Unexpected authors in results: bug or misunderstanding?

[Matthijs De Zwaan]

Dear all

I am looking at co-authorships between my institution and a number of Japanese universities. I select only articles, preprints, books and book chapters, mainly to exclude paratext. The query is here; I download results to a json using the API.

Looking at my results, I see a lot of Norwegian looking author names, for example 'Aktuelt Per Halvorsen', https://openalex.org/authors/A5094738184. I see those names coming up both in my list of authors from my institution and in the list of Japanese partners. On the author page, the author has only two works, both of which are labeled paratext. The Tidsskrift for Norsk psykologforening comes up a lot in the works for these suspected authors, which seems to be a Norwegian language clinically-oriented journal, but I don't understand Norwegian very well. Finally, if I explicitly add a suspected author to my query, I get zero results. This is confusing to me. I suspect that the paratext publications discuss/abstract some of the papers in my results and somehow end up in my results file. Can anyone here explain? Do I misunderstand how the data is constructed, or is this a bug that is worth reporting to OA?

Other author ids that I believe should not be in my data, but are:

'https://openalex.org/A5094738195',

'https://openalex.org/A5094738164',

'https://openalex.org/A5094738183',

'https://openalex.org/A5094738200',

'https://openalex.org/A5094738188',

'https://openalex.org/A5094738188'

Thanks for your time and effort!

Matthijs

[Samuel Mok]

If you open the list of works by   'Aktuelt Per Halvorsen', https://openalex.org/authors/A5094738184 (this url: https://api.openalex.org/works?filter=author.id:A5094738184), you'll see he is matched with 2 works in the OA database. Both of them are of the mentioned journal. If you look at the list of authors in the 2nd work, you'll see the issue: it is a long, long list of authors, with weird and garbled names and 'raw_affiliation_strings'. This is because this item is an entire 'journal', if you click the doi you'll see a long list of articles w/ authors and affiliations. Those are all added to this item, but not properly -- so this has introduced quite some errors in the database. 

This is clearly a parsing error, it'd be good to let OpenAlex know through the forms on the site.

Cheers,

Samuel

--
You received this message because you are subscribed to the Google Groups "OpenAlex Community" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openalex-commun...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openalex-community/85ae68bd-bf91-4640-b932-ce38b95bc097n%40googlegroups.com.