Priority of tagging style in NSI

[ghost]

Often, NSI has a different style of tagging than was originally used or the actual situation. While this is unavoidable, it should be corrected as much as possible.

But in #7792, it is rejected due to the presence of Wikidata. However, it is not acceptable why it should be rejected despite the existence of the original tagging, when there is no consensus that Wikidata should take precedence.

I was originally in favor of using Wikidata, and I'm not saying that everything should be done the same way, but shouldn't it at least take into consideration the tagging that was originally used and the actual situation in those areas?

[bhousel]

I still don't understand why you're making such a big deal about this Steak Gusto thing.

[UKChris-osm]

What needs to be corrected? Why is using Wikidata not acceptable? What is the "actual situation"?

"Gusto" and "Steak Gusto" seem to be slightly different restaurant types run by the same company, which is something the NSI has catered to for a while, and is something I think is being done correctly in this instance, similar to "Tesco" and "Tesco Express", where both are owned and run by Tesco, but have distinguishing features that separate them, even if only slightly, such as selling many of the same items, but with a reduced range, different prices, or different special offers.

If the company itself are currently using "Gusto" and "Steak Gusto" separately, with a separate name, logo, menu, map locations etc, then I don't see what's wrong with the way NSI is offering presets?

The NSI won't always be correct, but it is only a suggestions dataset for users, to speed up adding known brands. A mapper should still be ensuring they are adding to the map correctly. If some mappers are tagging a POI as a "Gusto" when it should be a "Steak Gusto" (or vice versa) then I can understand the annoyance, but I feel if that's occurring it should be something discussed with those mappers making a mistake rather than simply branding them all under one umbrella.

[Cj-Malone]

Rel: my ramblings a few weeks ago about store types/brands https://github.com/alltheplaces/alltheplaces/issues/4772

Correct me if I'm wrong @Lonerat, but you're considered that brand has been redefined from "generic brand" to "specific brand"?

I don't think I was around when brand became a thing in OSM, and I definitely wasn't part of the discussion. So I can't comment on it being redefined or the original definition. But it seems to me brand should very much be specific to the POI. Maybe operator or owner are more useful for this?

[soprano1125]

Personally, I have no objection to following "JA:Naming sample", but I feel uncomfortable about changing even the value of brand:wikidata.

The reason is what @UKChris-osm -san said "If the company itself are currently using "Gusto" and "Steak Gusto" separately, with a separate name, logo, menu, map locations etc, then I don't see what's wrong with the way NSI is offering presets?" (Everyone has said what I want to say, so I will only comment on this.)

I'm curious about how the brand of "Steak Gusto" listed in "JA:Naming sample" was decided on "Gusto", but since it doesn't seem to have changed since it was posted, I think it's a good time to propose a change(Applicable history).

---

By the way, "JA:Naming sample" has a brand item, but there is no brand:wikidata item, so there's no need to go into detail about it, is there?

[ghost]

Why is using Wikidata not acceptable?

As I have said, I am not opposed to using Wikidata itself. But I think it is problematic to override the tagging in order to use it.

you're considered that brand has been redefined from "generic brand" to "specific brand"?

That's almost similar to what I intended, but maybe rather, "the brand that is generally recognized" to "a specific brand".

To begin with, as stated in the OSM Wiki, initially, it is the name tag that is expected to describe the actual store name/business type, and the brand tag is originally used when the brand name is more recognizable (of greater interest to data users) than the store name.

So, in the original way, for example, in the case of "Komeri", there are several Komeri business types such as "Komeri Power" and "Komeri Hard & Green", but the most commonly recognized by the public is "Komeri", which is also broadcast in TV commercials as "Komeri". Therefore, shouldn't the brand tag be Komeri, the main brand? And the store name/business type can be represented by the name tag. This is almost the same for Steak Gusto, and the concept in Naming sample is probably following this.

In the case of the Wikidata approach, the name tag and the brand tag will have almost the same value in most stores, so the brand tag will not be very meaningful. Also, whether it should be done in the future or not, it does not seem to be accepted yet in OSM to describe them all in Wikidata. In the first place, the two datasets have different origins, purposes, and target ranges, so it would still be unreasonable to make them link together to that extent in the current situation.

Therefore, it is not in accordance with the "actual situation" that this cannot be expressed within the OSM, is it?

My opinion on the brand:wikidata tag change is as I said in #7792.

I still don't understand why you're making such a big deal about this Steak Gusto thing.

So this is a general issue and I am not obsessed with Steak Gusto.

By the way, "JA:Naming sample" has a brand item, but there is no brand:wikidata item, so there's no need to go into detail about it, is there?

If you are saying that brand:wikidata should be preferred because it is not in the Naming sample, I would disagree: OSM is a different data set than Wikidata, and unless its use of Wikidata is explicitly defined, the use of the brand tag should not be compromised. If you mean to change only the brand tag, as in my compromise, I agree, but I have some concern that it will be different from the actual value, as stated in the brand tag and the brand:wikidata tag.

[ghost]

There are other brands with similar cases. I already wrote, but for example, Big Boy and Steak Big Boy. Using conventional practices, the brand tag for both should be "Big Boy" as well as Komeri and Steak Gusto. This is because, in general, the recognized brand name is "Big Boy". "Steak Big Boy" is a later derivation, and some "Big Boy" have switched to "Steak Big Boy". Even if there are distinctions such as menus and business categories, the perception of the brand itself remains much the same.

Even in the example of Gusto, there was a brand called S-Gusto, although it is no longer exists. This should also be the same way.

