W3C home > Mailing lists > Public > public-rdf-dawg@w3.org > July to September 2010

Re: Graph naming in Service Descriptions (ACTION-266; discussion of comment DS-1)

From: Axel Polleres <axel.polleres@deri.org>
Date: Tue, 20 Jul 2010 16:21:40 +0100
Cc: "Steve Harris" <steve.harris@garlik.com>, "SPARQL Working Group" <public-rdf-dawg@w3.org>
Message-Id: <40E4FB4C-DCBB-4D22-B8C3-4AAB2A4A4A7E@deri.org>
To: Gregory Williams <greg@evilfunhouse.com>
I was now browsing through the mails and the discussins again, but I still feel - sigh - to dumb (or caught in our current definitions) to grasp what is the issue...

for me, it seems clear that any kind of inference can only and solely affect basic graph pattern matching... at least within the scope of our current charter.
 
that is, e.g. given Andy's example...

On 13 Jul 2010, at 11:08, Andy Seaborne wrote:
> If the engine is capable of handling the fact in the dataset that:
> 
> <http://www.example/named-graph> owl:sameAs
> <http://www.example/another-named-graph>
> 
> in the dataset and so making
> 
> GRAPH <http://www.example/named-graph> { ... }
> 
> work on
> 
> GRAPH <http://www.example/another-named-graph> { ... }
> 
> then that seems OK to me.
> 
> Similarly for GRAPH ?g {}

... there seems absolutely no way that 
 
 GRAPH <http://www.example/another-named-graph> { ... }

would be treaded synonuymous to 

 GRAPH <http://www.example/named-graph> { ... }

because the sameAs statement (BTW, in which graph is the sameAs statement contained?) can't affect the GRAPH name...
... and that's good, because also, there is absolutely no problem with having

 <g1>: <g1> owl:sameAs <g2>

 <g2>: // some totally diffferent statements, e.g.
       :s :p :o
  
in your dataset. 


 So, let's take some example queries... and assume owl entailment regime

   SELECT ?X 
   FROM NAMED <g1>
   FROM NAMED <g2>
   WHERE{ GRAPH <g2> {?X ?Y ?Z}

 the current understanding would be that ?X would bind to <g2> and all other kinds of axiomatic stuff coming from OWL about the graph *named* <g2>... nothing more.
 There is no inferences across graphs in the dataset - this is the assumption we have been working on all the time now - is that what is being questioned here? 
 If so, I have some fear that if we decide to mess with entailments beyond BGP matching, we open a can of worms. It is an interesting problem to discuss, but to my understanding 
 beyond the scope of this WG.

Awaiting Sandro's explaining mail and examples,

Axel
 




On 14 Jul 2010, at 18:15, Gregory Williams wrote:

> On Jul 13, 2010, at 2:41 AM, Steve Harris wrote:
> 
> > I don't think that we should try and gracefully handle situations where people assert nonsense in other graphs, that's just opening can of worms.
> 
> Agreed.
> 
> It seems Damian has now retracted[1] his entailment issue, so I think the consensus here is that it's a non-issue. Sandro's issues -- the naming of sd:name and the modeling of the named graphs in the service description -- are still open, though.
> 
> .greg
> 
> [1] http://lists.w3.org/Archives/Public/public-rdf-dawg-comments/2010Jul/0007.html
> 
> 
> 
Received on Tuesday, 20 July 2010 15:22:16 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:15:43 GMT