Re: ISSUE-30: How does SPARQL's notion of RDF dataset relate our notion of multiple graphs?

On Apr 18, 2011, at 20:02 , Pat Hayes wrote:

> On Apr 18, 2011, at 8:16 AM, Ivan Herman wrote:

>> Hm. I do not know...
>> Given that SPARQL is loose in the way they use URI-s from graphs in datasets, I would not be shocked by this. The question is what would the SPARQL 1.1 entailment regime say about this. 
>> But, again: I wonder whether we have to say anything in formal terms at all about SPARQL's behaviour, except to make it clear that (<g>,G) is _not_ a shorthand for <g> identifying G.
> But, Ivan, this thread began when *you* suggested using RDF to assert/declare something about SPARQL behavior. Which is exactly what is ruled out by this looseness. All Pierre-Antoine and I are doing is trying to take your idea and make it work in RDF. Are you now saying it was a bad idea all along, or are you saying that we don't need to do it using RDF, or what?
> Pat

I stand corrected, I indeed raised that:-) And, indeed, I am having second thoughts whether it was a good idea. Good idea in the sense whether it is necessary to add this to any of our specs. I am not sure what it would bring to the community.

However... if we decide something should be said than I do understand your point about the URI-s, so I see that the literal approach can work.


Ivan Herman, W3C Semantic Web Activity Lead
mobile: +31-641044153
PGP Key:

Received on Tuesday, 19 April 2011 08:10:39 UTC