Another example, not exactly like these, but an example of the use of brand tags, is a mobile phone store. https://github.com/osmlab/name-suggestion-index/blob/98a0d04dfbb8ec192743ec855640b81cfbc6045c/data/brands/shop/mobile_phone.json#L1455-L1474 For example, in this case, the brand tag is currently the same as the name tag, but the recognized brand is "docomo" (or "NTT docomo"), not "DoCoMo Shop". Naming sample is also done the same way.

Just to be clear, I am not saying that "the same brand tag should definitely be used when the same management company operates a brand with a similar type of business".

For example, there are Hard Off, Off House and Hobby Off. These are operated by "Hard Off Corporation" and often operate in the same building, but they should be branded differently. This is because they are generally recognized as different brands, and they are also generally broadcast separately in TV commercials.

[soprano1125]

I think that there are more than a few people who feel uncomfortable when looking at the icon when editing with an editor such as "iD", even if the value of wikidata has been changed. (I feel that NSI sets icon images linked to Wikidata and SNS icon images linked to them.)

Talking about "Steak Gusto" again, wouldn't it be weird if the "Gusto" icon appeared when you added the "Steak Gusto" POI? To prevent this, you should suspect that Wikidata is something else. If Wikidata equates "steak gusto" with "gusto", I think we need to discuss how to handle icon differences.

For the time being, it seems uncertain whether the current NSI can solve the problem, so why not consider whether it can be achieved with NSI 7 (#7387)?

[ghost]

I agree with the problem that @soprano1125 pointed out.

I believe this is already an issue with some brands that are already registered, such as Komeri and Viva Home. Viva Home is not to that much, but at least in the case of Komeri, each business type has its own logo, so it would be useful to be able to display that. Right now, the icon of a Facebook announcement account dedicated to gardening is applied, and it is confusing because it shows the logo of what doesn't exist as a business type.

I am not opposed to about attaching a wikidata itself (rather I think it should be connected to a specific brand wikidata as long as it does not override the tagging). So, for example, I thought it would be most desirable if there was a tag that could connect Wikidata to the store (business type) itself, rather than the "brand".

And it would not be impossible to solve it by getting the icon from Wikidata attached that tag. My other suggestion is to make a feature/setting on the NSI and editor software side that can override the icon. However, if possible, it would be good to be able to attach Wikidata for both main brand that is generally recognized and specific brand so that data users have more choices.

[ghost]

For now, at least for brands in Japan that are already registered in NSI with a specific brand Wikidata, may I suggest that when modifying them to match the originally used tagging or other, only the brand tag should be changed, as in @soprano1125's opinion and my compromise?

[ghost]

Although not a restaurant, in this example, brands may be represented within the brand tag by separating them by semicolons. So for example, we might be able to describe as brand=ステーキガスト;ガスト. If there are no objections, I might create this as a draft PR.

[bhousel]

So for example, we might be able to describe as brand=ステーキガスト;ガスト. If there are no objections, I might create this as a draft PR.

Sorry, we don't support semicolon separated brand tags in this project..

[ghost]

So for example, we might be able to describe as brand=ステーキガスト;ガスト. If there are no objections, I might create this as a draft PR.

Sorry, we don't support semicolon separated brand tags in this project..

If so, then I think only the brand tag (not brand:wikidata tag) should be changed to Gusto. As for the machine, it can read Wikidata to get both information if needed. I don't much see the need to put the Steak Gusto described in the name tag into the brand tag.

[bhousel]

Again.. Steak Gusto and Gusto are 2 different things as far as our index is concerned. They have different logos, different menus, but most importantly - different Wikidata. This project identifies brands by their Wikidata QID.

It's the same as how McDonalds Q38076 and McCafe Q3114287 are 2 different things.

The brand tag is just a human readable string that lets people know what the brand:wikidata identifier points to.

[ghost]

As I have said, the brand tag is used within the OSM database, not necessarily for Wikidata. So shouldn't it not necessarily be used to indicate what the brand:wikidata tag refers to? Also the original tagging should not be ignored for that.

[bhousel]

Also the original tagging should not be ignored for that.

Nobody is "ignoring original tagging"! In fact, we collect the original name tagging from the OSM planet and use it to seed this index.

As you can see, we've collected both names are used by mappers. These numbers are updated as of 2 days ago. https://github.com/ideditor/nsi-collector/blob/main/dist/osm/names_all.json

[ghost]

No, it has been ignored. The original tagging is JA:Naming sample, and we already know that there are actually multiple conflicts with NSI. This issue is what happened in fixing it in the first place.

The brand=ステーキガスト started being used because of the start of NSI and there is no consensus with the community about it originally.

Most of those that have been added now are mostly updates due to editor warnings, and there is not much awareness that they are due to NSI.

[ghost]

Nobody is "ignoring original tagging"! In fact, we collect the original name tagging from the OSM planet and use it to seed this index.

To begin with, when this item was collected by Planet in the first place, only the name tag was added, not the brand tag or anything else. Then, the brand tag was mechanically added (4164fbd) with the same value as the name tag.

I'm sorry if I have come across as somewhat accusatory, but in fact, this does not mean that the original tagging was adequately taken into account/not ignored, and it can't be a sufficient reason why Wikidata should be prioritized, even for brand tagging.

Therefore as for the priority of tagging, the tagging that was originally used should be considered first, although of course it should be modified each time there is a problem or improvement. (So I still have my doubts that that PR was closed in the middle of a discussion where opinions were being expressed)

Then a decision should be made as to how to match the brand tagging.