search

pelias / openaddresses

Pelias import pipeline for OpenAddresses.
MIT License
51 stars 43 forks source link

unit-number-extractor: remove unit number from name.default #504

Closed missinglink closed 2 years ago

missinglink commented 2 years ago

this PR implements https://github.com/pelias/openaddresses/issues/503

missinglink commented 2 years ago

acceptance tests look good, happy to merge this once there's a corresponding PR for pelias/labels

missinglink commented 2 years ago

this may also require a change to the deduplication logic in order to avoid merging all unit numbers into a single record:

Screenshot 2022-02-28 at 17 51 48
orangejulius commented 2 years ago

I have to admit I kinda like the new results with no unit numbers. Let's merge this and talk about how to deduplicate addresses with unit numbers separately?

Maybe we do something clever like deduplicate all the units into one record unless the query looks like it specifies a unit number itself?