[Talk-us] Make Nominatim results smarter - The search for New York City Metropolitan Area does not return any results

Fri Mar 10 09:47:14 UTC 2023

*What did you search for?*New York City Metropolitan Area
https://nominatim.openstreetmap.org/ui/search.html?q=New+York+City+Metropolitan+Area
<https://mailtrack.io/trace/link/54bb1b404eec9d5120cae81ad84fe2926c9762d4?url=https%3A%2F%2Fnominatim.openstreetmap.org%2Fui%2Fsearch.html%3Fq%3DNew%2BYork%2BCity%2BMetropolitan%2BArea&userId=3485220&signature=716f81543ac08ce4>
*What result did you expect?*
I believe returning something is better than nothing.
To get at least New York City or to get at least something.

>From what I see for metropolitan areas there are around 10 countries in OSM
that specify what admin_level means metropolitan areas (usually 7) based on
the wiki
https://wiki.openstreetmap.org/w/index.php?title=Tag:boundary%3Dadministrative&uselang=en
<https://mailtrack.io/trace/link/03afbac6cbba4a0e61a84d30827681587265f8fb?url=https%3A%2F%2Fwiki.openstreetmap.org%2Fw%2Findex.php%3Ftitle%3DTag%3Aboundary%253Dadministrative%26uselang%3Den&userId=3485220&signature=902c8326121cae98>
There are a lot of Metropolitan areas in the world and there are valid use
cases where the user would search the city name and include Metropolitan
area.

As of February 28, 2013, the United States Office of Management and Budget
(OMB) defined 1,098 statistical areas for the metropolitan areas of the
United States and Puerto Rico.[14]
As of 2018, there are 74 zonas metropolitanas in Mexico. 75.1 million
people, 62.8% of the country population, live within a metropolitan
area.[13]
Pakistan has nine metropolitan areas with populations greater than a million
There are 30 officially defined "metropolitan municipalities" in Turkey.
https://en.wikipedia.org/wiki/Metropolitan_area
<https://mailtrack.io/trace/link/d4ce33f652af9ef3e403a897847a7659ebd245e4?url=https%3A%2F%2Fen.wikipedia.org%2Fwiki%2FMetropolitan_area&userId=3485220&signature=bc860aeee47b9df2>
In the end, I will provide some examples of how other companies are
returning data for the following search. New York City Metropolitan Area

✅ Here

curl 'https://geocoder.ls.hereapi.com/6.2/geocode.json?apiKey=
<https://mailtrack.io/trace/link/12035644f3253ee1d7be9640e716186267055837?url=https%3A%2F%2Fgeocoder.ls.hereapi.com%2F6.2%2Fgeocode.json%3FapiKey%3D&userId=3485220&signature=32faf244c82ddd16>$API_KEY&searchtext=New%20York%20City%20Metropolitan%20Area'
{"Response":{"MetaInfo":{"Timestamp":"2023-03-10T00:20:23.754+0000"},"View":[{"_type":"SearchResultsViewType","ViewId":0,"Result":[{"Relevance":1.0,"MatchLevel":"street","MatchQuality":{"City":1.0,"Street":[0.85]},"Location":{"LocationId":"NT_Ww9Q.27G21bpO1B9Vbd0rB","LocationType":"point","DisplayPosition":{"Latitude":40.60584,"Longitude":-74.11999},"NavigationPosition":[{"Latitude":40.60584,"Longitude":-74.11999}],"MapView":{"TopLeft":{"Latitude":40.60644,"Longitude":-74.12003},"BottomRight":{"Latitude":40.60456,"Longitude":-74.11955}},"Address":{"Label":"Area
Pl, Staten Island, NY 10314, United
States","Country":"USA","State":"NY","County":"Richmond","City":"Staten
Island","District":"Emerson Hill","Street":"Area
Pl","PostalCode":"10314","AdditionalData":[{"value":"United
States","key":"CountryName"},{"value":"New
York","key":"StateName"},{"value":"Richmond","key":"CountyName"},{"value":"N","key":"PostalCodeType"}]}}}]}]}}%

   1. ✅ Apple Maps suggests the City

[image: image]
<https://mailtrack.io/trace/link/7e55f442fc5158ef82191464b7e36f59da090c9e?url=https%3A%2F%2Fuser-images.githubusercontent.com%2F2330999%2F224191480-4b4c731c-4278-4ab2-b7c7-168c7b35710c.png&userId=3485220&signature=0861e8a377dc4194>
3. ❌ OpenCage suggests a city in Italy for the search

[image: image]
<https://mailtrack.io/trace/link/e179b2af78135f46dbfa956c190541bfe7b3b03b?url=https%3A%2F%2Fuser-images.githubusercontent.com%2F2330999%2F224191811-f01b3ec0-c2bd-4d1e-9736-da2e8c4c9d26.png&userId=3485220&signature=18f177640cbfc9c7>
4. ✅ Google Maps suggests a pin in the middle of New York City

[image: image]
<https://mailtrack.io/trace/link/25fd78052bd653bb4e6c4ea722d3f239b9d0b243?url=https%3A%2F%2Fuser-images.githubusercontent.com%2F2330999%2F224192259-c359bd1e-bf04-4ee3-bdf1-4754ccb7338b.png&userId=3485220&signature=f92c6afded4b7c77>
5. ✅ Bing Maps suggest a centroid in the middle of the metropolitan area,
probably the most accurate.

