W3C home > Mailing lists > Public > public-rdf-wg@w3.org > March 2013

Re: reopen ISSUE-97 and reclose next week - should interpretations be relative to a vocabulary

From: Pat Hayes <phayes@ihmc.us>
Date: Thu, 7 Mar 2013 11:31:28 -0600
Cc: public-rdf-wg@w3.org
Message-Id: <94E6B204-F042-4361-8B05-D7A0F40FE263@ihmc.us>
To: Antoine Zimmermann <antoine.zimmermann@emse.fr>

On Mar 7, 2013, at 8:58 AM, Antoine Zimmermann wrote:

> Peter,
> 
> Le 06/03/2013 19:30, Peter Patel-Schneider a écrit :
>> I propose that we reopen ISSUE-97 and resolve it differently at the
>> teleconference next week.
>> 
>> ISSUE-97 concerns whether
>> 
>> ex:john ex:age "22"^^xsd:integer
>> {xsd:integer}-entails
>> ex:john ex:age "+22"^^xsd:integer
> 
> This has been decided to be solved independently of ISSUE-97.
> Thinking about it, the idea to make the mapping IL partial is good if interpretations do not depend on a vocabulary, since it does not force all literals to denote.
> 
> 
>> or the empty graph RDFS-entails
>> ex:john rdf:type rdfs:Resource
>> 
>> In the 2004 version of RDF these entailments do *not* hold, the first
>> because there are {xsd:integer}-interpretations where "22"^^xsd:integer is
>> in the vocabulary but "+22"^^xsd:integer is not.   I believe that all RDF
>> implementations do *not* work this way.  Instead they take the very
>> reasonable idea that all interpretations involved in the reasoning must
>> interpret all the vocabulary in both the LHS and the RHS.
> 
> I don't believe that all implementations make "ex:john rdf:type rdfs:Resource" true in all cases.
> 
> Consider the following. According to SPARQL 1.1 with RDFS entailment regime:
> 
> ASK WHERE { <myURI>  a  rdfs:Resource }
> 
> must return false if <myURI> is not a term in the dataset.

?? Really? That is crazy in any RDF semantics, including the 2004 semantics. All URIs denote, even if they aren't in a particular dataset.

Can you point to where this behavior is specified?

> The easiest way to implement that is to use a reasoner that implements RDF 2004 Semantics.

I don't think so. The 2004 semantics doesn't have any provision for a URI to fail to denote, and in RDFS, the universe *is* the interpretation of rdfs:Resource, so the denotation of <myURI> is in it. 

> If one uses a reasoner that implements RDF 1.1 Semantics (as currently written), one has to be careful with such cases.

I don't think its a matter of being careful, just one of getting the specs right. :-)

Pat

> 
> 
>> The semantics should be modified to make it clear that entailment should
>> work the way that everyone thinks it does.  The current editor's draft of
>> the semantics changed interpretations so that they are no longer relative
>> to a vocabulary, but instead interpret all IRIs and all well-typed literals
>> (and no ill-typed literals).  I believe that no changes are needed in any
>> other WG document, and that the change conforms to the universal
>> implementation of RDF.   This change has the desired effect and should be
>> adopted by the working group.
> 
> Fair enough, we can agree on this with a resolution next week, if everyone approves.
> 
> 
> AZ
> 
>> 
>> Peter F. Patel-Schneider
>> Nuance Communications
>> 
> 
> -- 
> Antoine Zimmermann
> ISCOD / LSTI - Institut Henri Fayol
> École Nationale Supérieure des Mines de Saint-Étienne
> 158 cours Fauriel
> 42023 Saint-Étienne Cedex 2
> France
> Tél:+33(0)4 77 42 66 03
> Fax:+33(0)4 77 42 66 66
> http://zimmer.aprilfoolsreview.com/
> 
> 

------------------------------------------------------------
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 Thursday, 7 March 2013 17:31:55 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:25:54 GMT