- From: Pat Hayes <phayes@ihmc.us>
- Date: Thu, 7 Mar 2013 12:22:23 -0600
- To: Antoine Zimmermann <antoine.zimmermann@emse.fr>
- Cc: RDF WG <public-rdf-wg@w3.org>
On Mar 7, 2013, at 9:52 AM, Antoine Zimmermann wrote:
> There is a problem with the definition of merge in the draft.
>
> I'm using math notations instead of a concrete serialisation syntax because I want to show things at the abstract syntax level, which is what RDF Semantics relies on.
>
> Let us take a blank node b from the set of blank nodes. Let us consider the two graphs G1 = {(<s1>,<p1>,b)} and G2 = {(<s2>,<p2>,b)}.
You have the same bnode in both graphs, so they must be in the same scope, right? For example, both are subgraphs of a larger graph, or both in the same dataset.
>
> Let us ask ourselves whether {G1,G2} entails:
>
> G = {(<s1>,<p1>,b),(<s2>,<p2>,b)}
>
> The answer is trivially NO wrt the current semantics of the ED.
If those really are the same b, then the answer is YES, and I claim that it should be.
> Yet, reasoners typically would, in this case, merge G1 and G2 to get a single graph and work out the non-entailment. However, with the new definition of merge, this breaks and fails. Reasoners must standardise apart the bnodes and that's what the merge should be.
No, reasoners should not have to standardize bnodes apart. They may have to standardize bnode *identifiers* apart, but that is just what you would expect when using local identifiers.
Pat
> --
> 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 18:22:53 UTC