[profiledesc] role assumes closed world of the document

[azaroth42]

The profiledesc ontology has a role associated directly with a resource which it does not control. For example:

<https://.../dcat-ap-v11> prof:resource <http://data.europa.eu/...c096> .
<http://data.europa.eu/...c096> prof:resourceRole roles:Guidance .

This role is only true for dcap-ap-v11, and the same resource might have other roles in other contexts. For example a Guidance document for one profile might also include a constraints description for a different profile. Or more likely, a PartConstraints for one profile might be a FullConstraints for another simpler profile.

There are various solutions:

• Named Graphs. Not supported by many systems, and you only get to use the ace once (a triple can be part of only one named graph).
• Reification. Instead of resource, have a node between the Profile and the document that has the role associated with it. Compare the Web Annotation model for SpecificResource. Much more verbose, but good when the vocabulary of roles is open ended.
• Many Relationships. A simpler pattern but harder to extend -- each role is encoded in the predicate between the Profile and the document, eg hasPartConstraints vs hasFullConstraints. Good for implementations when the vocabulary is closed and relatively small.

Reification:

<profile1> prof:hasResourceWithRole <rr1> .
<rr1> prof:hasRole roles:PartConstraint ;
         prof:forDocument <constraint1>.

Relationships:

<profile1> prof:hasPartConstraintResource <constraint1> .

[nicholascar]

We’re bashing out how to assign roles to Agents with respect to Entities for DCAT now with the PROV WG mailing list. It looks like we will need a new reification pattern there since PROV doesn’t allow direct Agent to Entity roles (only Agent to Activity).

If we specify a more time invariant Agent to Entity hadRole or similar there, we could use it here too.

I really don’t want to see a fixed set of relationships or Named Graphs used!

[andrea-perego]

Transferring this issue to https://github.com/w3c/dx-prof/ , as it is specifically related to the PROF vocabulary.

[rob-metalinkage]

transferred.

[rob-metalinkage]

This is actually a duplicate of #39 and #46 - since a prof:ResourceDescriptor is essentially a reified statement about role of artefacts, which may already have different roles in different contexts as requested. so the issue is actually about long v. short naming options for the predicates and classes involved.

[rob-metalinkage]

Closed as duplicate

[kcoyle]

I think you should ask @azaroth42 if this issue is resolved. Those other issues do not seem to address his comments directly.