W3C home > Mailing lists > Public > public-rdf-wg@w3.org > May 2012

Re: Why empty named graphs make sense

From: Pat Hayes <phayes@ihmc.us>
Date: Tue, 29 May 2012 20:38:37 -0500
Cc: Andy Seaborne <andy.seaborne@epimorphics.com>, RDF-WG <public-rdf-wg@w3.org>
Message-Id: <75627068-CDEA-4700-B101-9A297CAB14AE@ihmc.us>
To: Sandro Hawke <sandro@w3.org>

On May 29, 2012, at 7:41 PM, Sandro Hawke wrote:

> On Wed, 2012-05-16 at 17:33 +0100, Andy Seaborne wrote:
>> 1/ Scenario:
>> 
>> A site publishes a page, in RDFa, with triples in it.
>> 
>> The site replaces that page, with one saying, "changes in progress, 
>> please come back later" with no triples.
>> 
>> Recording this seems like a reason to have empty named graphs.
> 
> I agree it's useful, I just don't know how to make it work with
> partial-graph semantics.   See "subset semantics" in recent message,
> http://lists.w3.org/Archives/Public/public-rdf-wg/2012May/0650.html
> 
> If 
>     <g1> { }
> and
>     <g1> { :a :b :c }
> entails
>     <g1> { :a :b :c }
> 
> then information was the first graph conveying?   I think it was
> conveying that <g1> had at least no triples, which isn't conveying much
> at all.

Right. Semantically, an empty graph is an empty conjunction, which is trivially true. So it amounts to a kind of RDF tautology. And a tautology, indeed, conveys no information, when asserted. Still, it isnt incohwerent, just kind of dim.

Pat


> 
> I think this use case could be addressed with a vocabulary instead of
> syntax, as you suggest I think for dealing with quadstores.  That is,
> instead of allowing in a dataset:
> 
>   <g> { }
> 
> we could put in the default graph something like:
> 
>   <g> a rdf:EmptyResource.
> 
> It's certainly not as elegant, though.   (and not compatible with
> existing SPARQL systems.)
> 
> Oh.   I guess with partial-graph semantics (aka subset semantics) one
> could also make it complete with a count of the triples.  So
> 
>   <g> { :a :b 1,2,3 }.
>   <g> rdf:tripleCount 3.
> 
> would tell us <g> has those three triples and ONLY those three triples.
> With that, the complete-graph-semantics dataset:
> 
>   <g> { }
> 
> could be conveyed by the partial-graph-semantics dataset:
> 
>   <g> rdf:tripleCount 0.
> 
> Still not elegant, but it might be about as simple as possible.
> 
>    -- Sandro
> 
> 
>> 2/ The default graph can be empty.
>> 
>> 3/ A point made has been that N-quads can't represent an empty named graph.
>> 
>> A syntax fix could be:
>> 
>> _ <http://space> .
>> 
>> or other style indicate a space name.
>> 
>> 
>> To my reading, the spaces document works if "quads" are removed, not 
>> that I'm suggesting that.
>> 
>> How about defining as a convenience of useful vocabulary rather than 
>> core definition?
>> 
>> 	Andy
>> 
>> 
> 
> 
> 
> 

------------------------------------------------------------
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 Wednesday, 30 May 2012 01:39:11 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:25:49 GMT