W3C home > Mailing lists > Public > www-webont-wg@w3.org > April 2002

RE: ACTION: task force unasserted triples

From: Pat Hayes <phayes@ai.uwf.edu>
Date: Wed, 24 Apr 2002 13:13:30 -0500
Message-Id: <p05101532b8ec9e04fa12@[65.217.30.94]>
To: "Massimo Marchiori" <massimo@w3.org>
Cc: www-webont-wg@w3.org
>I was going to reply at length to [1], but then [2] came
>out (thanks Jeremy!), so I'll just quickly throw this synthesis in, and
>await
>(busy day today with Query's publications, blub...).
>Specific question to all the DTTFers and others:
>Fact: [2] just came out.
>So, Pat, Peter, is this really the issue?

Im not sure, see my reply to Jeremy. (That is, Im not sure what 
exactly Jeremy is saying.)

>  If so...
>Reply: welcome to the stratification proposal [3]

And I don't know what that proposal amounts to, or how it solves the problem.

>, that doesn't require
>RDFCore to do anything...
>
>So, putting my W3C process hat on:
>Can somebody please clearly address the point in [4]?

I presume you mean this:
"The first primary goal of the task force should be just to determine 
whether there is some problem with RDF, or it is instead the case 
that these are just our (WOWG) problems."
and
"let's focus on the "why in earth RDFCore should do some modification 
to the RDF model?" problem."

OK, let me add my 2c worth to this issue. First, if the Core WG wants 
to be stubborn, and I have to concede that it has every moral right 
to be, then I think that this could be considered to be not RDFs 
problem at all. That is, RDF is a simple content language with a 
clear syntax and semantics, and that's all the RDF core WG is 
expected to say about it. Other *users* of RDF, such as WebOnt, might 
want to use it in ways that go outside the official RDF spec: that is 
their business, not RDF core's business. Like I say, that is a viable 
position and I think a very defensible one, and with my RDF core WG 
hat on, it would be my own preferred answer. In other words, its not 
RDF-core's business to provide WebOnt with a way of ignoring part of 
the RDF spec.

HOWEVER, there are two problems with that response.

First, that would put WebOnt into the position of having to concede 
that OWLs use of RDF does in some way (that WebOnt will make clear, 
presumably) go beyond, or deviate from, the account of RDF given by 
the official RDF spec. It isn't much of a deviation, and indeed it 
might not be an actual deviation at all (depending on quite now 
strictly worded the RDF spec turns out to be), just a kind of 
declaration or admission that parts of OWL-in-RDF need to be 
protected from unlimited inferences made by an RDF engine. I have no 
problem with this, but there are those who do.

The second problem is based on the perception (which is widely, but 
not universally, held) that this isn't a isolated OWL problem, but in 
fact a more generic, structural problem with almost any extension to 
RDF; and since RDF is not being touted as a limited-use mechanism, 
but as some kind of very general-purpose foundational architecture, 
that this structural problem really is RDF's business. On this view, 
it is up to RDF to put its house in order, and the current 
discussions about how OWL might solve the problem by limiting its 
set-theoretic expressiveness, or whatever, are kind of missing the 
central point, which is that there is something basically broken 
about RDF's very limited view of every triple being asserted, and the 
only thing that one can do with an RDF graph is to assert it.

Still, even if one believes the second point (which I do), one could 
still take the position that this kind of a change is out of scope 
for the current RDF core WG, and should be postponed for a later WG 
to discuss, as many 'tricky' issues in RDF have been already; the 
charter of the core WG is severely limited and many people feel it 
has overstepped it already. The counter-point to this position is 
that this particular issue needs to be resolved by WebOnt in the very 
near future, so if it is left to WebOnt then there is a risk of rival 
WGs producing different solutions, many of them too late.  There is 
also the observation I would add personally, that I find it 
incredible that such a large number of very smart people can waste so 
much time over a tiny technical issue which is in any case 
dumb-as-dirt obvious (Asserting an expression does not always assert 
all its sub-expressions. If we have no way to indicate 
sub-expressions, we have a problem. Solution: invent some way to not 
assert sub-expressions.)

That was a comment, not a proof [5].

>Putting my mathematician's hat on:
>Can somebody please give me some solid mathematical argumentation wrt the
>point in [4]?

Which point was that again?

Pat

>All I've seen so far are problems of OWL, and the link to RDFCore has been
>proved
>using instances of [5] (up to you to figure out what instances... ;)
>
>I do have some answers, and I think I get where Peter and Pat want to go,
>and also for
>what deep reasons there are so much trouble to address [4], but before
>losing time trying to interprete
>their thought and wasting everybody's time, I'd like to get some more
>detailed answer.
>
>Please give me that sense of relief Pat was talking about in [1]....!!! :)
>-M
>
>[1] http://lists.w3.org/Archives/Public/www-webont-wg/2002Apr/0253.html
>[2] http://lists.w3.org/Archives/Public/www-webont-wg/2002Apr/0269.html
>[3] http://lists.w3.org/Archives/Public/www-webont-wg/2002Apr/0151.html
>[4] http://lists.w3.org/Archives/Public/www-webont-wg/2002Apr/0251.html
>[5] http://helios.unive.it/~franz/proofs.html


-- 
---------------------------------------------------------------------
IHMC					(850)434 8903   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola,  FL 32501			(850)202 4440   fax
phayes@ai.uwf.edu 
http://www.coginst.uwf.edu/~phayes
Received on Wednesday, 24 April 2002 14:13:36 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:57:49 GMT