osmlab / name-suggestion-index

Canonical common brand names, operators, transit and flags for OpenStreetMap.
https://nsi.guide
BSD 3-Clause "New" or "Revised" License
712 stars 870 forks source link

Discrepancy issue between JA:Naming sample and NSI in Japanese brands #7115

Open ghost opened 2 years ago

ghost commented 2 years ago

Many Japanese brands in NSI are not based on JA:Naming sample (especially brand tag). Examples: AEON and its affiliates, BOOK・OFF, SEIYU, ENEOS, COCO'S, etc.

According to the mailing list of the Japanese community, brand tags are basically to use the notation of the logo. In Japan, logos are usually written in alphabetical characters, so brand tags are usually in alphabetical characters.

However, due to the establishment of NSI, most of the brand tags are now written in NSI's notation (local language, i.e. Japanese). For POIs before the establishment of NSI, or added by mappers who edit based on the naming sample, the notation is based on the naming sample.

Therefore, I do not know whether the NSI should be modified or the Naming sample side should be adapted.

Naming sample is made among the Japanese community, so it is mostly based on a local concept (of course, there are some subtle points, since anyone can edit it). In this respect, NSI is often seen where it does not match the local way of thinking.

LaoshuBaby commented 2 years ago

In fact, seldom have mapper from Japan community contribute to NSI. Last time @blendy-map had done some changes and add 2 geojson, and I haven't remember other people ever raise this issue.

I had ask for help in OSM-Japan's slack channel, but looks like nobody reply to me: https://osm-japan.slack.com/archives/C010MC9914K/p1664029723796949 图片

In my opinion entry about local community should follow local guideline, for example, Hong Kong's entry keep bilingual because Chinese and English are both official language in HK.

And there still a lot of Japan entry haven't got a wikidata, in bus.json, in community_centre.json, in school.json, etc..

Many current east asian entry are maintained by me, about Japan and South Korea, because NSI seldom have contributor from local community. I know a little Japanese but isn't communicatable. So I think we need the help of Japan community several times.

And I also want to know other maintainer's advice.

1ec5 commented 2 years ago

Is this about romaji versus kana? Which writing system would be less confusing to users on a rendered map, in search results, etc.? We would certainly welcome corrections to make the suggestions more in line with user expectations. If necessary, the other spellings can be relegated to subkeys such as brand:ja-Japn.

ghost commented 2 years ago

Rather than "romaji vs. kana," it should be "English (or more accurately, logo notation) vs. Japanese." For example, for AEON, the brand tag would be "イオン" or "AEON". As to which notation is superior, I honestly don't think there is a clear superiority or inferiority.

Personally, however, I think that there seems to be a local rule that the brand tag should conform to the logo notation, and since the name tag is generally written in Japanese, I think that the brand tag should conform to the logo notation instead. The definition of the brand tag also states that "the brand name is easier to recognize than the store name." And I think the logo notation is easier to recognize.

However, the naming sample and NSI do not match in some areas, such as the way name tags are attached, so I think this point should be discussed as well.

One of the problems with this is that, even if corrected, it would require a major update effort at OSM, since the NSI way of writing is well-established.

By the way, I sent a mail about this for now, since the Japanese mailing list seems to be more active. I too believe that cooperation is needed especially from the Japanese community. However, it will be difficult because NSI is not well known in Japan and the Japanese community is not that active, making it difficult to contribute in many ways compared to OSM.

bhousel commented 2 years ago

However, due to the establishment of NSI, most of the brand tags are now written in NSI's notation (local language, i.e. Japanese). For POIs before the establishment of NSI, or added by mappers who edit based on the naming sample, the notation is based on the naming sample.

Therefore, I do not know whether the NSI should be modified or the Naming sample side should be adapted.

Naming sample is made among the Japanese community, so it is mostly based on a local concept (of course, there are some subtle points, since anyone can edit it). In this respect, NSI is often seen where it does not match the local way of thinking.

If this is true, I apologize.. The purpose of NSI is to allow the community to have their features to be mapped consistently in whatever way they want them to be. This is why we seed NSI with tag data that we collect from the OSM planet file - we are trying to match the tagging already in widespread use.

If the Japanese community wants their POIs mapped with name and brand tags a certain way, we should modify NSI to conform to that convention. Just let us know what to change!

Also to echo what @LaoshuBaby said above, we really need local users involved in this - I don't think we have any regular contributors to NSI who are experts in Japan naming convention.

See also #6788 - we were previously setting the name tag a certain way to avoid triggering JOSM and KeepRight warnings, but I removed this rule since that validation warning seems to contradict how some local communities (Japan, Hong Kong, and others) actually name things.

ghost commented 2 years ago

See also https://github.com/osmlab/name-suggestion-index/issues/6788 - we were previously setting the name tag a certain way to avoid triggering JOSM and KeepRight warnings, but I removed this rule since that validation warning seems to contradict how some local communities (Japan, Hong Kong, and others) actually name things.

I think that is correct. Thank you very much.

I have not received the reply yet, so I do not know what the Japanese community's policy is. Incidentally, I have actually only been a member of OSM for a few months, so I am inexperienced and don't have much of an idea of what the community might be thinking at this point.

If this is true, I apologize.. The purpose of NSI is to allow the community to have their features to be mapped consistently in whatever way they want them to be. This is why we seed NSI with tag data that we collect from the OSM planet file - we are trying to match the tagging already in widespread use.

Maybe I shouldn't be the one to say this, but I don't think you need to apologize. One reason for this series of discrepancies is probably because the data was registered with NSI before it was collected from the OSM Planet or etc. Another reason, again, because there was probably almost no contribution from the Japanese community due to a variety of factors.

Anyway, if we are going to fix it, it will require a very extensive updating process, both in OSM and NSI. In one AEON alone, nearly 250 POIs use the NSI notation. FYI, here is a history of brand tag usage rates in the case of Aeon, taken from Ohsome. スクリーンショット 2022-10-02 122653 Therefore, since there seems to be no clear superiority or inferiority at present, and since the workload is expected to increase, I think there may be the idea of aligning with the NSI notation (Japanese notation.)

ghost commented 2 years ago

As a reminder, the Japanese community is currently discussing this on Slack.