Re: "Multi-Edge Support in RDFn" slides

On 16/12/2022 15:10, Antoine Zimmermann wrote:
>
>
> Le 16/12/2022 à 12:58, Pierre-Antoine Champin a écrit :
> >>   that means the same everywhere.
> > just like an IRI, a literal or a blank node means the same 
> everywhere (mind you, I'm talking about blank nodes, not blank node 
> identifiers, who have a local scope).
>
> What do you mean by "a blank node means the same everywhere"?
> Literals and IRIs denote the same thing everywhere, yes. Conversely, 
> bnodes do not denote anything. Bnodes only indicate the existence of 
> things, and what thing exists as indicated by a bnode depends on the 
> RDF graph being considered.

You are right, I got carried away. "everywhere" for blank nodes is 
restricted to "in the same graph".

Also, you are right that "means the same" does not mean the same (!) for 
IRIS+Literals and for blank node. But still: a blank node asserts the 
existence of (at least) one and the same thing that must satisfy all the 
triples where it occurs.

>
> It happens that the CG report interprets quoted triples as bnodes 
> whose identity depends on the RDF-star graph being considered. So if 
> you consider different RDF-star graphs containing the same quoted 
> triple, the quoted triple in the context of the first graph indicates 
> the existence of a different things than the quoted triple in context 
> of the second graph.
>
> See https://w3c.github.io/rdf-star/cg-spec/editors_draft.html#mapping 
> item 2.2).

Along the same line as my response to Peter about uniqueness (end of 
[1]), I think we should distinguish the semantics of RDF-star, and the 
semantics of the underlying RDF graphs defined by the mapping -- which 
only partially captures the former. The full semantics of RDF-star is 
split between the mapping and the RDF entailment of the "unstarred" graphs.

You claim that "the CG report interprets quoted triples as bnodes", I 
disagree with this framing. The CG reports uses bnodes as an artifact to 
specify the semantics of RDF-star. The constraints that can't be 
captured by the bnodes (e.g. the uniqueness of quoted triples) is 
handled by the mapping -- because we can't handle it in the RDF 1.1 
semantics.

Granted, this "splitting" of the semantics may seem hacky. I would be 
happier with a clean MT semantics defined from the ground up, inspired 
by RDF's semantics, with the addition on quoted triples. We tried to do 
that in the CG, we didn't succeed, but we may try again.

   pa


[1] 
https://lists.w3.org/Archives/Public/public-rdf-star-wg/2022Dec/0037.html