Wrong "is instance" for preprint

[fnielsen]

For instance, https://www.medrxiv.org/content/10.1101/2021.08.15.21262067 results in manuscript (document written by hand). https://www.wikidata.org/w/index.php?title=Q109622851&oldid=1529788551

I would have expected that is should be something else, e.g., preprint.

[fnielsen]

I forgot to check whether this is already fixed. We are seeing the error in Scholia WDscholia/scholia#1704

[fnielsen]

Current master:

$ citation-js -t https://doi.org/10.1101/2021.07.28.21261159 | python -m json.tool | grep '"type"'
        "type": "manuscript",
                "type": "@doi/api",
                "type": "@csl/list+object"

[larsgw]

The CrossRef API is (still!) returning invalid CSL-JSON. They are using the type "posted-content" (should be "article") with the subtype field (undocumented, they should use genre) set to "preprint". At the time it was a bit unclear what "posted-content" should be matched to, and I used this mapping. I'm still not sure if I can always map "posted-content" to "article", who knows what other subtypes there might be.

[fnielsen]

Would this be better: https://www.wikidata.org/wiki/Q2376293. It is also call a "manuscript".

[larsgw]

The CSL documentation says this about manuscript

An unpublished manuscript; Use for both modern unpublished works and classical manuscripts; For working papers, preprints, and similar works posted to a repository, use article.

So I think the issue lies more with the CrossRef data. I can make a workaround in Citation.js, though I'd prefer CrossRef fixed their API of course.

By the way, the mappings to Wikidata are based on the exact match (P2888) statements (query)

[fnielsen]

In Scholia we got manuscript: 'Q87167'

This should/could be manuscript: 'Q2376293',.

I am not immediately able to see where that line comes from in the citation-js code

[larsgw]

This seems to be fixed in v0.5.3, not sure why I kept it open.