Language and location based content and categories

I think this is a good assessment. if something pertains to a specific region or country like Rio Grande do Sul in Brazil, they could mark it with a country tag.

Maybe the header can rotate between languages? It could be a feature that people could turn on and off.

Let’s hope that this proposal will put a stop to the rude behavior that consists in switching from one language to another in a given topic. That was part of what I did not like in the previous forum, and that is coming back on this one.

Maybe it should be explicitly mentioned as discouraged in the forum guidelines?

Hi all. Late to the party on this new topic, excuse my delayed reply.

Regardless of what ends up being the topic structure of this forum, to have a translator plugin is critical to integrating our global community which is a huge goal of this forum. It ought be a high priority to implement.

After bringing it up in the previous topic I heard concerns of the imperfection of machine translation (for the record, give DeepL a try it’ll blow your mind) but this is just conjecture - the only way to see if this would work for our community is to try it out.

I agree fully with the comment above. Is the point of this forum not to bring us all together? Before we segregate our community by language and silo our conversations away from eachother based upon a barrier that in 2022 can be addressed by technology, we owe it to ourselves to try a translation plugin.

After testing out the plugin, it may be clear that language-specific spaces are the only way to go forward for our community, and that’s OK! At least we tried! And, we can still keep the translation plugin for curious users who want to see what other language communities are discussing!

For sure there may be implementation issues wrt to translating topic titles or getting notifications of posts in other languages, things getting confusing with multiple languages in a category…I’m not convinced this will solve everything.

But if we are to create language-specific spaces now there is no going back to delete categories, we already know how this goes in OSM, I’ve no hope we can reach consensus to change something back after it’s already been done. Just the reality.

To not give the translation plugin a try will be a huge missed opportunity, and I see no reason why we don’t try it. There really is no risk. And, I’m happy to help in the tech side of things to implement it, just let me know.

3 Likes

Went ahead and created another topic specifically about the translation plugin, as it’s not just relevant to forum category structure but the site as a whole.

Testing the translate plugin is something we would like to do eventually, right now we have still a few big things to sort out with the Help OSM and old forums transition.

We already know that a translator plugin is not going to replace language specific spaces, as some have already commented here. Thinking from a major European-language perspective it’s easy to think that a machine translation can solve it, but there are a ton of African and Asian languages where machine translation is not doing a great job.

Testing a translator plugin is not incompatible with enabling people spaces in their languages where there is nothing in the middle doing a okish-to-poor job adapting their tone and expressions.

We also have some commenting here that it could work, if only we tried. Furthermore to support the language plugin and to have language-specific spaces is not mutually-exclusive, however to start the forums with language-specific spaces is an irreversible decision and ought not be made with haste. Is this a community decision or an admin decision?

Perhaps this is something we can consider once we try it out? As opposed to assuming. Especially try DeepL, it’s really that good. Do we have any non-major European language peoples here who can help us understand that perspective? How does language segregation help these folks as opposed to regional categories and allowed to post in any language they want, with ability to translate?

Enabling people spaces in their own languages can also happen if we go by location-based (which in cases of many non-major languages, is a proxy for a language space) instead of language-based. And in the case of a post in a location-based category in a non-major language there is no need for translation between speakers, only in the case of a non-speaker wanting to understand.

I’m all for people having spaces they are keen to post on, I just hope to do so with least fragmentation as possible and I think that can be done without segregating based on language, if only we make the small effort to install a plugin and just give it a try.

@mDav well since you mention it. I think theoretically, this is something that can be 100% solved with technology (with the known caveat that auto-translation is not perfect).

The issue is that the currently available translator plugin for discourse (GitHub - discourse/discourse-translator) does

  1. only translate when triggered, i.e. not automatically everything but you have to click on a (small) button next to the Reply button to do that - for a single post. This button can easily be missed. In contrast, comments in Google products usually do it the other way round - show the auto-translation and have a button to show the original. I think this is better, but costs more money of course.

  2. More importantly, the translator plugin does not support translation of the topic title as far as I know. But exactly this title should always be translatable - in fact, it should always automatically be translated because it is most important for finding content.

  3. Since the translation is manually triggered, I don’t think it is well integrated with search which of course would be important in the long run for discoverability.

