Re: ISSUE-23: A specific proposal

On 4/25/15 11:25 AM, Michel Dumontier wrote:
> It's not good practice to annotate somebody else's URI. Is there no 
> other mechanism by which I declare property descriptions against a 
> target type?

To me, one of the strenghts of having URIs is to be able to 
address/repurpose/refine resources that others have published. I don't 
see how to avoid a (forward or backward) reference to the URI of the 
target type. Do you have suggestions?

Thanks,
Holger


>
> m.
>
> Michel Dumontier, PhD
> Associate Professor of Medicine (Biomedical Informatics)
> Stanford University
> http://dumontierlab.com
>
> On Thu, Apr 23, 2015 at 7:40 PM, Holger Knublauch 
> <holger@topquadrant.com <mailto:holger@topquadrant.com>> wrote:
>
>     On 4/24/2015 12:18, Michel Dumontier wrote:
>
>         in any case, there are three fundamental issues, as I
>         currently see it
>         1. that the specification should indicate how a shape can be
>         defined in terms of an existing vocabulary, rather than be
>         intrinsic to the vocabulary definition (although I don't mind
>         if this is shown as in Example 1)
>
>
>     If (for whatever reason) you don't want to put your constraints
>     into a vocabulary file, you could use named graph/file imports to
>     store different shape definitions in different named graphs.
>     That's similar to taking an OWL ontology and reusing its class
>     definitions in another file (that owl:imports the ontology). The
>     SHACL validation is started with a parameter, which is the named
>     graph containing the shapes for this session. So e.g.
>
>     File1
>
>     ex:Person
>         a rdfs:Class ;
>         rdfs:label "Person"
>     .
>
>     File 2:
>
>     ex:Person
>         sh:constraint [ ... ] .
>
>     Then start validation with File 2 as shapes graph.
>
>         2. that the valueType should be an IRI for a class or a shape,
>         and we should drop sh:shape.
>
>
>     I already responded to that and disagree that we should merge them
>     together.
>
>         3. that a simple SPARQL query should or should not return that
>         data are instances of shapes regardless of the validation.
>
>
>     The system designer has a choice between using sh:Shapes and
>     Classes-as-shapes. There are all kinds of engineering solutions,
>     including named graphs, to take complete control.
>
>     Holger
>
>
>