- From: Peter F. Patel-Schneider <pfpschneider@gmail.com>
- Date: Mon, 9 May 2016 18:32:11 -0700
- To: kcoyle@kcoyle.net, public-data-shapes-wg@w3.org
On 05/09/2016 06:17 PM, Karen Coyle wrote: > > > On 5/9/16 4:32 PM, Holger Knublauch wrote: >> To avoid confusion, SHACL has two constructs where rdfs:subClassOf >> triples matter, and both are in the data graph: >> >> - sh:class >> - sh:scopeClass >> >> In neither of them any form of inferencing is needed - I would be the >> last person on this mailing list to make SHACL require inferencing. All >> we require is *querying* or pattern matching like SPARQL does it, and >> SPARQL provides the rdfs:subClassOf* construct as a convenience, while >> the same can easily be programmed in any other execution language. >> >> Holger > > This seems to be an approach that isn't articulated in the spec - that SHACL > requires *querying* "like SPARQL does it." Is that what others understand? I'm > not objecting to this, I just haven't heard it stated like this before. > > kc With recursion, it is not obvious that SHACL can be implemented via SPARQL querying. Without recursion, SHACL can be implemented via SPARQL querying. However, querying, and particularly SPARQL querying, is inference. peter
Received on Tuesday, 10 May 2016 01:32:41 UTC