Re: What is "the serious bug in entailment semantics" found by J. Perez"?

Hi,

about

http://lists.w3.org/Archives/Public/public-rdf-dawg/2006JulSep/0085.html

On 8 Aug 2006, at 15:31, Seaborne, Andy wrote:
> I can only find this as:
> http://lists.w3.org/Archives/Public/public-rdf-dawg-comments/
> 2006Jun/0052

I first thougth that it was only a problem in FredZ's definitions (the
comment that AndyS mention) but then I realize that it was a problem too
in the currect spec (rq23 or rq24), because "for simple entailment the
scoping set B is the set of all RDF terms in G'" with G' the scoping graph
which is graph equivalent to the dataset graph.

Only to state clearly my point (up to now I have only presented it
off-lists to Enrico), I'm not saying that the definitions are wrong, they
are just definitions. I'm only saying that in the way the definition of
"BGP E-matching" is currently stated, there are claims in the spec that
are false and confusing. For example, the last paragraph in section 5.2 in
rq24 has the intention of an "implementation hint", but it does not follow
the formal definition. Another focus of problems (and I presented the same
example to Enrico) is in section 10.3 of rq24 when talking about
constructing graphs (by CONSTRUCT). The specification says (not explicitly
but at least is what I understand) that one can use the query ?s ?p ?o to
obtain a (sintactic) "copy" of a graph in the case of simple entailment,
but in my example this query (with graph template ?s ?p ?o) produces the
graph { (a, p, a), (X, p, Y), (X, p, X), (Y, p, Y) } following the current
definitions.

I think then that the definitions in section 5.1 must be changed to avoid
these problems, or the claim in section 5.2 must be revised to strictly
reflect what an implementation has to do to find solutions in the case of
simple entailment according to the formal definition.

Regards,
- jorge

Received on Tuesday, 8 August 2006 15:42:33 UTC