Harvest attribution roles #3149
Conversation
Support FOAF.name for dct:publisher, dct:creators, dct:contributors
maudetes
changed the title
magopian
force-pushed
the
feat/add-roles-in-harvest-extras
branch
from
3d8fe90 to
609150c
Compare
magopian
force-pushed
the
feat/add-roles-in-harvest-extras
branch
from
609150c to
e200393
Compare
…services when generating the fixtures Also auto-set a role on the contact point if there is none - if email or contact_form present: `contact` - else `creator`
…V:qualified_attribution`
udata/core/dataset/rdf.py
Outdated
| if contact_point.role == "publisher" | ||
| ]: | ||
| # There's already a publisher, so the owner should instead be a qualified attribution. | ||
| owner_role = PROV.qualified_attribution |
There was a problem hiding this comment.
qualifiedAttribution points towards an Attribution and not directly an agent I think.
See maybe this section in W3C DCAT as well as the DCAT-AP one and finally the GeoDCAT-AP one.
I am not 100% clear on the best predicate to use for an organization if we have a dedicated publisher contact point.
There was a problem hiding this comment.
Yeah, having gone through the links you provided, here's a non-exhaustive list of possible roles:
What do you think? Did I miss any? Do you have a favorite?
There was a problem hiding this comment.
Let's go for geodcat:distributor for now? Since it's the organization distributing the data on data.gouv.fr, I would say it's the closest role.
|
Thanks you very much @maudetes for the detailed review, I believe i've addressed all your concerns, or answered with comments, would you mind having another look to validate that my changes are in line with what you were thinking, and letting me know what you think of the remaining things to decide? |
udata/core/dataservices/rdf.py
Outdated
| @@ -43,7 +43,11 @@ def dataservice_from_rdf( | |||
| dataservice.base_api_url = url_from_rdf(d, DCAT.endpointURL) | |||
| dataservice.endpoint_description_url = url_from_rdf(d, DCAT.endpointDescription) | |||
|
|
|||
| dataservice.contact_point = contact_point_from_rdf(d, dataservice) or dataservice.contact_point | |||
| # TODO: what are the type of contact points supported on dataservices? | |||
There was a problem hiding this comment.
I think we can go with this indeed :) Not sure we need the list(), I think a generator would do?
udata/core/dataset/rdf.py
Outdated
| if contact_point.role == "publisher" | ||
| ]: | ||
| # There's already a publisher, so the owner should instead be a qualified attribution. | ||
| owner_role = PROV.qualified_attribution |
There was a problem hiding this comment.
Let's go for geodcat:distributor for now? Since it's the organization distributing the data on data.gouv.fr, I would say it's the closest role.
* `contact_point` is now renamed to `contact_points`, a list * Also import "rpg" dataset fixture * Update the fixtures without nesting the dataset in reuses, discussions, resources and dataservices * Update the results.json with the latest changes in opendatateam/udata#3149
Part of datagouv/data.gouv.fr#1588
dct:publisher,dct:creators,dct:contributors.dct:publisher, so if there's a contact point with the rolepublisher, the owner should beprov:qualifiedAttributioninsteaddct:publisher,dct:creators,dct:contributorsin RDF (with correct properties depending on role)play around to add a suggest API on these contacts