- From: Thomas Baker <tom@tombaker.org>
- Date: Sun, 1 May 2016 19:23:48 +0200
- To: Irene Polikoff <irene@topquadrant.com>
- Cc: RDF Shapes <public-rdf-shapes@w3.org>
Hi Irene, On Sun, May 01, 2016 at 11:59:30AM -0400, Irene Polikoff wrote: > However, fundamentally, SHACL operates under the closed world assumption. > So, it could never concern itself with whatever exists or doesnąt exist in > the world in general - it is always only about the information that has > been been made available/submitted to a SHACL engine. Negation as failure > and so on. Sure - I'm completely with you re: CWA. That was, indeed, the idea behind the DSP work circa 2007-2008. In SHACL, as I picture it (which is not necessarily what the spec says), CWA should operate with respect to the triples asserted in the graph of data at hand. > This could be fixed by saying something like: > > "Class-based scopes define the scope as the set of all instances of a > Class present in the data graph." Yes, that's better, though in the spirit of a CWA view on a given data graph, might it perhaps be more precise to refer not to a set of instances, but to a set of nodes? > Perhaps, it is worth making a point of explaining this somewhere early on > and then, there is no need to constantly clarify points like this > everywhere in the document. If a reader approaches the spec with the OWA > set of mind, they are bound to make wrong conclusions or become confused. > Thus, it is important that a reader puts themselves into the CWA set of > mind. Yes, I agree - and I agree that it would be worthwhile to make the point early. My point, however, is that when the spec says: "Class-based scopes define the scope as the set of all instances of a class." it is using the language of RDF Schema. Full stop. If this is not the intended meaning (and it shouldn't be), then the spec should use different words. Tom -- Tom Baker <tom@tombaker.org>
Received on Sunday, 1 May 2016 17:24:24 UTC