2014-05-24 12:58 GMT+02:00 Peter Wendorff <wendo...@uni-paderborn.de>:
> Hi Christian, > The BANO data is licensed under ODbl, but the database does not conform > to the Contributor terms, right? > > That's why importing or adding it to OSM is not possible IMHO. > > BANO is not designed to be used as an import source for OSM. It uses sources (opendata, cadastre) that can be imported in OSM. These sources have been used over the past years and will still be used to add addresses in OSM, one at a time, or street by street which will take years before completion. So, BANO is there to allow to use all the available address data right now without having to wait years to get them cleanly added to OSM. This has one good side effect... reduce the pressure on massive address imports in OSM and give us the opportunity to improve the quality thru survey prior to add it to OSM. Adding it to (the osm installation of) nominatim on the other hand is a > bad idea, I think: > IMHO nominatim.openstreetmap.or should rely on OSM as a data source as > pure as possible. Adding different data sets might be useful to add > information not possible or out of scope of the osm database (like > importance factors using the wikipedia links), but BANO is a different > thing. > > If the BANO addresses are free and open enough for OSM, we should > encourage to work on bringing them to the OSM database itself (provided, > they are GOOD enough as well). How should anybody in France get > motivation to collect more addresses for the osm database itself, if > anything is available in BANO already? > > I'm pushing the french community to be more motivated on having all streets and all streets name in OSM (BANO allowed us to detect 40% missing streets or street names) more than adding addresses without surveying them. I agree to motivate to collect BETTER address data... it we collect the same data with the same quality level it looks to me more losing our time than anything else. > Nominatim as well as the maps presented by the project on osm.org IMHO > should show the strengths of OSM, and not lie about it by presenting > stuff that's not part of OSM (like addresses from different databases). > > regards > Peter > > P.S: Of course this is not meant to deal with the software as such, but > for the installations/instances "officially" part of OSM. Feel free to > set up an BANO nominatim server to show the strength of BANO instead. > In that case remove geonames and TIGER address search from osm.orgNominatim instance. I agree not to add these address data in the osm.org mapnik rendering to avoid confusion. I'm planning to add BANO dataset to improve OSM-FR tiles. 2014-05-24 13:23 GMT+02:00 Frederik Ramm <frede...@remote.org>: > > I think that Nominatim already has a built-in process where you can take > TIGER shape files and it runs them through an shp-to-osm processor and > then imports them into the local database just as if they came from OSM; > and I think there's even a supersede mechanism that will make sure that > house numbers that come from OSM are given preference over those from > the separate repository. So I believe it should not be difficult to > modify the TIGER-specific import code to conver BANO data too. > > That's what I was thinking. > Maybe it's worth coordinating with those that run the American > OpenAddresses database (Ian Deees?) who might have similar ambitions to > plug their data directly into Nominatim. > The is a BANO pull request on openaddresses.io to list it as an address data source. -- Christian Quest - OpenStreetMap France
_______________________________________________ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
