- From: Holger Knublauch <holger@topquadrant.com>
- Date: Tue, 6 Oct 2015 09:39:57 +1000
- To: public-data-shapes-wg <public-data-shapes-wg@w3.org>
- Message-ID: <56130A4D.8000609@topquadrant.com>
On 10/2/2015 18:51, Dimitris Kontokostas wrote: > > > On Thu, Oct 1, 2015 at 11:49 PM, Holger Knublauch > <holger@topquadrant.com <mailto:holger@topquadrant.com>> wrote: > > On 10/1/15 5:04 PM, Dimitris Kontokostas wrote: >> Now that it is more clear I would like to propose my resolution >> of issue-86. >> >> I suggest the spec mentions something in the lines of the >> following sentence >> ontology or vocabulary designers that want to publish SHACL >> constraints along with their schemas are encouraged (or SHOULD) >> either define the associated shapes in the same document with the >> schema or link to them through the sh:shapesGraph property. > > sh:shapesGraph would not have the right effect. It would basically > only say that the ontology (e.g. definition of skos:Concept) > itself would have to follow the shape definitions. I believe it > should be owl:imports instead. > > > owl:imports and inline declaration have the exact same effect. the > problem with those is that if a user imports an ontology in the data > graph, she automatically loads the shapes as well which might not be > what the user expects / wants. I believe for a large number of use cases, esp for people beginning with SHACL, we should promote the simplest possible design in which - like with OWL - instance, class, property (and shapes) all live in the same logical graph and owl:imports are used. This avoids many issues and lowers frustration and negative press like "this is over-engineered complicating stuff, let's not bother with SHACL". Many people will already wonder why we came up with this whole parallel universe of "shapes" anyway. I think what we need is a careful analysis of less than a handful of set-ups that we can reasonably support, and put best practice recipes for those set-ups into our documentation. So I retract my previous comments, and now believe that we indeed need to do something here with this ticket. Holger > > I agree that sh:shapesGraph does not have the exact semantics in this > case but when the ontology is loaded in the data graph, sh:shapesGraph > gets proper semantics again. > We can redefine the semantics of sh:shapesGraph or introduce a new > property for this > > Overall, I don't think we need to specify or recommend anything > here. There will be different design patterns emerging, and we > cannot anticipate yet which variation people will prefer, how > SHACL will relate to OWL etc. > > > If we do not recommend anything we are forcing people to write shapes > inline or use owl:imports which might have an undesired effect. > I am ok with that, although I would prefer to have an alternative. > > Dimtiris > > > > Holger > > > >> >> This is independent of Peter's suggestion and if the WG thinks >> that Peter's suggestion should also exist in the spec I would >> vote +1 on this as well. >> >> Dimitris >> >> On Thu, Oct 1, 2015 at 8:43 AM, Dimitris Kontokostas >> <kontokostas@informatik.uni-leipzig.de >> <mailto:kontokostas@informatik.uni-leipzig.de>> wrote: >> >> >> >> On Thu, Oct 1, 2015 at 5:35 AM, Peter F. Patel-Schneider >> <pfpschneider@gmail.com <mailto:pfpschneider@gmail.com>> wrote: >> >> I do not see that SHACL needs any connection between a >> shapes graph and an >> ontology definition. >> >> For purposes of designing a collection of shapes, having >> access to an ontology >> that provides axioms about the classes in a collection of >> shapes is handy. >> However, validating SHACL shapes or documents against a >> data graph or node in >> a data graph does not need any link going from the shapes >> graph to an ontology >> graph. A SHACL validation engine does need to have >> access to ontology axioms >> to determine whether a node in the data graph is a SHACL >> instance of a class, >> but this is best done by including a graph with the >> required ontology axioms >> into the data graph. >> >> I therefore vote 0 for a) and -1 for the other options. >> >> >> Peter, >> >> I also do not think that shacl needs a link to an >> ontology/vocabulary. >> The issue subject is indeed not clear but the intent was >> about the reverse relation: ontology/vocabulary to shacl >> >> e.g. skos could define their additional constraints [1] in >> shacl and my issue was about how could e.g. skos publish >> these constraints >> >> Dimitris >> >> [1] http://www.w3.org/TR/2009/REC-skos-reference-20090818/#L2422 >> >> >> I would vote +1 for a proposal like: >> >> PROPOSED: The SHACL spec states that there is no need for >> a link from a SHACL >> shapes graph to an ontology graph and does not define >> such a link. The SHACL >> spec further states that there is nothing in SHACL to >> prevent a SHACL shapes >> graph from including ontology axioms or importing >> ontology axioms, but that >> such inclusion or importation has no effect on >> determining whether a node in a >> data graph is a SHACL instance of a class. The SHACL >> spec states that >> ontology axioms that affect SHACL are either part of the >> data graph or >> included from the data graph. The SHACL spec mentions >> that SHACL shape >> graphs are often best developed in conjunction with a set >> of ontology axioms >> and that tools for the development of SHACL shapes may >> want to provide >> mechanisms for viewing axioms from a separate ontology. >> >> This proposal clearly makes the required distinction >> between what is required >> for SHACL validation and thus should be part of the SHACL >> language, and what >> is useful for SHACL development and thus should not be >> part of the SHACL language. >> >> >> peter >> >> >> On 09/10/2015 01:09 AM, RDF Data Shapes Working Group >> Issue Tracker wrote: >> > shapes-ISSUE-86 (dimitris): Associating shapes with >> ontologies or vocabularies [SHACL Spec] >> > >> > http://www.w3.org/2014/data-shapes/track/issues/86 >> > >> > Raised by: Dimitris Kontokostas >> > On product: SHACL Spec >> > >> > Related to ISSUE-44, this is issue is about ways to >> associate an ontology or vocabulary to a set of shapes. >> > >> > Possible ways to resolve it >> > a) SHACL spec says nothing about associating >> ontologies/vocabularies with shapes >> > b) SHACL spec suggests the use of owl:imports >> > c) SHACL spec suggests the use of sh:shapesGraph >> > d) SHACL spec suggests shapes are defined in the same >> file with the ontology/vocabulary >> > e) SHACL spec suggests a combination of (d) with (b) or (c) >> > >> > >> > >> >> >> >> >> -- >> Dimitris Kontokostas >> Department of Computer Science, University of Leipzig & >> DBpedia Association >> Projects: http://dbpedia.org, >> http://http://aligned-project.eu, http://rdfunit.aksw.org >> Homepage:http://aksw.org/DimitrisKontokostas >> Research Group: http://aksw.org >> >> >> >> >> -- >> Dimitris Kontokostas >> Department of Computer Science, University of Leipzig & DBpedia >> Association >> Projects: http://dbpedia.org, http://http://aligned-project.eu, >> http://rdfunit.aksw.org >> Homepage:http://aksw.org/DimitrisKontokostas >> Research Group: http://aksw.org >> > > > > > -- > Dimitris Kontokostas > Department of Computer Science, University of Leipzig & DBpedia > Association > Projects: http://dbpedia.org, http://http://aligned-project.eu, > http://rdfunit.aksw.org > Homepage:http://aksw.org/DimitrisKontokostas > Research Group: http://aksw.org >
Received on Monday, 5 October 2015 23:40:32 UTC