[Geocoding] Reverse-geocoding with large dataset

[Frederik Ramm]

Tao,

   like Troy said, set up your own Nominatim and you're good to go.

It is worth noting that if you reverse geocode 1.5 million GPS points
with OSM and store the result in a database of your own, then the data
you have gained from OpenStreetMap (not your original GPS points, just
the additional information that comes from OSM) is quite likely to be
covered by the ODbL license, which may affect your responsibilities in
further processing the data (for example, if you were to publish a heat
map that shows all spots where people went faster than X on a dirt
track, then you would be required to quote OSM as a data source and
release the OSM-derived part of your data on request).

This is just to clarify that if you download OSM data and run your own
Nominatim, you're free of all API restrictions but the license still
applies to the data you're using in your own database!

I'm writing "is likely to be covered" above because there's currently
some discussion in the project to what extent the license should cover
geocoding cases. But you're definitely on the safe side if you assume
that what comes out of your 1.5 million requests is covered by the ODbL.

-- Having said that, another technical detail: Since your GPS points are
likely part of GPS tracks and not totally independent measurements,
reverse geocoding them individually might yield funny results - e.g.
someone travels along the motorway but due to a GPS glitch a single
track point lies on the dirt track along the motorway. If you were to
use a track matching algorithm instead, where the most likely path in
OSM is sought for every GPS track you feed in, then such errors will
occur less often because the algorithm is clever enough to detect that
it is not possible to travel from the motorway to the dirt track and
back. Such applications can be built with the routing engines
GraphHopper or Project-OSRM for instance.

Bye
Frederik

-- 
Frederik Ramm  ##  eMail frederik at remote.org  ##  N49°00'09" E008°23'33"