pelias / openaddresses

Pelias import pipeline for OpenAddresses.
MIT License
52 stars 44 forks source link

street name normalization: contract diagonal directionals #479

Closed missinglink closed 3 years ago

missinglink commented 3 years ago

this is an extension to https://github.com/pelias/openaddresses/pull/477, please merge 477 first.

it adds https://github.com/pelias/openaddresses/commit/e90700e4ddc0ac5dd0a955dbdc3bcfcf1993bfd2 which contracts diagonal directionals such as 'Southeast' to 'SE' which solves this issue:

Screenshot 2021-01-29 at 12 48 02

Logically it's maybe a bid odd that we expand S->South but contract Southwest->SW. I chose to do it like this because the 'long form' of the diagonal is very verbose.

Opening this as a separate PR so it doesn't block merging https://github.com/pelias/openaddresses/pull/477 Will require a rebase after https://github.com/pelias/openaddresses/pull/477 is merged.

missinglink commented 3 years ago

[edit] aagh actually this won't immediately solve the issue in the screenshot as the results are a mix of OA and OSM and we're not currently applying the same normalisation logic to OSM.

missinglink commented 3 years ago

closing: https://github.com/pelias/openaddresses/pull/486#issuecomment-902931572