Science news articles are labeled "Dataset"

10.1126/SCIENCE.ADD1068 resolving to monkeypox-outbreak-questions-intensify-cases-soar seems to be some kind of science news article from Science. In Crossref this is listed as a “dataset”, search with from_ui=yes&q=10.1126/SCIENCE.ADD1068 There are multiple other of these articles that are labeled “dataset”, searching with q=AAAS+Articles+DO+Group&from_ui=yes

I was not expecting that the article would have the type “dataset”, but perhaps the type “journal article” or similar.

I use our Scholia service to enter data into Wikidata via the DOI. Scholia uses the citation-js Javascript libraries that queries the Crossref XML data via the API. When I use this approach, citation-js constructs the entry as an instance of dataset, see wikidata:/w/index.php?title=Q112116175&oldid=1645355163

It is unclear to me whether this is an error or strangeness in Crossref. Or whether it is possible for Crossref or Scholia to do anything about it.

Our issues are tracked at github:citation-js/citation-js/issues/163 and github:WDscholia/scholia/issues/1981

For what it’s worth, this is by the decision of the publisher, not by CrossRef.

Please see also the thread by @sckott recently for more information:

Hello @fnielsen,

Crossref metadata is provided to us by our publisher members. The publisher of the DOI has decided on the content type for those DOIs I am afraid. Crossref cannot make changes to the metadata unless instructed in some cases by the publisher of the works.

Do you have a list of the DOIs that you think should be categorised with a different content type. I could try contacting the publishers for more information.

Many thanks,

A search for “AAAS Articles DO Group” returns about 1.8 million records, and an API query for “AAAS Articles DO Group” in query.container-title (,title,URL,type,issued&query.container-title=AAAS%20Articles%20DO%20Group) returns 400179 results. I’m pretty sure the latter results are the items with incorrect metadata. E.g. 10.1126/article.39958 ( is an article published September 3rd 1997 in the journal Science (presumably), but is listed as a dataset published August 13th 2021 in “AAAS Articles DO Group”.


Following up on this, it looks like news items are in “AAAS Articles DO Group”. Research articles are registered as articles with good metadata (abstracts, references, ORCID IDs). For example,

Crossref doesn’t have a content type for news items but the news items metadata could be improved.


I missed this reply, but thank you. That’s unfortunate.