W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > October 2002

Re: draft question: option C

From: Graham Klyne <Graham.Klyne@MIMEsweeper.com>
Date: Thu, 17 Oct 2002 15:38:05 +0100
Message-Id: <5.1.0.14.2.20021017151056.03852500@127.0.0.1>
To: Dan Connolly <connolly@w3.org>
Cc: Brian McBride <bwm@hplb.hpl.hp.com>, Pat Hayes <phayes@ai.uwf.edu>, w3c-rdfcore-wg@w3.org

Dan,

It is my recollection that the idea of datatype URIs being both classes and 
properties has been floating around for some time [*], and this is the 
first time I've been aware of dissent with it (not counting the message to 
rdf-logic as a record of dissent).

I think datatype as class and properties represent current WG thinking, so 
if there is a problem with this I think we should hear what that problem is.


[*] a recent reference is: 
http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Sep/0125.html

and from the datatyping draft that was agreed would be the basis for work 
going forward:
[[
4. RDF Datatyping Model Theory

The RDF Model Theory explains the fundamental model-theoretic concepts like 
interpretation, universe, extension etc. used for interpreting the 
semantics of RDF graphs. This section assumes familiarity with these basic 
concepts.

Suppose I is an RDF interpretation of a graph E. Then I is datatyped (with 
respect to a set D of datatypes) if the following is true for any datatype 
URI Reference ddd (with I(ddd) in D):

(1) ICEXT(I(ddd)) = {x : <x,y> in IEXT(I(ddd))}

I.e. the class extension of the datatype class is the value space of the 
datatype.

(2) For any typed literal ddd"LLL",   I(ddd"LLL") = L2V(I(ddd))("LLL")

I.e. the typed literal node denotes the datatype value having the lexical 
representation "LLL" according to the lexical to value mapping defined for 
the datatype ddd
]]

Condition (1) is quite explicit about this.
The same appeared in some of Pat's earlier documents on this subject.

#g
--

At 08:38 PM 10/16/02 -0500, Dan Connolly wrote:

>On Wed, 2002-10-16 at 14:16, Brian McBride wrote:
> > At 10:25 15/10/2002 -0500, pat hayes wrote:
> >
> > [...]
> >
> > >>I want to be sure that whatever spec we come up with,
> > >>I can continue to use the datatype property idiom...
> > >>     <k:Thursday r:about="#_thu10">
> > >>         <dt:date>2002-10-10</dt:date>
> > >>     </k:Thursday>
> > >>         -- http://www.w3.org/2002/10dc-uk/itin3.rdf
> > >>
> > >>So far, our (published WD) specs have been consistent
> > >>with a view that classes and properties are disjoint. (In
> > >>SWAD, we use that assumption for lint-style checking.)
> > >>The 6Sep decision seems to conflict with the
> > >>use of the datatype property idioim under
> > >>the disjointness-of-properties-and-classes
> > >>assumption.
> > >
> > >I was not aware that there was any such assumption. On the contrary, in
> > >fact: the MT has been designed to allow the possibility of a class and a
> > >property being the same. If this is an assumption, maybe we should 
> reflect
> > >it formally in the language. Certainly that would make the Webont work a
> > >little simpler.
> >
> > We discussed this.  It was felt that asking users to distinguish between
> > two uri's for a datatype, one for the class and one for the property would
> > be unnecessarily confusing.
>
>Er... it was also felt that confusing the value space with
>the mapping is unnecessarily confusing; I'm not sure if that's
>part of the discussion you're talking about, since you
>don't give a citation, but it is a matter of record:
>
>" The proposal treats datatypes as if they were the classes which
>     conventionally have the same name (eg "integer"), but according to
>     the XMLSchema spec, they are not."
>-- http://lists.w3.org/Archives/Public/www-rdf-logic/2001Feb/0091.html
>
>
> > The notion that classes and properties were not disjoint has been around
> > for many months
>
>Which notion, exactly? I see two:
>
>         (a) the RDF specs don't guarantee that
>         classes are disjoint from properties
>
>         (b) the RDF specs guarantee that
>         classes and properties intersect.
>
>(a) has been around for a while; (b) has been with us only
>since 6 Sep.
>
>Even in situation (a), some folks can advocate
>"keep your properties separate from your classes"
>as a best practice (sorta like ala # vs /) while the spec
>remains silent on the issue.
>
>With the WG proposing (b), it forces those
>of us who advocate keeping properties and
>classes separate to object.
>
> > and has found general acceptance.
>
>I don't think (b) has found general acceptance.
>It hasn't found acceptance among the developers
>I work with.
>
> >  I'm not convinced that
> > SWAD's lint application is sufficiently strong justification to reopen 
> this.
>
>Well, perhaps it's not worth re-opening the issue at this
>point; we're clearly not going to make everybody happy with
>any datatypes design.
>But I couldn't let your "everybody thinks this is just fine"
>go unqualified.
>
>I'm not sure I'm going to be able to get this design thru
>last call without objections from the folks I work with.
>
>--
>Dan Connolly, W3C http://www.w3.org/People/Connolly/

-------------------
Graham Klyne
<GK@NineByNine.org>
Received on Thursday, 17 October 2002 10:34:22 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:52:27 EDT