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

Re: RDF Datatyping Working Draft

From: Patrick Stickler <patrick.stickler@nokia.com>
Date: Sat, 06 Apr 2002 09:12:29 +0300
To: Jeremy Carroll <jjc@hplb.hpl.hp.com>, Pat Hayes <phayes@ai.uwf.edu>
CC: RDF Core <w3c-rdfcore-wg@w3.org>
Message-ID: <B8D46C7D.1291D%patrick.stickler@nokia.com>
On 2002-04-04 14:36, "ext Jeremy Carroll" <jjc@hplb.hpl.hp.com> wrote:

> 
> Patrick,
> 
> if I have understood your comments about datatype interpretation correctly
> then I believe that:
> 
> ----
> 
> <ex:age> <rdfd:range> <xsd:integer> .
> <Jane> <ex:age> "25" .
> 
> datatype entails
> 
> <Jane> <ex:age> _:b .
> _:b <xsd:integer> "25" .
> 
> 
> ---
> 
> something that Pat seemed to shy away from saying in section 5.
> 
> This continues to worry me because then in something like webont or DAML we
> can't say that ex:age has a unique value, because <Jane> has two really
> different ages: 25 the integer (i.e. the same age as <John> and <Judy>) and
> "25" the string (i.e the same as the <bar> of <foo>).

One must take into consideration that equivalent values may have
different representations. This is true not only of datatyped values
being expressed by different idioms, but also of URI denoted values;
where e.g. one may begin with http:// and another with http:// etc.

A unique property assertion is really an assertion about the values,
which constitutes a hypothesis that all representations equate to
the same value, and that hypothesis is then testable in a context
where all variants are recongnized/understood.

To say that a property takes unique values is not, I think, to
say that those values must take an identical, global representation.

The only way that equivalence can be tested "below the line" in
RDF is if there is absolute canonicity of representation -- i.e.
all values have URI labels and all URIs are canonical, so that
equivalence is distilled by the tidying process into intersections
of identitical object nodes. Without this absolute canonicity, the
truthfullness of any set of property values for a unique property
can only be tested by extra-RDF applications.

Thus, although there is an issue here, it doesn't seem to be an issue
with the datatyping proposal. It is rather a broader issue of variant
representations and denotations of the same resource/thing and to
what extent RDF or RDFS is aware of such variation and any equivalence
between such variants.

Patrick

--
               
Patrick Stickler              Phone: +358 50 483 9453
Senior Research Scientist     Fax:   +358 7180 35409
Nokia Research Center         Email: patrick.stickler@nokia.com
Received on Saturday, 6 April 2002 01:09:55 EST

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:47:23 EDT