[image: image]
<https://mailtrack.io/trace/link/bd5f7bdd4abdc530f62ea5558b7a40c507b986e3?url=https%3A%2F%2Fuser-images.githubusercontent.com%2F2330999%2F224192666-3de47f39-f11a-49f9-8f47-2da5574423fd.png&userId=3485220&signature=a9caa9131e39d64f>
6. ❌ openrouteservice show a lot of locations all over the world

[image: image]
<https://mailtrack.io/trace/link/76a4d594a21d509cabfc6762f6f30a9cda47966f?url=https%3A%2F%2Fuser-images.githubusercontent.com%2F2330999%2F224193316-d1242639-51ff-4098-9634-8330ec4faceb.png&userId=3485220&signature=45f0b47e3d8ada05>

>From the Open Source Geocoder almost all of them fail to give relevant
results 7. Photon shows a wrong first result https://photon.komoot.io/
<https://mailtrack.io/trace/link/091aedc2108efd7ba17e5b7d37036e815f365f4e?url=https%3A%2F%2Fphoton.komoot.io%2F&userId=3485220&signature=b46f205f0df12add>[image:
image]
<https://mailtrack.io/trace/link/4617c44ec409a0b15b7367129b3461549c1ee6c3?url=https%3A%2F%2Fuser-images.githubusercontent.com%2F2330999%2F224193754-ffdece1b-5180-4462-94aa-4403a613fc46.png&userId=3485220&signature=9777c19bc80486c3>
8. ❌ Geocode Earth

[image: image]
<https://mailtrack.io/trace/link/c363bc6ec586b5171f63b88bc8f0089d788abeb0?url=https%3A%2F%2Fuser-images.githubusercontent.com%2F2330999%2F224194115-4d18b65f-05e2-46e3-bbb3-61d7b530deb9.png&userId=3485220&signature=a82ee6612605ce77>
9. ✅https://www.oscar-web.de/
<https://mailtrack.io/trace/link/084a5702da94b3817f9622fc72254c510bbf7a62?url=https%3A%2F%2Fwww.oscar-web.de%2F&userId=3485220&signature=7084463a575f4f40>
Mostly by luck it shows a point in Jersey City

[image: image]
<https://mailtrack.io/trace/link/6909dfba3f3aaf70cbe07d4bd2f8e36dc16b217e?url=https%3A%2F%2Fuser-images.githubusercontent.com%2F2330999%2F224194256-81308cab-636f-4ee7-84b0-4e8cbcbe1a18.png&userId=3485220&signature=404137dcc8b747c1>
10 ❌ [Gisgraphy](https://github.com/gisgraphy/gisgraphy
<https://mailtrack.io/trace/link/d1e96c8694c58f88fd7d7e1a2521d933ef14e4e7?url=https%3A%2F%2Fgithub.com%2Fgisgraphy%2Fgisgraphy&userId=3485220&signature=8ad3a7709c05760e>)
([demo](https://services.gisgraphy.com/public/geocoding.html
<https://mailtrack.io/trace/link/fd35620e75fe30403ec59a83f384b082194900e2?url=https%3A%2F%2Fservices.gisgraphy.com%2Fpublic%2Fgeocoding.html&userId=3485220&signature=d0b2f87b58346b78>
))

[image: image]
<https://mailtrack.io/trace/link/8b4f802a940d2c286cbd50f35dabfe8a272cd805?url=https%3A%2F%2Fuser-images.githubusercontent.com%2F2330999%2F224194650-055488f6-3f72-4618-9fa3-d4cef6cee05c.png&userId=3485220&signature=7eafed2e1ff90a9b>
Other Geocoders 11. ✅ OsmNames
https://osmnames.org/#q=new%20york%20city%20administrative%20area
<https://mailtrack.io/trace/link/2ce8ef1fe597aa8034d07802da27b07e45adafae?url=https%3A%2F%2Fosmnames.org%2F%23q%3Dnew%2520york%2520city%2520administrative%2520area&userId=3485220&signature=9b19f799e50314b3>

[image: image]
<https://mailtrack.io/trace/link/cc4d59d2b60fc04e3b98b6d12fa11decd70f1325?url=https%3A%2F%2Fuser-images.githubusercontent.com%2F2330999%2F224195071-11ca565c-1761-457b-ad27-3d6f7f64ec5c.png&userId=3485220&signature=8d0b1d27b1f86258>
12. ✅ [LocationIQ](https://locationiq.com/
<https://mailtrack.io/trace/link/eb5bc445c634eaced004ee5249a5058d6995b4b8?url=https%3A%2F%2Flocationiq.com%2F&userId=3485220&signature=08844774af393ec9>
)

[image: image]
<https://mailtrack.io/trace/link/44983fb890605d62b525e74455a3d3b3bd2b9dcf?url=https%3A%2F%2Fuser-images.githubusercontent.com%2F2330999%2F224195349-2b3b8ca5-c66d-4738-a4cd-5cca889f09fe.png&userId=3485220&signature=7f959938e50e5a9e>

This is a discussion that I wrote from an issue that I started here, but
was declared out of scope as nominatim policy is if it is not in OSM, we
should not give an answer to it.
https://github.com/osm-search/Nominatim/issues/3002
<https://mailtrack.io/trace/link/030acdecd34c4bf1d3ffec6be5e16f9a771451b6?url=https%3A%2F%2Fgithub.com%2Fosm-search%2FNominatim%2Fissues%2F3002&userId=3485220&signature=9784b79f33423d79>

Also, we do not learn anything from what people are searching on nominatim
in the name of privacy. I would argue that we could provide just the most
searched queries, which are searched at least 1000 times in a year and that
do not return any results.
This would allow us to learn popular searches that real people are
searching for, with a focus on the searches that appear a lot of times and
nominatim returns 0 results.

Florin Badita
Founder, OpenStreetMap Romania FB community, Corruption Kills
TEDx Speaker | Forbes 30 Under 30 Europe | Senior Landecker Democracy Fellow
Book a time to have a small chat
<https://calendly.com/florinbadita/small-chat>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-us/attachments/20230310/6fcf0e82/attachment-0001.htm>