Issues with OpenRefine Reconciliation for Terms With Differing English-P and American English-P

0 views
Skip to first unread message

ella thompson

unread,
Feb 6, 2026, 2:39:25 PM (2 days ago) Feb 6
to Getty Vocabularies as Linked Open Data
Hi!
I am running into some trouble trying to use OpenRefine's Add Columns from Reconciled Values functionality when working with some AAT terms that have parents whose preferred English and American English terms are distinct. For instance, for aat/300037029 lances (cavalry weapons), I can't use the /aat_hier property to add a new column because one of the terms in the hierarchy is aat/300180692 staff weapons, which has a different preferred term for English and American English. When I try to add the column, I get this response:

 {"errors": [{"status": 422, "source": {"details": "['rods for torah scroll', 'staff weapons'] is not of type 'string'"}, "title": "Invalid data", "detail": "Malformed or missing data"}]}

It seems like the Getty reconciliation service is looking for the preferred terms of either English or American English for each entity in the hierarchy, and when there are multiple candidates it is unable to handle the resulting list. Using the following SPARQL query, it seems that there may only be 39 such terms:

SELECT ?concept ?gvplabel1 ?gvplabel2
WHERE
{
   ?concept a gvp:Concept .
   ?concept xl:prefLabel ?gvplabel1, ?gvplabel2 .
   FILTER ( regex(STR(?gvplabel1), "-en$") )
   FILTER ( regex(STR(?gvplabel2), "-en-US$") )
   FILTER ( !contains(STR(?gvplabel2),STR(?gvplabel1)) )
}
ORDER BY ?concept
LIMIT 100

Thanks so much for your help!
Best,
Ella
Reply all
Reply to author
Forward
0 new messages