search

w3c / dxwg

Data Catalog Vocabulary (DCAT)
https://w3c.github.io/dxwg/dcat/
Other
144 stars 55 forks source link

Human-readable description location #369

Open nicholascar opened 5 years ago

nicholascar commented 5 years ago

This issue was created in the Profile Guidance document and is listed in it. Once consensus on addressing it is reached here in comments below, the results will be added to the document and the issue closed.

(Antoine:) One may argue that human-readable description (if it's provided as a full document) does fit better in a separate section. But upon re-reading the structure suggested by Karen I was not sure of the 'granularity' (of the description) meant here.

kcoyle commented 5 years ago

I was thinking more about data elements such as: labels, definitions, and instructions. These, to my mind, would be included in the profile and would describe entities (probably classes in RDF), properties, and value descriptions.

The full human-readable document would be described in the section on publication in my mind. What is tricky about that is that the human-readable document may be the profile itself (as in DCAT-AP whose only form is as a PDF document) or it could be a transformation of the profile, like the output of an XSLT if the profile is expressed in XML.

I think we have this problem of defining "what is the profile?" when it may exist in different flavors. DCAT and profileDesc could help us here because the profile could be an abstraction with one or more manifestations that have different properties. The next challenge is that some manifestations may have more than one property or role, such as something like the DCAT-AP that has terms, definitions, cardinality, and validation rules all in one "file" or "document". Plus there's the need to distinguish between conceptual and implementable, and a document like the DSP could be both. I think this isn't as easily defined as the difference between datasets and distributions.

agreiner commented 5 years ago

Wait, doesn't DCAT-AP include ShEx and ShaCL versions? They are very hard to find, but I seem to recall seeing them in a github. I tend to think that the actual profile should be the machine-readable description and the human-readable description (PDF or html) should be considered a profile description rather than a profile per se. @aisaac ?

rob-metalinkage commented 5 years ago

From what I can tell the SHACL expression was non-normative - it was "retro-fitted" later to aid in implementing the DCAT-AP profile, which was described in a document form.

Machine readable expressions are potentially only partially expressive of a profile. So the Profile is a identifiable thing, and its expressions (like dcat:distributions) need to be related with enough metadata to understand the role of that relation.

the profiles ontology allows for a description of the profile and its interoperability implications w.r.t. other profiles, then is agnostic about what forms (if any) are available to express the profile constraints.

nicholascar commented 5 years ago

There is a mechanism within the Profiles Ontology for giving resources within a profile a role (ResourceRole) and a role could be guidance and defined to be human readable and the vocabulary of roles already present in the ontology already contain a guidance role.