Isn't the OSM tagging scheme is too difficult?(OSM 태그 체계에 대하여)

General talk Tagging general discussion
, , , ,
1

(It would have been nice if someone who was fluent in English could have made this opinion, but I’m not sure I’ve made myself clear enough. If anything is unclear or lacking in my comments, I’ll add it as I hear from you. If there’s anything in my writing that puzzles you, it’s most likely purely due to my poor English. Please do me a favor and use a plain language. Also, I’ll add in Korean in hopes that it will help.)
Despite my terrible English, I have posted a few questions and asked for opinions.
Only recently did the core of all those questions become clear to me.
It was whether the OSM system of tagging was too complicated, subjective, or abstract.
I believe that someone taking their first steps into OSM should be able to tag easily and that it shouldn’t be a fatal mistake.
However, since the OSM tag system is quite difficult, there are actually many tags that are incorrectly attached, and I think this reduces the value of OSM data.
I think that only a few clear, simple and intuitive elements should be organized systematically, and the rest of the abstract or subjective properties should be selectable.
For example, ‘I’ has a classification system called ‘biological-animal-mammal-human’. This is a relatively objective, systematic, and straightforward classification.
I don’t think attributes like ‘male’ or ‘Korean-born’ or 180cm tall are suitable as central classifications.
Furthermore, things like what color fur is and what color skin is are not objective at all, they are relative and complex.
And, how many years after a building is built does it become a historic building, and who decides if it is a work of art or not?
In addition, I think that OSM editors’ geographical bias leads to cultural bias, which leads to a clash of concepts. (‘Culture’ is an inevitably subjective element.)
I believe that the OSM tagging scheme should be simple, clear, objective and intuitive.
That way even beginners are less likely to make mistakes, and at least there will be fewer mis-tags, if not detailed.

I don’t think this is a problem that can be solved right now.
What I hope is that this discussion deepens and a consensus is formed, and in the meantime, if a good alternative is provided, I think it is a problem that should be attempted someday.
We look forward to your insightful comments.
Thank you.

Related articles for reference

How to tag an arch symbolising entry to area
What tags should be attached to structures that are neither gates nor even gatehouses?
Talk:Tag:man made=torii
Other articles related to the ‘tagging’ keyword

  • in Korean…
    제 영어 실력이 형편없음에도 불구하고, 저는 그 동안 몇 차례 질문과 의견을 구한 적이 있습니다.
    최근에서야 그 모든 의문의 핵심을 분명히 깨달을 수 있었습니다.
    그것은 OSM 태그의 체계가 너무 복잡하거나 주관적이거나 추상적이지 않나 하는 것입니다.
    OSM에 처음 발걸음을 한 사람도 쉽게 태그를 찾아서 붙일 수 있어야 하며 그것이 결정적인 실수가 아니어야 한다고 생각합니다.
    하지만 OSM 태그 체계가 상당히 어렵기 때문에 실제로는 잘못 붙인 태그가 많으며, 이것이 OSM의 데이터로써의 가치를 떨어뜨린다고 생각합니다.
    분명하고 단순하며 직관적인 요소 몇 가지만 계통적으로 정리가 되고 나머지 추상적이거나 주관적인 성격은 선택해서 쓸 수 있어야 한다고 생각합니다.
    보기를 들어서, 역사적인 건물은 지어진 지 몇 년이 지나야 역사적인 건물이 됩니까? 예술작품인지 아닌지는 누가 결정합니까?
    덧붙여, OSM 편집자의 지리적인 편중이 문화적인 편중으로 이어지고, 그것이 개념의 충돌로 이어지고 있다고 생각합니다.('문화’라는 건 주관적일 수 밖에 없는 요소입니다.)
    저는 OSM 태그 체계가 간단하고 명료하고 객관적이며 직관적이어야 한다고 생각합니다.
    그래야 초보자도 실수할 확률이 낮으며, 세밀하지는 않더라도 적어도 잘못 태그하는 일은 적을 것입니다.
    이 문제는 지금 당장 해결할 수 있는 문제는 아니라고 생각합니다.
    제가 바라는 것은, 이런 논의가 깊어지고 동감대가 형성이 되며 그러는 가운데 좋은 대안이 마련된다면 언젠가는 시도되어야 할 문제라고 생각합니다.
    여러분의 통찰력있는 의견을 기다립니다.
1 Like
2

People agree in general with this. The problem is that world described by OpenStreetMap is very often complex, confusing, counterintuitive and mapped by people with various biases, what makes it extremely difficult.

As result this task is quite difficult.

Taxonomy classification actually is extremely complex and tricky and far less objective than many would expect. See Species concept - Wikipedia and Ring species - Wikipedia and Giraffidae - Wikipedia and Synonym (taxonomy) - Wikipedia and https://www.criver.com/eureka/reclassification-bacteria-happens for just few aspects of complexity present also there.

Many things in OSM seemed to be having relatively objective, systematic, and straightforward classification, and then it turned out that it completely fails just few hundreds kilometers away.

