Topic on User talk:Yurik

How much translation data is available in OSM?

2
JMatazzoni (WMF) (talkcontribs)

Hi Yurik, As you know, we're getting close to releasing the internationalization feature (T112948). I want to be able to tell wiki users just what this will mean for them. To a large degree, that will be dependent on the data that is available from OSM. In other words, we can provide the capability to show a map with labels in your language, but if OSM doesn't have labels in your language, then it won't display (unless you add it).

I know you're experienced in these areas, so I have a couple of questions, if you don't mind.

  • General question: in your experience, how much translation does OSM have? Just country and capital names and major tourist attractions? Or does it go much further? (Obviously, the biggest issue is between different scripts: I can learn that "Rue" = "Street," but if I'm looking at Chinese characters for the street names, I'm lost.)
  • Is there a way to preview how various maps would look in a different language? Changing my "preferred language" setting in openstreetmap.org, doesn't work. I see that I can click on individual map features; is there a more general way? And is openstreetmap.org actually showing everything it has?
  • Is there a way to query the system somehow and get figures on this? E.g., to know how much of each language has been added for each country? Or to say for map Paris or Rome, what is the amount of non-French and non-Italian language content available?

It's important because I don't want to oversell this feature to users and then have them be disappointed with the results. Thanks so much for your help.

Yurik (talkcontribs)

Hi @JMatazzoni (WMF), The easiest way to check how a certain name:xx is distributed is using taginfo. Here's all tags containing "name". name:en is the most common, with 2.6 million tags, and if you click it and click map tab, you will see it's fairly well distributed.

But the most other languages are far from it - name:ru, 2nd most common, seems to be only present in the eastern Europe - probably because editors there tend to add it together with name.

In general, several map companies have said they have to rely on Wikidata for internationalization. The process is fairly simple - just use Wikidata tag to lookup the name if its not present.

There is no way AFAIK to preview it, at least not on OSM site. OSM is using a fairly dated tech, and cannot easily show multiple languages.

Reply to "How much translation data is available in OSM?"