Topic: How forum detect automated translation? (Read 526 times)

This is true for the most smaller languages, where google translate has not enough content to fine tune the translation. For some of the languages, it's not possible to use the google translate becuase of the poor performance...

Well is very simple, generally the mod of the section know the language where the moderate (I hope) in that case if you report a message translated with google or other tools they can read and find the errors done by Google, for example on the Italian section we sometimes have f....ed up translations done clearly with those tools, verbs are often wrong from the translator.

Been a little busy yesterday to reply to this topic. I have approached both r1s2g3 and erikalui for helping me out in revealing the fake translators and the result was Overview of Fake Hindi Translators in the Indian Board. Probably most of them were aware of the fake translations taking place in the Indian Board, but not much reported them nor the action were taken by moderator and the DT to tag them or remove those topics. r1s2g3 triggered the wrong translation work and erikalui spent their time to provide the evidences. The translators are getting out of control in the board, and even though if I report those posts they are not being removed. I do mention my topic as a reference in the comment while I report but to the contrary they are not being removed.


I am aware that Benson wasn't really taking any measures when he was a MOD to prevent the abuse and with the increase in the ICO's and Bounties in 2017-2018 the fake translators have risen enormously in the Indian Board too. The translators are receiving easy money by just shitting all over the board with their filthy translation spams.

As far as I have figured out Greek moderator mitzie and other global mods like rickbig handles my reports most of the time. I am quite sure few of the local board mods are acting as a global moderator in the shadows without having an official title. Since there isn't really a moderator for the board right now, we need to depend upon global mods and DT people by providing enough evidences and helping them to clean the board.

The translated topics in the Indian Board are quite easier to find out since they are most probably being translated by Google Translators which doesn't add any real meaning to the topic.

---

Situation 1 :

The Indian Board is literally very terrible right now, but they are atmost readable to an extent. I report quite often and within a day or two they would be handled. Brand New Members who are not Indians are using the situation (A board with no MOD) to promote their shit. At times, a group of spammers just like the Chitauri clan rush inside the board and spam with the same messages unless they are being reported and nuked.

Situation 2 :

2 weeks back, a group of Brand New accounts came to the board, necro-bumped 3-4 year old dead topics and posted a bitcointalk link. Once I clicked through the bitcointalk link, they linked to a fake phishing Electrum software download. All of a sudden I reported all these topics (should have been 5+) and they were removed the following day. The problem is that, we Indians are diversified and very few of them are non-spamming people. Most of them indulge in Bounties, Scams and spamming the forum endlessly in the Altcoin Subs.

Such situations happen quite a few times in the Indian Board too but they can be left aside. I have seen a user posting translated announcements in more than 3-4 languages but they seem to translate only one project. So they should have been an employee of the project and post topics related to the project which cannot be considered as a translation abuse.

P.S If theymos cannot appoint a MOD, he could better delete the board. It is becoming worse than the Off-Topic and Archival board 

A request was made in Meta here: https://bitcointalk.org/index.php?topic=3125174.20

But no mod was elected as it is said that Indian section doesn't have much activity and since many languages are spoken here, the mod may not be able to handle such reports. I can see some other Local sections also without a mod like the German section and hence members are handling such issues and reporting them.

I tagged one translator after checking his recent thread. this user  rangnatht clearly using Google translator for translations.

I will check rest of them tonight and will proceed accordingly.

^
No mod ?
That's a terrible situation to be in. Before Halab was named mod of the French section, a few of us had to take care of that "fake translators" problem. We wasted a lot of energy there.
I suppose you made a request to Theymos about having a new one ?

I used to report to Benson first who was the Indian mod but he also used to take no action and it will be tough for the global mods to deal with it as they don't understand Hindi. Only once one topic got deleted for using automated tools but that was because the project manager complained I guess else all others go unnoticed.

That is why you don't see an French moderator moderating the Arabic local board, because that will result in terrible judgment, there is no way for someone who doesn't speak the language to know how was a text translated unless it's super obvious.

automatic translation can be accurate in terms of understanding , you take a text in French, translate it to English, most of the time you will get a "clue" of what is it about, but going from English to French and posting the results - it is going to be ugly.

a week or two ago, there was a topic translated to Arabic about some new ICO , and OP was asking for "prime minister"  ( i think he was looking for a manager) but google failed him.

if the Moderator did not understand Arabic, he would not have deleted that post.

Another option is also reporting the user to the bounty manager for the project he translated for as well as to DT members like members already mentioned in the thread. That way he can get blacklisted and reconsider his actions in the future. 

