W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > May 2003

Re: Possible movement from RDFS to RDF (was: Re: ACTION 20030425#4 [was Re: timbl-03])

From: Jos De_Roo <jos.deroo@agfa.com>
Date: Tue, 6 May 2003 21:52:28 +0200
To: "pat hayes <phayes" <phayes@ai.uwf.edu>
Cc: fmanola@mitre.org, w3c-rdfcore-wg@w3.org, w3c-rdfcore-wg-request@w3.org
Message-ID: <OFD46445DE.E69F5A06-ONC1256D1E.006D000C-C1256D1E.006D2C98@agfa.be>


-- ,
Jos De Roo, AGFA http://www.agfa.com/w3c/jdroo/

                    pat hayes                                                                                           
                    <phayes@ai.uwf.edu>       To:     fmanola@mitre.org                                                 
                    Sent by:                  cc:     w3c-rdfcore-wg@w3.org                                             
                    w3c-rdfcore-wg-requ       Subject:     Possible movement from RDFS to RDF (was: Re: ACTION          
                    est@w3.org                 20030425#4  [was Re: timbl-03])                                          
                    2003-05-05 10:57 PM                                                                                 

>Yes;  as noted in an earlier interaction with Brian, this also applies
>to RDFS.  As I said, the Primer would have to cover this point in some
>way (given that at the point where the Primer talks about Collections,
>it hasn't gotten to Schema yet).

Right. I felt a similar pressure, which is why there is a text
comment in the RDF section saying that RDFS will add some more
constraints to the RDF vocabulary.

The relationship between RDF and RDFS entailment is kind of sneaky,
in fact. I put into the 'pure' RDF semantics all those conditions
that could be stated 'directly' in terms of triples using just the
pure  RDF namespace.  But this 'directly' is a delicate distinction,
since one could state this condition without actually mentioning
rdfs:domain, eg by imposing a semantic condition connecting
rdf:first, rdf:type and rdf:List, without mentioning anything in the
RDFS namespace at all

if <x,y> in IEXT(I(rdf:first)) then <x, I(rdf:List)> in IEXT(I(rdf:type)).

and similarly, by the way:

if <x,y> in IEXT(I(rdf:next)) then <x, I(rdf:List)> in IEXT(I(rdf:type)).
if <x,y> in IEXT(I(rdf:next)) then <y, I(rdf:List)> in IEXT(I(rdf:type)).
if <x,y> in IEXT(I(rdf:subject)) then <x, I(rdf:Statement)> in
if <x,y> in IEXT(I(rdf:predicate)) then <x, I(rdf:Statement)> in
if <x,y> in IEXT(I(rdf:object)) then <x, I(rdf:Statement)> in

Its just that this is all so much easier to say directly once the
rdfs:domain/range is introduced, that it seemed kind of silly to
write it into RDF, even though one could reasonably claim that these
are all RDF rather than RDFS conditions, since they apply to the RDF
vocabulary and can be stated as restrictions on it.  However, there
isn't any reason why you need to be so fussy in the Primer, seems to
me (??): you could just talk about things being in domains and ranges
before formally introducing rdf:domain and rdfs:range. This would
then allow you to introduce the idea when talking about RDF, then
refer back to it to 'explain' what rdfs:domain and rdfs:range mean.

Or, if folk prefer, I could add these conditions to the RDF
semantics, stated in the above style as semantic conditions, and
supply appropriate RDF closure rules (which would of course be
redundant in RDFS, but that's OK)

Any votes for this? It occurs to me that it might also help de-fuse
some of the howls about reification and collections having 'no

It would be quick to do at this stage, so don't feel that it would be
a great slow-up or anything; and I could claim it as editorial since
it really is only a matter of presentation, and nobody outside this
WG is going to give a damn about the finer points of the RDF/RDFS
distinction, I would guess.  If I hear two yeses and no noes, I'll
put it into a revised draft :-)


>Graham Klyne wrote:
>>  At 09:20 30/04/2003 -0400, Frank Manola wrote:
>>  >Or rather, whether this is true *in RDF* (as opposed to in OWL).  The
>>  >point is that, as I read the Semantics document, the only semantic
>>  >condition imposed on the collection vocabulary is that the type of
>>  >must be rdf:List.  This, of course, doesn't apply to the subject of an
>>  >rdf:first, so the inference Tim wants drawn would seem to be a
>>  >extension which might be true for OWL, but not necessarily for RDF per
>>  >se.  It seems to me the explanation would have to cover this
>>point in some way.
>>  In the semantics doc, there's also:
>>  [[
>>  rdf:first rdfs:domain rdf:List .
>>  rdf:rest rdfs:domain rdf:List .
>>  ]]
>>  -- http://www.w3.org/2001/sw/RDFCore/TR/WD-rdf-mt-20030117/#rdfs_interp
>>  from which the rest flows through RDFS entailments.
>>  #g
>>  -------------------
>>  Graham Klyne
>>  <GK@NineByNine.org>
>>  PGP: 0FAA 69FF C083 000B A2E9  A131 01B9 1C7A DBCA CB5E
>Frank Manola                   The MITRE Corporation
>202 Burlington Road, MS A345   Bedford, MA 01730-1420
>mailto:fmanola@mitre.org       voice: 781-271-8147   FAX: 781-271-8752

IHMC                                                (850)434 8903 or
(650)494 3973   home
40 South Alcaniz St.                                (850)202 4416   office
Pensacola                                           (850)202 4440   fax
FL 32501                                                 (850)291 0667
phayes@ai.uwf.edu                   http://www.coginst.uwf.edu/~phayes
s.pam@ai.uwf.edu   for spam
Received on Tuesday, 6 May 2003 15:57:42 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:57:27 EDT