How to resolve DOI from Datacite

Hi,

The following discussion thread led to a pull request closed without being merged:
https://forums.zotero.org/discussion/26243/datacite-as-well-as-crossref-for-doi

So I ask for this feature again: would it be possible to add a resolver for the almost 13 millions Datacite DOIs (as well as Crossref DOIs)?

Maybe this (deprecated) tool could be a good starting point: http://www.handle.net/firefox_hdlclient.html.

Thanks in advance for your answer.
  • We've long done this, it's just temporarily broken, should be fixed again shortly.
  • It worked one item: 10.5075/epfl-thesis-6806

    What does Zotero rely on to resolve DOIs?
  • I am very curious about this too, from the standpoint of importing records which have non CrossRef DOIs.

    Our institution has a data repository, and each item is assigned a DataCite DOI.
    https://databank.illinois.edu/datasets

    From the list view of all records in our repository, Zotero Firefox plug-in offers "Save to Zotero (DOI)" as an option. However, it only recognized/imported 5 of 72 records. The five imported records have a related resource with a CrossRef DOI (e.g., a related journal article with a CrossRef DOI is listed in the record). Zotero imported these are journal articles with the CrossRef DOI of the related article to the DOI field. One of these records happened to have 2 related articles, and it was imported as type=document with the DataCite DOI in the "Extra" field, an no DOI in the DOI field.

    However, from the detailed view of individual records, Zotero Firefox plugin offers "Save to Zotero (Embedded metadata)" and it will import the record as type=journal article and puts the DataCite DOI in the proper field. Those DOIs do resolve back to the record in Illinois Data Bank as they should. In this scenario, Zotero ignored the CrossRef DOI of the related article (as it should). So, it works great to pull in our dataset citations with DataCite DOIs and to resolve those DOIs--perfect except for the lack of a dataset item type...

  • Zotero uses the DataCite API (data.datacite.org) to resolve datacite DOIs and then imports DataCite JSON https://github.com/zotero/translators/blob/master/DataCite.js#L71

    I think adding DataCite DOIs should work consistently now using the "Add by identifier" button. Let us know if you run into any problems with specific DOIs.

    Recognizing DataCite DOIs on pages should work (it used to work in Firefox pre-57) but we aren't able to quite make this happen on pages with "mixed" DOIs currently using connectors, so it only shows CrossRef ones.

    We know in general how to fix that, but are waiting on some metadata fixes by CrossRef before we can switch APIs for all DOI calls (to using doi.org with content negotation, should you care about the technical details).
  • You can set the item type for imported items to "dataset" using the Extra field:
    https://www.zotero.org/support/kb/item_types_and_fields#additional_item_types_and_fields

    It would be nice if the Save to Zotero button would add that automatically, but the pages there don't provide any metadata indicating that the items are datasets, not journal articles, so I believe that it would require a custom translator. I'm not sure how likely that is for the Illinois Data Bank and/or the many other data repositories. @adamsmith
  • Where embedded metatadata follows recommended practices for datasets (https://doi.org/10.1101/097196 ) using DC.type content="Dataset" (or "dataset"), Zotero will add type: dataset to Extra. Both for Zotero compatibility and generally as good data repository practice, I'd recommend following these. (Zotero currently doesn't support the JSON-LD/schema.org import yet but we'll hopefully do so some time later this year).

    I think we should be able to automatically add the type: dataset for datacite DOI direct import, too.
  • Just tried the "Add by Identifier" button.
    It imports items as "document" and puts the DOI in the "Extra" field.
    DOIs are added to the citations as URLS when you create a bibliography.

    It's great that you can import many records at once by pasting a list of DOIs in.

    I had not realized data in the "Extra" field was machine actionable--very good info to have.

    Does having the DOI in the Extra field cause any problems for adding the "type: dataset" into that field? Do you need to delimit multiple "extra" entries in some way?



  • Put multiple Extra entries on new lines, all at the top of the Extra field (or with at most one non-CSL-variable line at the top of the field).
  • edited February 8, 2018
    Does having the DOI in the Extra field cause any problems for adding the "type: dataset" into that field? Do you need to delimit multiple "extra" entries in some way?
    They need to be newline delimited, i.e.
    type: dataset
    DOI: 10.1234/5678


    There's a limit to how deep into that field Zotero will go, I think, but it'll definitely pick up 3 different fields and you probably shouldn't cram more in there.(edit: turns out this isn't true; there's no limit see bwiernik's comment below). This is a workaround, obviously, while we wait for real item model changes in 5.1, which will definitely include both a dataset item type and DOIs for all item types.
  • (You can add as many variables as you like, so long as they are all at the top of the field.)
  • @adamsmith Thanks a lot. It works fine.
  • When I import a report with the DOI: 10.4232/10.bpg.1.0 via the "Add by identifier" button, I get the following title entry.
    [{"lang"=>"de", "text"=>"Best Practice Guide for the Registration of Resources with da|ra"}, {"lang"=>"en", "text"=>"Best Practice Guide for the Registration of Resources with da|ra"}]
    Can this be fixed?
  • edited March 16, 2018
    That's a data entry error
    (see the raw data at https://data.datacite.org/application/vnd.citationstyles.csl+json/10.4232/10.bpg.1.0 )

    We wouldn't want to try to clean this up since the syntax is explicitly escaped. I tink GESIS would be the right place to report this.
  • the date field is blank in Zotero although it's provided for the following item:
    https://data.crosscite.org/application/ld+json/10.21982/m1rv-hq77
    maybe the date format ("date-parts", with only year) is problematic?
  • edited February 12, 2022
    (please delete)
Sign In or Register to comment.