The plugin is freely licensed and could be forked so that it can be improved to meet our requirements better. These improvements in the fork could be contributed back upstream (if the maintainers want such features, no way to ask since they have the issue tracker deactivated). The plugin is written in ruby, a simple and easy script language.

Edit: It looks like their “issue tracker” is their own forum with everything tagged with translator - Topics tagged translator (quite nice that there is such a permalink for tags: it’s <forum url>/tag/<name of tag>)

2 Likes

@westnordost - thanks for that! I too believe that it is possible to customize discourse / translator to adapt for our specific community needs and remove the need for segregation based on language. We do have quite a few developers around these parts, and the OSM website is also written in Ruby, so it’s not totally out of the question. What’s more the question is if the “powers at be” agree to support it or not.

If I were to dream we could for surely implement a way to translate topic titles, auto-translate posts, and only send notifications for topics in a user language, etc… and solve all the potential issues with using the translator plugin which have been brought up so far.

However I hasten to talk about what-if’s at this stage and at this point wish to focus on the low hanging fruit, which is just installing the translator plugin as-is (or, even install this fork with DeepL support) just so the community can see the potential here and decide for themselves whether it’s worth investigating. I myself am not even convinced it will work, but I am convinced it’s worth a shot!

I continue to be surprised by the lack of interest from the admin team to implement such a plugin at this stage considering how low effort it is to install (soooo easy according to the docs). Just give me 10 mins of ssh access and I’ll do it myself, even.

It remains unclear to me what the overall category structure we’re going for here, but I like what you mention about tag permalinks, why not have languages as ISO-2 tags for each post (ie fr for french), which is auto-assigned by the language detection of the translator? Then all french posts are at /tag/fr ?

There is so much potential to tailor fit this Discourse to our unique community, if only we open ourselves to that possibility before it’s too late.

I see a bigger problem here. Especially from the perspective of a European, a translator based on Google, Microsoft, Amazon, etc. is not necessarily so compliant with the GDPR. Here you also have to be careful…
I have no experience with the quality of LibreTranslate. DeepL is powerful and very good, but there is also a cost factor.

GDPR should not be an issue here.

The translation is done by Discourse from the server itself, so the external translation API only knows what is send to it by Discourse: text to translate, server IP address, nothing else about who wrote that text.

Don’t forget that the same text is publicly available for anyone with much more informations :wink:

3 Likes

Implementing this plugin has a cost per translation, that we still haven’t discussed and also we haven’t evaluated any other requirements. The current tools we are transition over here don’t have this functionality and that’s why we would like to control the scope of the current work and don’t add additional requirements yet.

Also, Discourse allows us to organize and reorganize topics as we need, and we are currently leaning towards allowing an organic growth, opening categories when the communities need it. The forum governance team is meeting at the end of this week to digest all the feedback and take some decisions.

I think it would be important to hear in this discussion from people whose main language is not a major European one, and understand if they need o not a language-based space, and not block them because others think a translator plugin is enough.

I’m from Sweden, but as many here /we understand and speak English reasonable good. The younger people better then we old (I’m 67). The traffic in the current Swedish forum is very low, and I suspect this partly because we don’t need a specific forum for our language. Besides that, the most active osm group here is on Facebook.

So I assume an auto translation Swedish<–>English is not important for Swedish osm mappers.

Lol, had the same thought. Appreciate GDPR for being the first thought though, it’s important stuff! FWIW, I know at least DeepL is GDPR compliant.

You mention opening categories when there is organic community desire…but not about closing them, or merging them. That’s what I mention here, once we create language-based categories there is no going back. Not that it’s technically impossible, but that at least this community has such trouble with consensus it will just not change once the categories have been made.

It’s like the ele tag in OSM, it should be elevation but there’s too much use and inertia already so change just ain’t gonna happen, point blank. We here have learned to be cautious and consider all angles before adoption of new things so as to avoid situations like this.

What’s wrong with starting the categories as not language-based, and then if it really does become cumbersome, communities can request we create them later on?

By not opening language-based categories at the outset isn’t blocking communities from creating their own categories. As you said we can open categories as needed down the line. This is about the structure which has been created from the outset, which is an admin decision not a community decision and is why I bring up this topic now.

