Update DOI generation instructions (author info) for HF

[egrace479]

I did not realize that CITATION.cff worked on Hugging Face, but it seems (Select "Cite this dataset" under the DOI on this repo) that including one may override the auto-generated DOI citation (e.g., on TOL-10M). In which case, I think it's a good idea to include for HF as well.

I still think including the BibTeX is useful since it will provide pre-formatted version of all citations (the actual repo, paper(s), references, etc.), but if simply including a CITATION.cff ensures proper citation dropdown when a DOI is generated, that's great.

Originally posted by @egrace479 in #25 (comment)

Todo:

• Test if adding a CITATION.cff to a HF dataset repo works to control the DOI dropdown citation (will require adding the file to a dataset or model with a DOI).
• Update guidance accordingly.

[hlapp]

Nice find!

[egrace479]

It doesn't seem like the CITATION.cff will have the desired effect, but Hugging Face just added the ability to list authors when generating a DOI (see announcement).

So, it doesn't seem like it would be worth making this recommendation, but we should note how to get a proper author list when generating a DOI on HF.

[egrace479]

I have tested the DOI generation for a new dataset now (STRI-Samples) and it seems to be working precisely as intended.

[egrace479]

There does not appear to be a method of seeing earlier "revisions" outside of the settings section, so we should probably include some kind of note/guidance for referencing earlier versions as appropriate (ex with 2018-NEON-Beetles).

It also doesn't seem to require changes to the repo to generate new DOIs, offering another option of just generating a new one for earlier repos that had "HDR Imageomics Institute" listed as author instead of the actual authors.

[egrace479]

Actually, we link to the Hugging Face DOI documentation (on DOI page), which is pretty straightforward (including a screenshot of the actual pop-up with adding authors).

I'd say the only remaining question is if we want to address either of these two points, otherwise I'd say we can close this.

There does not appear to be a method of seeing earlier "revisions" outside of the settings section, so we should probably include some kind of note/guidance for referencing earlier versions as appropriate (ex with 2018-NEON-Beetles).

It also doesn't seem to require changes to the repo to generate new DOIs, offering another option of just generating a new one for earlier repos that had "HDR Imageomics Institute" listed as author instead of the actual authors.

@hlapp?

[hlapp]

I think we can leave first of the two points in the queue until we have a case (or cases) to better understand when this might be encountered. If it's very rare (which it might well be), I'm not sure it's worth addressing.

For the second point, I'm unsure what would be both hard to find and also at least somewhat commonly encountered about this to make it worth documenting.

So in short, I think we can close this for now.