Re: the idea of a 'reserved' vocabulary

>First, what Graham said:
>
>On Fri, 2002-06-14 at 02:13, Graham Klyne wrote:
>[...]
>>  >Since reserving a vocabulary effects the meaning of RDF, the authority to
>>  >declare a uriref or urirefs 'reserved' in this sense rests with the
>>  >W3C.  A uriref or set of urirefs is reserved only if it is declared to be
>>  >so by a W3C Recommendation. In particular, reserving a vocabulary cannot
>>  >be done by simply asserting on a webpage that it is to be considered
>>  >reserved. There is no way to state in RDF, or any language encoded in RDF,
>>  >that a uriref is reserved, or for any RDF document to entail this as a
>>  >consequence.
>>
>>  My more substantive comment:  some folks are going to have to implement
>>  this stuff, and the above statement doesn't really help them.  Therefore, I
>>  think the spec should state up-front the form of URIs that won't be
>>  asserted.  To alleviate the issues of URI-inspection, I think we could
>>  limit the form to something like:
>>
>>       http://www.w3.org/2002/06-rdf-unasserted#<foo>
>>
>>  where values of <foo> must be documented in W3C recommendation track
>>  documents.
>
>Second, would somebody please show how this helps with layering?

Well, this has been explored exhaustively in many emails now, but 
here goes once again. Again, let me emphasize that this isnt being 
proposed as a *solution* to layering, more a way of making a solution 
possible. I don't think it is RDF core WG's job  to solve the 
layering issue completely.

>i.e. show how it relates to the example in...
>
># Layering OWL on RDF: the case for unasserted triples
>Jonathan Borden (Thu, May 30 2002)
>http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002May/0145.html

That document assumes throughout that unasserted triples are 
required, so there is no need to comment further.

>
>and/or the 5 "Indicative Statements and Inferences" in
>
>http://lists.w3.org/Archives/Public/www-archive/2002May/att-0022/01-_all.htm
>

Well, in case (1) it would obviously help if the 
daml:first/daml:rest/daml:nil vocabulary were reserved, since then 
the inference would require only that two rdf:type assertions entail 
a single daml:intersection assertion, with the daml interpretation 
from the DAML model theory; which requires only a semantic link 
between rdf:type and daml class membership, which would follow 
directly from a straightforward mapping between the model theories 
(or axiomatically if we mapped daml:class membership into predication 
in Lbase, like RDFS class membership).

Similar comments about the other cases. The point being that while 
the reserved vocabulary doesn't completely solve all the layering 
issues - for example, it says nothing about whether unnamed 
restrictions exist or not - it does clear away some of the 
undergrowth and allow those decisions to be made independently of the 
syntactic encoding of DAML/OWL in RDF. For example, in the 'subclass 
inference' example, it wouldn't require that

John rdf:type Student .
John child Joe .

entail

_:3 daml:rest <daml:nil> .

Much of the tangle surrounding layering is the need to get these 
'silly' inferences to happen in the cases where they are required, 
but not to happen in the cases where they give rise to paradoxes. If 
they are ruled out altogether, this tangle is cleared away. Once that 
is done, the layering issues are relatively straightforward; after 
all, these languages are all subsets of FOL.

>?
>
>I remain unconvinced of the utility, let alone necessity,
>of unasserted triples in any form.

Then you should take that issue up in Webont, since the Webont WG has 
specifically requested us to provide a dark-triples mechanism of some 
kind.

Pat

-- 
---------------------------------------------------------------------
IHMC					(850)322 0319   cell
40 South Alcaniz St.			(850)202 4416   office
Pensacola,  FL 32501			(850)202 4440   fax

Received on Friday, 14 June 2002 22:49:50 UTC