[Geocoding] [OpenStreetMap] #5175: Inappropriate translation of names in Nominatim search results

[OpenStreetMap]

#5175: Inappropriate translation of names in Nominatim search results
------------------------+-------------------------
 Reporter:  stanton     |      Owner:  geocoding@…
     Type:  defect      |     Status:  new
 Priority:  minor       |  Milestone:
Component:  nominatim   |    Version:
 Keywords:  i18n, l10n  |
------------------------+-------------------------
 When using the search function on the main OSM page, names in search
 results appear to use foreign names.

 In my case, I was searching for "supermarket, Colmar" (located in Alsace,
 France). My locale and browser language settings are both English/US but I
 was accessing it from an IP which reverse geocoding probably resolves to
 Germany.

 I got several results, all from Colmar and surroundings. All of them
 consistently used German toponyms for a number of places (Colmar itself,
 the département and the région). Examples:
  * Supermarket Intermarché, Route d'Ingersheim, Kolmar, Oberelsass,
 Elsass, Metropolitan France, 68000, France
  * Supermarket Monoprix, Passage Monoprix, Kolmar, Oberelsass, Elsass,
 Metropolitan France, 68000, France
  * Supermarket Aldi, Avenue Joseph Rey, Kolmar, Oberelsass, Elsass,
 Metropolitan France, 68000, France

 In my opinion, this behavior should be altered for a number of reasons:
  * '''Avoiding confusion:''' Toponyms can vary greatly between languages.
 While Colmar/Kolmar is a fairly easy one, nearby Sélestat/Schlettstadt is
 less obvious, and with cases like Rijeka/Fiume or Klaipėda/Memel it is
 nearly impossible to guess without sound background knowledge that both
 names refer to the same place. Even I don't know all the exonyms which my
 native language has for places outside its borders, and am more likely to
 know (and use) the local name – the exonym is more likely to confuse me
 than to help me.
  * '''Politics:''' Toponyms can be a politically sensitive matter, and we
 have seen numerous discussions in the OSM community on the content of the
 ''name'' tag in bilingual regions. Especially in territories that have
 changed hands multiple times over the course of the centuries, usage of
 exonyms can be perceived as a territorial claim – and OSM is certainly not
 the place to do that.
  * '''User choice:''' Users have only limited choice regarding their
 public IP address on the Internet. Furthermore, the location of a user
 does not necessarily reflect their language preference – think of expats,
 travelers, multilingual areas or people who are fluent in multiple
 languages. Thus, there are better options than IP-based reverse geocoding
 for guessing the user's language preference. (Granted, that's what Google
 does as well, but if Google were without flaws, we might as well use their
 maps and would have no need for OSM.) One is the browser language
 preference, transferred in the Accept-Language HTTP header. Browsers
 typically set it to the UI language, unless the user overrides it (which
 is a decision made by the user and thus should be honored) – so this is
 usually a fairly good indicator of the language the user wishes to
 communicate in. Another option is a language selector on the web site
 itself, which a lot of websites do.

-- 
Ticket URL: <https://trac.openstreetmap.org/ticket/5175>
OpenStreetMap <http://www.openstreetmap.org/>
OpenStreetMap is a free editable map of the whole world