W3C home > Mailing lists > Public > public-rdf-wg@w3.org > August 2011

Re: RDF-ISSUE-76 (empty-lexical-space): RDF Semantics and RDF Concepts disagree on definition of datatypes

From: Antoine Zimmermann <antoine.zimmermann@gmail.com>
Date: Tue, 30 Aug 2011 22:19:58 +0200
Message-ID: <4E5D45EE.6090900@insa-lyon.fr>
To: Pat Hayes <phayes@ihmc.us>
CC: RDF Working Group WG <public-rdf-wg@w3.org>, RDF Working Group Issue Tracker <sysbot+tracker@w3.org>
Le 30/08/2011 22:07, Pat Hayes a écrit :
> I got this slightly wrong. Here is the correct version:
>
> 1. All literals with this datatype are ill-typed.
> 2. The L2V mapping of ex:empty is the empty mapping (that is, it is the empty set of pairs), which is unique.
> 3. No literal can denote any value in the value space of ex:empty.

Point 3 is not correct. For instance, owl:real has an empty lexical 
space (as pointed out by Andy) and yet, there are plenty of literals 
that can denote a real number. "2"^^xsd:integer, "1.2"^^xsd:decimal, 
"1/3"^^owl:rational all denote elements of the value space of owl:real.

>
> This still seems ridiculous to me.
>
> Pat
>
>
> On Aug 30, 2011, at 2:43 PM, Pat Hayes wrote:
>
>> I would argue that this is, rather, an omission in RDF Concepts. The idea of a datatype with an empty lexical space only barely makes sense.
>>
>> Consider a datatype, let me call it ex:empty, with an empty lexical space. Then the following can be derived from the specifications in RDF Semantics.
>>
>> 1. There are no well-formed legal literals with this datatype.
>> 2. Any graph containing ex:empty as a datatype name is inconsistent.
>> 3. Any triple of the form [foo rdf:type ex:empty] is false.
>> 4. The L2V mapping of ex:empty is the empty mapping (that is, it is the empty set of pairs), which is unique.
>> 5. Therefore, ex:empty is unique: any datatype with an empty lexical space must be identical to ex:empty.
>>
>> If, as I suggest, this is ridiculous, the nonempty condition imposed by RDF Semantics is not only reasonable, but should be required. And as RDF Semantics and RDF Concepts are equally normative, this resolution of the disagreement can be taken either way. (If it is indeed a disagreement: I would say that not speaking of emptiness does not imply the possibility of emptiness, if another part of the spec imposes this extra condition.)
>>
>> Pat
>>
>>
>> On Aug 29, 2011, at 1:12 PM, RDF Working Group Issue Tracker wrote:
>>
>>>
>>> RDF-ISSUE-76 (empty-lexical-space): RDF Semantics and RDF Concepts disagree on definition of datatypes
>>>
>>> http://www.w3.org/2011/rdf-wg/track/issues/76
>>>
>>> Raised by: Richard Cyganiak
>>> On product:
>>>
>>> RDF Semantics requires lexical space and value space of datatypes to be non-empty:
>>> http://www.w3.org/TR/rdf-mt/#dtype_interp
>>>
>>> RDF Concepts (and XSD 1.0 and 1.1) express no such restriction.
>>>
>>> As RDF Concepts seems to be the normative reference here, this should be considered a buglet in RDF Semantics.
>>>
>>> Related: RDF Semantics has a list of terms that it imports from RDF Concepts. This list should include “datatype” and related terms from the normative Datatypes section in RDF Concepts.
>>> http://www.w3.org/TR/rdf-mt/#graphsyntax
>>>
>>>
>>>
>>>
>>
>> ------------------------------------------------------------
>> IHMC                                     (850)434 8903 or (650)494 3973
>> 40 South Alcaniz St.           (850)202 4416   office
>> Pensacola                            (850)202 4440   fax
>> FL 32502                              (850)291 0667   mobile
>> phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
>>
>>
>>
>>
>>
>>
>>
>
> ------------------------------------------------------------
> IHMC                                     (850)434 8903 or (650)494 3973
> 40 South Alcaniz St.           (850)202 4416   office
> Pensacola                            (850)202 4440   fax
> FL 32502                              (850)291 0667   mobile
> phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
>
>
>
>
>
>
Received on Tuesday, 30 August 2011 20:21:41 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:25:45 GMT