Re: Proposal to resolve ISSUE-86

On Mon, Oct 12, 2015 at 1:36 AM, Holger Knublauch <holger@topquadrant.com>
wrote:

>
>
> On 10/12/15 12:09 AM, Karen Coyle wrote:
>
>>
>>
>> On 10/9/15 4:19 PM, Holger Knublauch wrote:
>>
>>> Also, some people will put everything into one file (option 1) so our
>>> tools need to live with that situation anyway. There is no harm in
>>> having the shapes as data. For some use cases, shapes *are* data.
>>>
>>
>> Putting shapes in the graph will be awkward (at best) when the shape is
>> closed. It would require one to write an ignore statement that includes
>> every shape property used.
>>
>
> I don't see how these topics are related. In practical terms, having
> shapes in the data graph only leads to unrelated triples IMHO, e.g.
>
> ex:MyInstance
>     sh:nodeShape ex:MyShape ;
>     ex:someProperty 42 .
>
> ex:MyShape
>     a sh:Shape ;
>     sh:constraint [
>         a sh:ClosedShapeConstraint ;
>         sh:ignoredProperties [ sh:nodeShape ] ;
>     ] ;
>     sh:property [
>         sh:predicate ex:someProperty ;
>         sh:datatype xsd:integer ;
>     ] .
>
> What different would the presence of the definition of ex:MyShape in the
> data make to the validation of ex:MyInstance?
>
> More generally: does anyone have cases where having shapes in the
> dataGraph causes problems?
>

In general no.
In the context of this issue, when someone implicitly loads the shapes of
e.g. foaf/skos (if they are defined inline or with owl:imports) and she
performs a validation with everything in one graph, then the shapes of
foaf/skos will be enforced in the data. If this is the user's intention,
everything is fine, if the user would like to deviate from these shapes, it
is very hard to isolate these shapes from the validation.
Additionally, the shapes will be validated as data. Although I can be vary
rare to cause a false violation result they might induce additional
constraints to validate the shapes. A SHACL engine might hardcode this
behavior or give an option to the user.


>
> Thanks,
> Holger
>
>
>
>
> That leads me to conclude, by the way, that the ignore function might need
>> to work on namesapaces, not just individual properties. And, of course,
>> that doesn't work for the base namespace.
>>
>> kc
>>
>>
>
>


-- 
Dimitris Kontokostas
Department of Computer Science, University of Leipzig & DBpedia Association
Events: http://wiki.dbpedia.org/meetings/California2015 (Nov 5th)
Projects: http://dbpedia.org, http://rdfunit.aksw.org, http://
http://aligned-project.eu
Homepage:http://aksw.org/DimitrisKontokostas
Research Group: http://aksw.org