+1

Far, far more effort was put into designing tagging schemes handling well objects in Europe than in Asia. It is quite clearly visible in many parts, including documentation, presets and tag use.

But it is the current state, so far!

Help with fixing this is welcome! That takes a lot of effort, for example I remember that discussion about extremely simple https://wiki.openstreetmap.org/wiki/Tag:man_made%3Dcarpet_hanger (see Proposed features/carpet hanger - OpenStreetMap Wiki and discussion on mailing lists etc) took quite a lot of time, but now there is an established way to map this feature that is relatively unique and specific to my area.

Key:shop - OpenStreetMap Wiki likely should list shop=rice and so on.

this is I think a good example waiting for someone to invent a good tagging scheme or documented something that is used and works well

4 Likes
3

That say’s it all. The real world we try to reflect in the OSM database is (not only very often, but in general) very very complex and confusing in many parts. There can be no overall simple tagging scheme covering all physical aspects (+ some none-physical like borders or routes as well) of the real world.

Again the tagging schemes have been developed over the years. There was no masterplan how to tag the world in all details. Errors have been made, duplicate tags developed, tagging schemes changed, tags deprecated and so on.

Additionally there is the problem of documentation. If there would be an experienced team dedicated and empowered to permanently work on writing and reviewing documentation things could be much easier. Such team does not exist and some mappers wouldn’t even like it, there was a recent topic dealing with this issue here.

From that point of view one has to say: as a result this task is not only quite difficult, but impossible.

By the way: Your english text is fine, no problem to understand your point, so don’t worry about that.

3 Likes
4

Let’s not forget that OSM is a great success despite (or maybe because of?) the issues.

I don’t agree that the tagging scheme is too difficult. It allows a degree of variation and uncertainty, which can be difficult to cope with and which generates friction, but the basic scheme is not that complicated.

5

We often can get closer to an ideal.

For start many features for now have no established standard tagging. Especially in areas without significant on the ground surveys by local people.

This makes marking them highly confusing.

Many established tags are not yet supported by various editors, and could be.

3 Likes
6

That takes a lot of effort, for example I remember that discussion about extremely simple Tag:man_made=carpet_hanger - OpenStreetMap Wiki (see Proposed features/carpet hanger - OpenStreetMap Wiki and discussion on mailing lists etc) took quite a lot of time, but now there is an established way to map this feature that is relatively unique and specific to my area.

how big is your area? Bars to beat carpets are not unseen in Germany and other European countries at least.

Anyway, I agree, we all want a simple system to clearly and unambiguously describe the world, but it is not immediately clear how to achieve it in a way that everybody agrees that it works, the more input there is from various contexts the better, we wouldn’t have developed detailed (?) tags for japanese bathing facility types without the Japanese.
If you feel a lot of things can’t be well represented with common tags then the reason could be there are still a lot of tags to develop. Don’t be shy or overwhelmed, one at a time :slight_smile:

1 Like
7

I assumed that this is limited to central/eastern Europe but maybe I was wrong!

+1

8

Thank you for your comments, everyone.
As you probably already understand, I don’t want the rules of OSM to be perfect (no one should, and they never will be!!!).
They are confusing and difficult to understand for beginners, late comers, or people from certain cultures, so I want to make them simpler, clearer, more intuitive, and more objective. ‘As much as it can be’, ‘a little more’ …
Mixing subjective and relative attributes with objective ones and allowing beginners to pick anything(forced to pick anything) is problematic.
At the very least, objective attributes should come first, and relative and subjective attributes should be secondary.
This is necessary for ease of use(editing), but I also believe that inconsistent conventions reduce the value of OSM as data.
Due to differences in user perception, having different tags for the same kind of feature will only lead to confusion when utilizing the information later.
In some countries, the OSM is being used quite well, but in others, there is a lot of work to be done to improve its use.
It’s hard to increase utilization if the data is not valuable as data.
(Especially in Korea, where there is a lot of government-managed information, there are quality map services that specialize in Korea, so OSM will never have a foothold if the map is just an image.)
An unutilized resource is not much different from garbage.

It’s hard to say what that should look like, but I think we should at least have the relatively objective, systematic, and simple attributes as the main focus, and the subjective, relative, and nuanced attributes that can be used for anything as a secondary choice.
If less objective, relative, and localized attributes become the norm, it’s bound to get messy and confusing later on.
Please understand that as I edit OSM, I’m becoming more and more confused as my proficiency increases.
Thank you.

9

In general I agree - though in many cases improving things may be tricky.

Artworks are really tricky, but we want to mark artworks. Distinguishing streams and rivers is highly useful - but exact border is again tricky. The same goes for road classes and so on.

10

I see it this way:

In OSM you can use any tag you like.

This is an unusual approach, and there are two ways to think about this.

You may find that it means everything is complicated. There is no one correct way of tagging something, because there are many different ways to tag the same thing. How are you supposed to decide which one is correct? Should it be building=gate or man_made=gateway? Or both? Or neither? Is it an artwork or a highway feature? You can have endless discussions on the forum about this. And the people who read the forum are the people who enjoy these discussions, so you will hear many different opinions.

