Add JSONLD @context to the top level .ipynb node

[westurner]

From #43 (comment) :

JSONLD is application/ld+json:

• https://www.w3.org/TR/json-ld/
• https://en.wikipedia.org/wiki/JSON-LD
• https://www.iana.org/assignments/media-types/application/ld+json

There is a @context mapping for nbformat JSONLD out there:

• https://gist.github.com/bollwyvl/e7f8136bd2ea5674dd00

.

• SCH: .ipynb JSONLD @context by @bollwyvl
  • Would it be okay to merge this upstream?
  • TST: Does this cover all of the ipynb nbformat v4 attributes?
  • TST: Tests (@context + .ipynb => jsonld => rdf) (PyLD, rdflib-jsonld)
• Host an @context at a stable, versioned URI
  • https://<url>/ns/v4/nbformat.context.jsonld
    • There may be a lot of (cacheable) lookups
      • CDN (GitHub Pages (Fastly), CloudFlare)
  • @ jupyter, what would be the best way to handle this?
• DOC,SCH: nbformat specification: heading, link(s)?
• ENH,SCH: Add @context: to the top level .ipynb in Jupyter Notebook

Questions:

• Would this be a breaking change?
  • Signatures would change; but otherwise,no, IIUC, adding @context to .ipynb would not be a breaking or even a major change, so waiting until nbformat v5 would not be necessary.

[rgbkrk]

Well the @context is finely encapsulated inside of the payload for application/ld+json. There is no change to the notebook format spec here.

[westurner]

Value: Transform Jupyter Notebook JSON (nbformat .ipynb) to JSONLD to RDF (linked data).

http://5stardata.info/

[westurner]

Well the @context is finely encapsulated inside of the payload for application/ld+json. There is no change to the notebook format spec here.

The change to the notebook format spec would be adding "@context": "<url>", to the .ipynb JSON to link to an official JSONLD @context . (edit) and adding "@type": ["http://schema.org/DigitalDocument", "http://schema.jupyter.org/JupyterNotebook"] in the top-level nbformat node

[EDIT]
Here's a patch to add a JSONLD @context, author, and license to the nbformat JSONSchema: master...westurner:SCH_add_metadata.

[westurner]

See:

• Add JSONLD @context to the top level .ipynb node #44 Add JSONLD @context to the top level .ipynb node (this)
• Fields for author/creator and title #45 Fields for author/creator and title
  • Fields for author/creator and title #45 (comment) Example JSONLD for jupyter:JupyterNotebook
    • https://schema.org/name ("title" -> Thing.name)
    • https://schema.org/CreativeWork
    • https://schema.org/author
      • https://schema.org/Person +
    • https://schema.org/creator
    • https://schema.org/funder +
      • https://schema.org/Organization +
    • https://schema.org/license +
    • jupyter:JupyterNotebook < (CreativeWork < Thing, )
• Add document_info to schema #46 Add document_info to schema
  • westurner@c298b08
• Capture more of the kernel spec in the notebook document #49 Capture more of the kernel spec in the notebook document
  • Draft spec for capturing a notebook environment #60 Draft spec for capturing a notebook environment
• Rethinking kernelspecs in notebook metadata #81 Rethinking kernel specs in notebook metadata

[rgbkrk]

This would be a major version bump that would break all the frontends requiring validation (notebook, lab, etc.), so this is a non-starter. Closing. Feel free to open if there's likely development activity around it.

[bollwyvl]

Yeah, closing is appropriate. I think the annotation/metadata stuff is quiet for now, so it's unlikely to get pushed from that angle. It's still probably a good idea to take things like json-pointer-able items seriously, but the LD spec has evolved a fair amount to support more of the challenging items (e.g. mimebundle) without requiring us to change anything. If a frontend/backend wanted to support a context, they can still do it out-of-band as an HTTP header. Actually specifying a `$schema` top-level key is another story...

…

On Mon, Jan 6, 2020 at 4:53 PM Kyle Kelley ***@***.***> wrote: This would be a major version bump that would break all the frontends requiring validation (notebook, lab, etc.), so this is a non-starter. Closing. Feel free to open if there's likely development activity around it. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#44?email_source=notifications&email_token=AAALCRCJ336K3WB63GFOOJTQ4OR43A5CNFSM4COX7QF2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEIG5GYY#issuecomment-571331427>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAALCRFE6RWCUSQOVRG35PLQ4OR43ANCNFSM4COX7QFQ> .

[meeseeksmachine]

This issue has been mentioned on Jupyter Community Forum. There might be relevant details there:

https://discourse.jupyter.org/t/sustainability-of-the-ipynb-nbformat-document-format/3051/2

[westurner]

Is it now [almost] possible to use the new features of JSON-LD 1.1 to write a sufficient @context for nbformat 5.1?

Looks like there may be a JSON-LD 1.2/2.0 for RDFstar* support; and possibly whatever is needed to add @context to .ipynb files backward-compatibly and non-breakingly.

Is this the latest draft of a JSON-LD 1.0 @context for nbformat v4?
https://gist.github.com/bollwyvl/09f0241b45fc3a98df8b

[westurner]

JSON-LD 1.1 (2020) > "Changes since 1.0 Recommendation of 16 January 2014":
https://www.w3.org/TR/json-ld11/#changes-from-10