[OSM-ja] Fwd: [OSM-talk] Bug fixes multilingual map

Hiroshi Miura(@osmf) miurahr @ osmf.jp
2012年 12月 3日 (月) 09:20:53 GMT


三浦です。

Pythonで、日本語のよみをromajiにするライブラリを作っています。
変換は、辞書を使っています。
https://launchpad.net/unihandecode

内部的には、KAKASIというライブラリを参考に実装しています。

http://blog.kenichimaehashi.com/?article=13048363750

Javaで実装されたKAKASIもあるので、JOSMプラグインにできるかもしれません。

三浦


On 2012年12月03日 11:47, Daniel Kastl wrote:
> Hi Douglas
>
> Right, it can't be 100% automated.
> So when I meant JOSM plugin, I thought that the automatic conversion will just be a proposal and needs to be corrected manually afterwards. But at least there would be an interface and tags didn't need to be added manually. And some conversions should be possible without problem, for example Kana to Romaji. 
> I know that Kanji can be ambiguous, but when you type you actually type the reading already and then convert it to Kanji. So the plugin had to listen to the key strokes before converting to Kanji.
>
> From personal mapping experience I just feel that adding multiple name tags with more or less identical content is annoying work. If it would be possible to do everything fully automated, it could be better processed afterwards. But because it needs user interaction and verification, it's better to have easy to use tools to make correct tagging easier.
>
> Daniel
>
>
> On Mon, Dec 3, 2012 at 11:31 AM, Douglas Perkins <douglas @ dperkins.org <mailto:douglas @ dperkins.org>> wrote:
>
>     ダグラスです。
>
>     It's very nice to see the internationalization support progress!
>
>     Automatically creating name:ja_kana, name:ja_rm, and name:en is probably not going to work.  Japanese kanji have ambiguous readings, so you can't go from them to a phonetic reading automatically.  And you can't go from a phonetic reading to kanji, because there are many kanji with the same reading.  Also, there are different ways of writing words in romaji, and currently there is not consenus on what name:ja_rm should be.  We probably cannot decide on just one romanization style, because different businesses or places may have their own preferred spelling.  Trying to automate the entry process for these tags (except for name, which can be an exact copy of name:ja), may well lead to a large number of errors.
>
>     残念ですが、name:ja だけと知っても、name:ja_kana も name:ja_rm も name:en が分からない可能性があります。
>     漢字の読み方がそんなに簡単なことではないですし。
>     ローマ字のスタイルはいろいろがありますし。
>     そうすれば、一つ一つ考えないとエラーが多くなるかもしれません。
>

-------------- next part --------------
HTML$B$NE:IU%U%!%$%k$rJ]4I$7$^$7$?(B...
URL: <http://lists.openstreetmap.org/pipermail/talk-ja/attachments/20121203/429fb238/attachment.html>


Talk-ja メーリングリストの案内