Plus, without having installed the translator plugin these folks are not able to make an informed decision of whether language-specific categories are necessary or not. We have at least one dissenting opinion from a non-major language speaker here who doesn’t think language-specific categories are necessary (thank you for your input @Msiipola !) so imo that warrants more discussion as opposed to a hasty decision.

I quite agree… with the opposite. And later it seems to be what you meant: I think a tag for the language should be sufficient, at least in the beginning.
It’s the way the Wiki is defined with automatic links to similar entries in other languages.
And then inside a section for the region/country specialties.
Key:priority - OpenStreetMap Wiki has automatic links to similar pages:
čeština Deutsch English español français italiano polski português](https://wiki.openstreetmap.org/wiki/Pt:Key:priority) 日本語
Here the French page has a specific section for France as the Vienna signs displayed on the English page:

Priority over oncoming traffic (B6) Priority for oncoming traffic (B5)
Vienna Convention road sign B6.svg Vienna Convention road sign B5-V1.svg

Is translated as follow:

Signalisation routière en France

Les véhicules prioritaires voient ce panneau : France road sign C18.svg

et ceux devant leur céder le passage sont avertis par ce signal : France road sign B15.svg


(No need to have a different layout, but it’s real example)

In order to decrease the risk of silos, it’s better to “split” just by language so that tags invented in Germany are also tested in Austria, Switzerland, Belgium, etc… then proposed globally.

BTW (*) we CAN’T and SHOULD NOT expect that let say a Spanish use a ES:Bicycle) tag without help. Bicicleta or ES:Bicicleta is more realistic. Don’t oblige people for requesting help in their own language to know the terms (tags!) in British English.

Yes Deepl is a tool tool. Not perfect, but a good one. It is missing some (!) languages, for instance if you want to translate from/to Ukrainian, you can use SimplyTranslate with the Google engine or the Reverso engine.
I believe we should have at least the 6 official UN international languages. From w:United Nations: The six official languages of the UN, used in intergovernmental meetings and documents, are Arabic, Chinese, English, French, Russian, and Spanish. Even if today there is not much traffic in Arabic, creating a place for this widely spoken language is important, even if it’s at first an almost empty shell.

Data Items may help here. @pyrog?

(*) By the way: try to avoid cryptic conventions too :face_with_hand_over_mouth:.

1 Like

Arabic is indeed in the Top6 of languages by speakers: List of languages by total number of speakers - Wikipedia with 274M total speakers, while Russian for example is 8th (258M), both long after Hindi (602M). As you can see, on a global scale also Bengali (273M), Portuguese (258M) Urdu (231M) and Indonesian (199M) are big languages.

If you look at the list of languages by native speakers, 27 languages have more than 50M native speakers and 91 languages more than 10 million.

1 Like

In fact KISS: use the languages on the main page of the Wiki:
Afrikaans asturianu azərbaycanca Bahasa Indonesia Bahasa Melayu bosanski brezhoneg català čeština dansk Deutsch eesti English español Esperanto euskara français Frysk galego hrvatski interlingua íslenska italiano kréyòl gwadloupéyen kurdî latviešu Lëtzebuergesch lietuvių magyar Nederlands norsk occitan polski português română shqip slovenčina slovenščina srpski (latinica)‎ suomi svenska Tiếng Việt Türkçe Zazaki Ελληνικά български македонски русский српски / srpski українська հայերեն עברית العربية فارسی پښتو नेपाली বাংলা සිංහල ไทย 한국어 中文(简体)‎ 中文(繁體)‎ 日本語

1 Like

My point is, let’s not block communities to request a language-based space if they want to.

If Germany or Italy just need a location-based category it’s fine, but if Spanish, Arabic or French prefer to also have a language-based category, let’s allow that.

2 Likes

wouldn’t it be possible to do some orthogonal categorization with both, location and language, maybe with tags? This way one could for example follow everything in German, regardless of location, or everything in German in Italy (German happens to be a recognized minority language in some areas of Italy).

This is always possible no matter what category organization we end up with, it’s just a matter of how you ensure each and every topic is tagged with its language, which is a manual task unfortunately.

1 Like