Define a Datacite namespace; Convert schema into the JSON-LD format

[Zack-83]

Context/problem

With the expansion of the semantic web it has become more and more important to set up metadata descriptions composed of fields defined univoquely in a semantic way in a so-calles "namespace".

DataCite does define univoquely fields via its XML and JSON schema, but it misses a semantic description and a resolvable namespace. This makes it very difficult to create metadata schemas compatible with DataCite referencing it and mapping to it. For instance, the AIMS metadata profile generator https://profiles.nfdi4ing.de/ is not able to upload any DataCite schemas in the currently available format. A schema validator such as https://validator.schema.org/ confirms that the field types are not defined enough.

Describe the solution you'd like

• Please define a namespace URI, choose a namespace prefix, and relate the field definitions to it or to other existing namespaces/ontologies. A previous version of it, now outdated, is https://sparontologies.github.io/datacite/current/datacite.html .
• Please publish the DataCite schema in a semantic format: TTL, NTriples, JSON-LD or RDF/XML.

Describe alternatives you've considered

The alternative for me would be to add all needed fields by hand, getting them from other namespaces (DCTerms, DCat, schema.org ...) and mapping them to DataCite separately in a program script and not in my schema description itself.

Additional context

A previous semantic implementation of the DataCite schema is https://sparontologies.github.io/datacite/current/datacite.html . Unfortunately it relates to a now deprecated version of the DataCite schema. I would find it good if DataCite itself maintained a semantic version of its own schema.

[tmorrell]

I'll start by noting that this isn't an official DataCite schema, it's a community-based schema. So we accept contributions from anyone, but don't have any ability to change DataCite's services.

The primary use case for the jsonschem is to validate json that is submitted to the DataCite rest API https://support.datacite.org/docs/api-create-dois. So we're need to validate json that looks like https://github.com/inveniosoftware/datacite/blob/master/tests/data/4.5/datacite-example-dataset-v4.json with those specific field names.

It would be nice to be able to also provide a semantic implementation, but I'm not sure about compatibility and what the maintenance burden would look like. I'd certainly be happy to review a pull request, but it does need to meet the criteria above as well as be something that we could maintain going forward.