Artist Images: the ISNI option: explorations


WikiMedia editors have suggested that having an ISNI would be sufficient to allow a WikiData entry.
And MetaBrainz(?) projects are happy linking to artist images that are linked to WikiData as such images have been checked as having suitable copyright permissions.
This interview with Laura Dawson from Bowker (the ISBN distributors) gives as good an overview of ISNI as I’ve found.

But the specifics remain shrouded in mist.

I’ve found a MB artist Hannah Epperson who has been given an ISNI, and also got the results of an ISNI search for “Hannah Epperson”. (Google does not have ISNI numbers indexed :zipper_mouth_face: )
"ISNI: 0000 0004 6307 260X
Name: Epperson, Hannah
Creation class: Language material
Creation role: performer
Titles: Upsweep
Sources: CDR"
The muzieweb URL is “de muziekbibliotheek van nederland”.
Who, I conclude, is functioning as a source for, or a subsidiary of, Koninklijke Bibliotheek who is included in the list of ISNI registration Agencies.

Going by the search result, the standard that appears to have been used (by Koninklijke Bibliotheek) to judge eligibility is the existence of one commercially released album.
(The album is physical, not digital. It is not clear whether this was relevant to eligibility for an ISNI.)


ISNI and other database identifiers fit under the topic of authority control.

The idea is you have a unique identifier in a database that uniquely identifies that person / idea / thing.
If two people share the same name there can both have an entry in the database and a unique identifier.

ISNI seem to be designed for national libraries where if you publish a book they will give unique identifiers to you as the author and list what you have created.

These databases tend to link to each other so if you know one identifier you can look up the record and find it in another database.

For example Daft Punk have a musicbrainz id, an ISNI, VIAF, wikidata identifiers.
If you look up the ISNI you can find the VIAF and from this you can find wikidata.
From wikidata you can find all the other id’s including musicbrainz.


I had the impression that ISNI was rolling out at bureaucratic slowness.

But no, it appears very live.
Epperson’s ISNI page there is a “Please help us improve this record”. I submitted Epperson’s Musicbrainz ID number about 48 hours ago.
And just got an email from an ISNI Quality Team member at the British Library telling me the MBID has been added.


Dawson says that “when an ISNI is assigned, we have either computationally proven the link between the name, the metadata, and the assigned ID—or that this data has been checked manually by someone.”

I’d like to know what constitutes “computational proof”. Maybe if MB, Wikidata, ISNI, and VIAF agreed on a rubric for proof, we could automate the syncronisation of links, IDs, and other metadata.


MusicBrainz is already one of the sources they use. We already have an (older) dump of ISNIs data we could use to look for what ISNI-MB matchups they have that we don’t and check/add the ones we miss. See the last item in these meeting notes:


So we’re going to import ISNIs en masse? That’s great!


Probably not, as we’re generally not very comfortable with 100% automated edits:

Also, that meeting note is from ~1½ year ago, and I’ve been contacted once about it, I think, but no one has actually done anything with the ISNI data, so…


That said, I don’t think anyone using the Wikidata script to import ISNIs is actually checking the ISNI by hand to make sure it’s good, the users just trust that it was correctly set up on Wikidata. So auto-importing the ISNIs (and other catalogue links) from linked Wikidata pages seems to me like it would be pretty much fine.


I have found one MB artist who has a working ISNI on WP and another separate one as a ISNI search result - for no apparent reason. But this doesn’t look to be a WP problem - rather somehow 2 ISNI’s were released for Oliver Goldsmith. Given the sample size before this was found, 4?, there may have been some problems in ISNI-Land - though both ISNI’s curently point to exactly the same set of works by Goldsmith.


At some point there were multiple ISNIs for JS Bach. I think it is possible they’ve now merged the pages, but have the same thing as we do, where old merged IDs still redirect to the right entry?


ISNI records can both be duplicates, but they can also contain information for what is actually multiple duplicate people. There’s a form on the ISNI site to report this information to them when you come across it. (Be sure to mention that you’re from MusicBrainz. :innocent:)