- From: Andy Seaborne <andy@apache.org>
- Date: Fri, 13 Dec 2013 16:37:51 +0000
- To: public-rdf-wg@w3.org
On 13/12/13 15:49, Antoine Zimmermann wrote: > Le 13/12/2013 13:35, Andy Seaborne a écrit : >> On 12/12/13 21:08, Antoine Zimmermann wrote: >>> The new draft addresses most of Pat's comments. I added 2 issues to it. >>> I also added a section that reflects Sandro's box dataset semantics, >>> following Guus' suggestion, and also made some changes in response to >>> Peter initial comments. >>> >>> >>> https://dvcs.w3.org/hg/rdf/raw-file/default/rdf-dataset/index.html >>> >>> >>> If it looks ok, we can vote by email to make it a FPWD. Sorry for not >>> having done this before yesterday teleconf. >> >> >> [[ >> 3.2 Default graph as union or as merge >> >> It is sometimes assumed that named graphs are simply a convenient way of >> sorting the triples but all the triples participate in a united >> knowledge base that takes the place of the default graph. More >> precisely, a dataset is considered to be true if all the triples in all >> the graphs, named or default, are true together. >> ]] >> >> This starts talking of dftGraph = union of named graphs, not a super >> set, but then says it is all the triples in graphs, named or default. >> >> Shouldn't the default graph be exactly the union (version 2: merge) of >> the named graphs? > > The semantics defines the truth of all datasets. Not only the datasets > with a default graph equal to the union of the named graphs. > > >> Or there needs to be some rewording because "default graph" is being >> used is different ways. > > In different ways? What way is different from what other way? In one case, there is no ground data in the dft graph, in the other there is. > > >> [[ >> This description allows two formalizations of dataset semantics, >> depending on how blank nodes spanning several named graphs are treated. >> ]] >> >> It is worth nothing that TriG and N-quads use document-scoped blank node >> labels. > > worth nothing? or worth noting? "noting" > In any case, the formal semantics is > agnostic wrt the serialisation syntax used. The definition of RDF > datasets allows for shared blank nodes across several named graphs, and > the semantics has to apply to all RDF datasets. > >> I don't know of any system that does NGs+dftgraph -> dftGraph of >> dataset. Are there any? > > Sesame and OWLIM apply reasoning on the union of all named graphs to > generate triples that are then inserted into the default graph. This > materialisation of inferences creates a dataset where the default graph > is a strict superset of the union of named graphs. But that's not what the example is - as you describe it, only triples that in NGs or inferred are in the inferred/materialisation dft graph. Any in the original dft graph are lost. > > >> 3.6 Quad semantics >> >> Worth mentioning about the empty graph. > > Ok, empty named graphs cannot be represented as quads, but this is > essentially a syntactic issue. Still, I'd say it could be noted > nonetheless. > > > AZ. > >> >> Andy >> >> >> >
Received on Friday, 13 December 2013 16:38:21 UTC