DCAT-AP-IT/DCAT-AP support

[lucadex]

Hi,
I'm currently starting an Open Data portal project in Italy. As my company is a "Drupal shop", and since we have limited personnel who can develop with CKAN, I'd like to give DKAN a try.
As in Italy the DCAT-AP-IT data model for harvesting is used, any idea if somebody has developed a custom schema for it? In case, is it available as open source?

More generically, what is the current state of support for the European DCAT-AP in DKAN? I've seen several posts but couldn't get a clear idea on that (to be honest I don't know how big are the differences respect the DCAT-US schema which is currently the default in DKAN).
I'd appreciate if somebody can help me on that.
TIA!
Luca

[dafeder]

Hi @lucadex - is there a JSON Schema definition for DCAT-AP-IT? While DKAN's schema support is not currently quite as flexible as originally envisioned, the schemas can be customized and should work if they meet the following requirements:

1. Described in valid JSON Schema
2. Basic record defined as "dataset".
3. Datasets contain array of "distribution" objects, and data file URLs are in a downloadURL field in distribution.

Would be happy to continue checking in on progress as we would like DKAN to at minimum be able to support other variants of DCAT.

Thanks,
Dan

[dafeder]

Also, @stefan-korn may have some insight on this?

[lucadex]

Hi @dafeder thanks a lot for the prompt feedback.

There is an OWL definition of the schema. Also there is a diagram here:
https://dati.gov.it/sites/default/files/2020-02/DCAT-AP_IT_UML.png
It seems on the one hand a subset and an extension of DCAT.

Apparently at least one Italian data catalogue based on DKAN supports this schema:
https://opendata.comune.fi.it/

If anyone is aware of a publicly release DKAN custom schema for DCAT-AP-IT please comment here.
Thanks!
L.

[dafeder]

@lucadex I would be very surprised if there is an existing DCAT-AP-IT schema for DKAN 😄 . I tagged @stefan-korn because I believe he is working on one for the German variant of DCAT-AP, which will of course be closer than DCAT-US. Either way, it should be doable as the most basic elements - dataset, distribution, downloadURL are identical across all variants, AFAIK. The tricky thing will be to make a DCAT-AP valid schema in JSON-Schema, since DCAT-AP tends to be expressed either in RDF/XML or very verbose JSON-LD.

[stefan-korn]

Yes, I am currently trying to provide DCAT-AP.de schema (https://www.dcat-ap.de/) for use with DKANv2. At this moment I am just changing the schema definitions in schema/collections as needed. So I am basically overwriting these definitions, which would actually decouple them from the ones provided with the DKANv2 default. This bothers me a bit and first I was thinking about having my own DCAT-AP.de definitions beside the original definitions from DKANv2 and maybe switch between the DKANv2 standard and DCAT-AP.de somehow. But there are a few difficulties in there, as some logic in DKANv2 is tied to specific schema parts (for example to the dataset definition) and it is not that easy to solve this as far as I can see.
I am very open for suggestions on how to integrate this better with the DKANv2 default definitions (as a variant, rather than an override) and I am also willing to provide the DCAT-AP.de schema as open source (not at this moment as I am currently in the midst of creating the schema, but surely later).

Regarding the RDF/XML and/or JSON-LD I am planning to provide output in these formats via some custom endpoints, so I can use the JSON-schema features from DKANv2 for storing and entering data. We had some similar approach in DKANv1 to provide DCAT-AP.de conform output.

The Italian data catalogue from the Firenze municipality is very interesting. They are using Drupal 9, but I am not sure if they are also using DKANv2. If so, it would of course be very interesting to get some insights on their solution.
Very interesting is also there use of linked data, for example https://linkedopendata.comune.fi.it/lodview/Organizzazione/1169_Comune_di_Firenze_-_Direzione_Ambiente.html - seems they use this tool https://github.com/LodLive/LodView for that.

[dafeder]

@stefan-korn not sure if this was clear or what you're doing but you can override the default schemas without editing the DKAN module itself; copy the schemas folder into your docroot and it will read them from there instead of the module.

This is of course not a very standard approach for a Drupal module; I'd like to move to a system where schemas are defined by modules. In that way DKAN could ship with DCAT-US and DCAT-AP modules, and DCAT-AP could demonstrate how to further extend for country-specific variants.