  Is there any way we can clarify terms here?

Peter seems to be using the term 'RDF' to mean 'the RDF layer only,
characterised by the RDF M&S and the RDF Model Theory'.  I may not have that
correct - Peter?  Is that appropriate?

Dan seems to be using the term 'RDF' to mean 'the RDF layer and everything
that has ever been encoded as a triple by any agent that uses the RDF
layer'.  Again, I may not have that correct - Dan?  Is that appropriate?

Dan *may* also be using the term 'RDF' to include 'things that have been or
could be deduced by an RDF-aware agent'; I'm less sure of this.  Dan?

So the problem seems to come down to terminology, which is somehow
appropriate for this group.  The core question appears to be "is an
inference, or an inferred triple, 'RDF' if it can only be inferred by using
some inference mechanism that is not inherent to the RDF layer and its model
theory?"  Peter's and Dan's answers appear both to be valid given their
different definitions of this quintessential element: RDF itself.

A number of other discussions around this area seem to have generated more
heat than light, and it appears that at least one of the reasons is that
different participants have different ideas of what 'RDF' is.  Is it worth
trying to agree a definition of 'RDF' that we can use as the basis of
further discussion, so that we can get onto discussion by reasoned argument
rather than by assertion?

Just my $0.02 (which can be paid neither by one dollar bill nor by four
quarters).

		- Peter

Received on Monday, 27 May 2002 06:05:36 UTC