[OSM-ja] Osmose-QA - Question about multilingual tags
Douglas Perkins
douglas @ dperkins.org
2016年 6月 26日 (日) 14:44:45 UTC
On 06/26/2016 10:31 PM, Frédéric Rodrigo wrote:
> Le 26/06/2016 à 10:14, Douglas Perkins a écrit :
>>
>> On 06/26/2016 02:15 AM, Frédéric Rodrigo wrote:
>>> Hello
>>>
>>> We are working on run Osmose-QA world wide, and Japan is one of the
>>> last country to fulfill the coverage. Osmose-QA is a quality tools
>>> reporting issues about OSM data, like Keeright or OSM-Inspector.
>>>
>>> http://wiki.openstreetmap.org/wiki/Quality_assurance
>>>
>>> http://osmose.openstreetmap.fr/ja/map/#zoom=11&lat=34.8174&lon=137.0197&layer=Mapnik
>>>
>>>
>>> We have a question about multilingual tags. Do you have some rules
>>> about usage of name, name:ja, name:ja_kana, name:ja_rm ?
>>> In particular in Osmose-QA we check if the default name is same as
>>> local name (name:ja).
>> Dear Frédéric,
>>
>> Please see these two pages for rules about names in Japan. Probably
>> it will answer most of your questions.
>>
>> https://wiki.openstreetmap.org/wiki/Japan_tagging
> As I understand, name must me equal to name:ja, English in parenthesis
> is deprecated but tolerate (automatic translation from Japanese from
> the wiki is not so good).
Ah, yes. Since 2013, the policy is that we do not put English in
parentheses. Before then, it was acceptable, and much of the old data
has been updated. According to that policy, there's "no need to overdo
it to fix the old names". Slow improvement.
>
> From this rule, when name is not equal to name:ja, Osmose-QA must
> report an issue ?
Maybe. I personally don't think so. See below.
>
>>
>> https://wiki.openstreetmap.org/wiki/JA:Naming_sample
> Some examples have name equal to name:en and not to name:ja.
> So in Osmose-QA must we check name is equal to (name:ja or name:en) ?
> Alternatively, can you provide a list of preferred name from English
> rather than Japanese ?
>
>>
>>> name = 沖浦 (Okiura)
>>> name:ja = 沖浦
>>> name:ja_kana = おきうら
>>> name:ja_rm = Okiura
> Must be reported as an issue.
>>> name = 広島県道75号三原竹原線 (Route Mihara Takehara)
>>> name:en = Route Mihara Takehara
>>> name:ja = 広島県道75号三原竹原線
> Must be reported as an issue.
>>> name = 新広島幹線 (Shin Hiroshima Kansen)
>>> name:en = New Hiroshima Trunk Line
>>> name:ja = 新広島幹線
>>> name:ja_rm = Shin Hiroshima Kansen
> Must be reported as an issue.
>>> name = セブン−イレブン東城川東店
>>> name:en = Seven-Eleven
>>> name:ja = セブン-イレブン
>>> name:ja_rm = Sebun Irebun
> Must be reported as an issue.
>>>
>>> name = カインズホーム (CAINZ HOME)
>>> name:en = CAINZ HOME
>>> name:ja = カインズホーム
>>> name:ja_rm = Kainnzu Homu
> Must be reported as an issue.
>>>
>>> name = ENEOS
>>> name:en = ENEOS
>>> name:ja = エネオス
>>> name:ja_rm = Eneosu
> Not sure about this one.
This is the tricky one. For some businesses (like ENEOS or Mobil),
signs and storefronts use western letters. So "name = ENEOS" and
therefore "name:en = ENEOS". For "name:ja" it's weird to use western
letters, so instead the Japanese phonetic alphabet is used.
There's nothing on the Wiki going into detail about this, AFAIK, because
this kind of situation is relatively rare.
In my personal opinion, the above ENEOS example is reasonably tagged,
and if your algorithm is set to accept name equals name:ja or name
equals name:en, you will minimize false positives. Perhaps some others
will provide their thoughts, too.
>
> Regards.
> Frédéric.
>
>
> _______________________________________________
> Talk-ja mailing list
> Talk-ja at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk-ja
Talk-ja メーリングリストの案内