- From: Peter F. Patel-Schneider <pfps@research.bell-labs.com>
- Date: Wed, 22 Mar 2006 09:49:45 -0500 (EST)
- To: phayes@ihmc.us
- Cc: public-rdf-dawg-comments@w3.org, franconi@inf.unibz.it
From: Pat Hayes <phayes@ihmc.us> Subject: Re: [OK] Re: [OK?] Re: comments on "SPARQL Query Language for RDF" (Non-respect for RDF Semantics) Date: Wed, 8 Mar 2006 11:32:46 -0800 > >On 7 Mar 2006, at 00:26, Peter F. Patel-Schneider wrote: > >>>>(This also depends on just how the scoping graph is determined.) > >>> > >>>Mmhh, I'd say that it does not depend on that: the answer is uniquely > >>>determined up to bnode renaming. Why are you saying that? > >> > >>Well, changing the scoping graph can change the permissable answers, or at > >>least that is what I believe based on the SPARL documents. > > > >The scoping graph is always isomorphic to the original dataset - the > >only things that may change are the names of the bnodes. This has no > >effect on the answer set, apart from having possibly different bnode > >names. > > Quite. The bnode names used in the answer document are scoped > separately from those in the query pattern, and need not (though may, > a possibility we do not endorse but do allow) be the same as those > used in the dataset, but they satisfy the same global patterns as the > bnodes in the dataset graphs. The purpose of having the scoping graph > at all is to ensure that the third condition is true; the reason for > requiring it only to be isomorphic to - rather than identical to - > the dataset, is to ensure the second condition; and the reason for > requiring it to be bnode-disjoint with the query is to ensure the > first condition. All of this has to do with bnode scoping, and none > of it would be needed if there were no bnodes, in which case the > definition could simply be that G (simply) entails B(Q). > > Pat This sort of discussion needs to be in the document - along with more examples. peer
Received on Wednesday, 22 March 2006 14:50:01 UTC