OpenStreetMap
New project to do i18n on Mapnik-based maps is here - "osm-wikinames"
Posted by Branko Kokanovic on 10 September 2020 in English.I worked for some time to create Mapnik-based tileserver where names are internationalized. I managed to do it by pulling data from names from Wikidata and Wikipedia. Project is released at: https://gitlab.com/stalker314314/osm-wikinames
You can see showcases at: https://gitlab.com/stalker314314/osm-wikinames/-/blob/master/showcase.md
Idea is to create new columns in PostGIS which get populated from wikidata/wikipedia and then, use those columns in renderer.
Do check it out if you find this useful and submit issues if you see any bugs, or this doesn’t completely do the job for your case!
Discussion
Comment from jimkats on 10 September 2020 at 23:34
That’s pretty amazing concept and illustration. And of course, as you shown in the examples, it all depends on how much completed are the wiki data about any place in any language.
Serbian in this case, indeed has a weirdly wide coverage of several pages in Wikidata and Wikipedia of any language, unlike other big languages do.
Comment from Sanderd17 on 11 September 2020 at 10:46
Wikidata only helps with big features as you said (regions, cities, …). You won’t find many translated streetnames (as we often have in Belgium) in Wikidata.
I would actually prefer a thematic editor/QA tool for this: go to an area, extract the important names of boundaries and cities (perhaps even up to villages or hamlets). Get a suggestion from Wikidata to what they can be translated, and approve it (or translate it yourself).
That said, the smaller you go, many names don’t have a translation but only a transliteration to the different alphabets. It’s sometimes odd what gets translated and what not. Like in Dutch, we never translate any American city or region, but we do translate big European cities and regions.
Comment from Branko Kokanovic on 12 September 2020 at 20:42
@jimkats Thanks! BTW, curious, why do you think it has wider coverage than other big languages? I though Serbian is nowhere special here? I know couple years ago we had project to transliterate lot of French places (villages), so you can get https://sr.wikipedia.org/wiki/%D0%91%D0%B5%D0%BB%D0%B5%D1%81%D1%82%D0%B0_(%D0%90%D1%80%D1%98%D0%B5%D0%B6) in Serbian wiki, but I am seeing other “big” languages also have articles about this village?
Comment from Branko Kokanovic on 12 September 2020 at 20:42
@Sanders17 Yes, this approach will only get you so far. No point to get street names from Wikidata:) If you go with QA tool to do translation/transliteration, where you would store that? Assuming that translating French village to Romanian is not disputed (I guess there are phonetic rules for each language pair combination how to transliterate it), it would be good that everyone can benefit from this, right? Since OSM policy is not to store it to OSM, why this QA tool would not store it to wikidata (and not reinvent some third place). Then, I think, since we don’t need OSM for any of this, we get to point that this thematic editor already exists and it is wikidata editor itself:)
WRT this American/European transliteration, I think it is historical thing. Looking at my own language (Serbian), just from the fact if it is transliterated or not, you can guess if this city/area/part of the world was historically important for Serbia or not:) Most of USA is simple transliterations (English phonetic rules), but Europe can get pretty hairy etymologically:)
Comment from Duja on 24 September 2020 at 13:51
Note that it’s not generally necessary that a place has a Wikipedia article to have a useful Wikidata transl(iter)ation. It is sufficient to have its /name/ property filled in for the corresponding language, just as in OSM. So in theory, in cooperation with Wikidata folks (notwithstanding potential licensing issues), someone could import and harmonize the existing OSM /name/ tags into Wikidata.
Comment from Branko Kokanovic on 24 September 2020 at 15:13
Yes, this code first tries to read name from Wikidata, and only if that do not exist, it will try to read from Wikipedia (which itself is lower quality, and needs to be sanitized further, like a need to remove parenthesis…). Yes, moving from OSM -> Wikidata is OK (IANAL, but looks like so), but I think that Wikidata is laready “richer” in info than OSM (e.g. on those entities where we have
wikidatatag, they is usually already name in Wikidata:)