But I like to think it means the opposite. Tagging is easy. There is no wrong way of tagging something, as long as you make an effort. Just pick one tag for your entrance arches or gateways, and stick to it. Convince other mappers in the area to use it too. And once it has been used a few hundred times by a few dozen people, make a wiki page. Create a github issue for your favourite renderer and ask them to show it, or if they won’t, make a new renderer that does.

At the end of the day, who cares if that small lane is a residential or an unclassified or a service? It’s on the map. Move on, go outside, map something else. Stop looking for a simple system. Embrace the complexity and the chaos :slight_smile:

3 Likes
11

Your comment is very much appreciated, but it’s lost its place.
It doesn’t matter what you tag them with anymore.
The ‘freedom’ and ‘flexibility’ of OSM, which are its greatest strengths, are the biggest obstacles to utilizing it.
For example, if someone wants to find you, they need to know something objective about you first, and you tell them something less objective or changeable, like that you’re tall, friendly, or have dark hair, they won’t be able to find you as easily.
Because of the freedom and flexibility of OSM, people who edit it end up tagging different things with the same thing, and that makes it harder to extract. One thing becomes a completely different thing.

12

From the point of view of a data consumer, having more than one tag that “means the same thing” (or more often “mean different things to the people who added them but means the same thing to me”) isn’t a major problem. What is a problem is if some people use tag X to mean thing Y and some use it to mean thing Z.

I can also think of plenty of examples where I’ve initially used tag X for something and then later found that more other people use tag Y; in that case I’ve changed my X to Y.

6 Likes
13

In some ways, I think this comment best expresses my intentions.
In the sense that it’s not just an image map, but a sum of data for variety of uses, I think these points are preventing the data from being utilized properly and turning it into garbage data (to put it mildly…)
What good is it to me if I have a very useful extension, but I can’t find it when I need it!

14

that happens but is unlikely to be caused by “any tag you like” that rather encourages divergent tagging

Specific examples of problems with OSM tagging may be helpful, in this case I am unsure what kind of tagging design you consider as problematic here

1 Like
15

@adreamy The act of trying to implement a rigid universal system would be doomed to failure, because perfect representation of all stakeholders in OSM is impossible. There would always be overrepresentation or underrepresentation of various groups and someone would cry foul. As others have also mentioned, I am certain that any attempt at a universal schema would be strained to effectively represent many endemic objects and situations. This is why the local community always has the final say. I myself, as a corporate editor have encountered tagging practices that seem extremely arbitrary or illogical in other continents and countries. What can I do? I can kindly and very politely reach out out and make suggestions for changes, but otherwise the onus is on the international editor to accept/adapt to the preferences of the locals. The localization of tags to include other languages besides English is a cause that many have taken up and are working on. Maybe you could join this effort to try to better the map of Korea if various English tagging schemas are insufficient.

1 Like
16

The perfect definition of OSM work … should be written as header on every wiki page … :wink:

17

Can you give concrete example where you’re having problems distinguishing which tag means what?
Than we can discuss that specific problem, and come to conclusion and try to document it better on the wiki or make a proposal for new things.

In other words, concrete problems can be agreed on and solved. But generalistic “OSM tagging is not done in a best way” is not actionable, i.e. there is nothing anybody can do to even try to improve it. :man_shrugging:
So, if there are hundred problems you’ve encountered, there need to be hundred specific threads open, one for each of them. Only then is there hope they can be improved: one by one.

What good is it to me if I have a very useful extension, but I can’t find it when I need it!

What exactly you can’t find, can you give example? I’ve seen several other threads, and good suggestions have been made. Sometimes, there is more than one way to do it, yes. That is not optimal, but it is not so horrible either. Use the one you like the best, or the most popular / documented one if you can’t decide.

Sure, “any tags you like” is not perfect strategy, but imagine what an alternative would be! As project originating in Europe, if OSM tags were strictly defined and centralized by some committee, it would be likely be quite impossible to map many things specific to Korea, until (if ever!) you’ve managed to mobilize huge amount of people in order to be able to pass the bureaucratic barriers needed for new tags.

What I would suggest as a strategy to improve mapping situation in your part of the world:
Translate Map features - OpenStreetMap Wiki to Korean. That is the most important tags to use (this is how we’ve done it for Croatia: Hr:Map Features - OpenStreetMap Wiki)

Then, as needed, define how to simply map things that are specific to Korea, or need to be mapped in special way, or what are recommended tags to use for which case (e.g. this is how we’ve done it for shopping-related things in Croatia: Hr:Trgovacki centri - OpenStreetMap Wiki)

Then point the users in your community to such documentation, and collaborate to improve on it. If there are unclear uses of things specific to Korea, discuss them and agree on how they should be tagged, and then document them in the wiki. If there are unclear uses for things used in the same way in the rest of the world, discuss them with wider community and document ways to map it (if more then one, document what are the differences and which way is preferred in which situations). That is the way to progress.

4 Likes