If we are talking about other languages translated to English literally anyone can spot the grammar mistakes of a word for word translated text, if someone has detected the text then the user can always report it, either way users found this way are either plagiarizing or just posting in English-language topics. If we are talking about English text translated to foreign languages it's the same thing since any native language user can detect the awful translation and report it right away.

The main problem are Bitcointalk ANNs of shitcoins, often done by lazy bounty hunters using automated translation tools to translate for various languages they don't know and get more tokens. As Lafu already said, we have a dedicated topic and the points posted by Halab are a very good start.

In addition I'm going through the post history of the user and it's suspicious if he translates into various languages where no connection is between (like Chinese, German, Arabic and Russian) and never did a normal post in a discussion thread of my language.
A direct comparison between the original ANN, the translated version and test-translated using Google Translate is also very helpful, like we did it here: https://bitcointalksearch.org/topic/m.43246222 or here: https://bitcointalksearch.org/topic/m.40965515
In German we have also some words where automated translations can be spotted, like "Bergmann" (means mineworker in German) (Miner) or other gibberish.

I don't see a single point as sufficient proof on it's own until it's not very clear. Most patterns are similar for fake translators and easy to spot when you know where you have to search.

As a result we list fake-translators because using automated translation tools is not allowed according the bounty rules. They are bounty cheaters and get a red tag if they are doing it repeatedly. Normally I give them some doubt and start tagging them after they have done 3 fake translations.
That's important to warn the bounty managers to stop hiring these scumbags for the next bounty.

ANNs of devs who are posting automated translations are just getting their topic removed because they don't cheat a bounty.

The problem is that we don't have a moderator in the German section because phantastisch was removed in November 2018. A native speaking moderator would be essential to ensure a correct handling of fake translations.

Yes, .Heisenberg_Hunter  requested to check the translation. I guess you can ask someone having DT powers to tag these translators.
You can still "Report To Moderator" , it will go on Global Mods queue in absence of local mods but I have no idea how they will be able to take action when they also do not know the language.

Heisenberg_Hunter has mentioned the list of translators who have used automated tools and I also helped him out. Still nobody replied regarding this and plus don't know which moderator to report to as there is no Indian mod now.

Google Translators performs better with machine learning (I believe Google actually use it for this tool), so it's hard to automatically use other tools to find fake-translations from users if they use provided translation from Google Translator (especially for popular language, as @hugeblack pointed out). That is why we need moderators for local boards.
However, another thing will appear. Moderators can not do it themselves, they don't have time to do it, and they even should not spend most of their time to do it. It belongs to community responsibility. Members of local boards should report such fake translations if they find them.
Another component that results in abundant of fake translations is laziness of developer team whom come from (mostly) scam projects. They simply use their free money (from their scam projects) to pay for fake translators. They even don't check, don't care about quality of translations.

You should distinguish between automatic translation and some language errors.
Second: Google Translator gives excellent results for *Most Spoken Languages* ​​and weak content of unknown/unused languages:

1. Translate some well-known words: This is an example of what happens when you translate some well-known ANN words:

We notice errors in the translation of Snapchat and Telegram because the words (Snap= شبح  Chat = محادثة) have several meanings.

2. Some shortcuts: peer to peer (P2P), A.K.A, ....etc
3. Full-text translation: Google Translate may not understand some words so it leaves it as it is, causing a problem understanding meanings.


4. The direction of symbols: Some languages are written from right to left and vice versa.

I think the easiest way is if the native speaker who reported the fake translation describes those words/sentences that are weird/wrong and why he/she thinks that the translation has been automated.

I've reported such translations before, here is an example:
The translation to Bulgarian, but many words were in Russian.

Somebody asked me to check few ANN in Indian section to check whether they are automatically translated . (Unfortunately , I never looked too deep on it.)
But my initial investigation was pointing that in Indian section, even native speakers are doing automatic translation to save the effort required in manual translation.

I sometimes see these being posted in our Croatian board. Usually ANNs or a thread advertising a website or a service. The forum cant detect these translations on its own, its the users of that particular board who report it and based on those reports the post gets deleted.
I dont know what kind of checks the mods do to check if the report is valid and automated translation tools have been used. Mods any more info on that?

I can see many users using automated tools for translating content but since there is no moderator in the Indian section, nobody checks these posts. Reporting them makes no difference as some of them change two-three words here and there to make it unable to detect the difference. There are some translators who claim to have a team but those team members all use automated tools in most bounties.

Easy to spot for a native speaker.
If you have to read the same sentence 3 times to understand it, you probably have a case of someone using a translating tool.
Some of the expressions used are also obvious proof of a fake translator. Some of them are just grotesque.
But a fake translator can also be a native speaker, one lazy fucker who just did his work very fast and didn't care much how it would look.