Name-suggestion-index: `npm run wikidata` is adding duplicate P8253 Name-Suggestion-Index Identifier claim values

[Robot8A]

See https://www.wikidata.org/w/index.php?title=Q98456614&action=history

[bhousel]

Yes, that is what the script does.. Why do you want to remove it?

[bhousel]

Oh sorry, I realize now what you mean. It's adding duplicates and it shouldn't do that.

[Identitaet]

So looking around it seems to only be a problem for this item (as far as I did look at least)

[peternewman]

It's not because there are two entries for the brand:wikidata is it:
https://nsi.guide/?t=brands&k=shop&v=confectionery&tt=Q98456614

But both linking to the same item, compared to e.g. Tesco, which have that, but not in the same category:
https://nsi.guide/index.html?t=brands&k=shop&v=convenience&tt=tesco

Nope, because M&S does that properly:
https://nsi.guide/index.html?t=brands&k=shop&v=department_store&tt=marks

Although that is the only one I can quickly find with brand and category matching identically.

[bhousel]

The issue seems to be caused by this line:
https://github.com/osmlab/name-suggestion-index/blob/56b287fbaec00c3d8c1d0eb02abb733cee51c04a/scripts/build_wikidata.js#L365-L366
Where we fetch all the existing P8253, the "simplify" part is oversimplifying them and only returning the first one. 😆

[bhousel]

Turns out that simplify was only returning one because the first one has a "preferred" rank. This makes sense!
The fix is to pass the keepNonTruthy: true option, to ensure we get all the claims...

[Robot8A]

Thanks to everybody!