-17

During review audits sometimes questions are written in a foreign language and determining what would be the fairest outcome for the person requires the auditor to either be fluent in a foreign language or copy/paste the text into an online transaction service.

Some questions could be migrated to one of our foreign language sites which is a friendlier approach than trashcanning their thoughts. Other questions are either spam or otherwise unsuitable and should be deleted.

In order to ease the auditor's job I request a feature where a translated version would be available either underneath the question itself or via a link.

My complaint is that I was reviewing questions on Stack Overflow and encountered a question in a foreign language. Per our policies I chose "Requires Editing" because the English language is to be used and our CoC requires us to "be nice" and "welcoming" - it doesn't obligate us to translate questions.

I failed the audit because (it claims) I should have rejected it as spam.

  1. We are not obligated to translate - if it was SPAM in English I would have rejected it, it is the obligation of the writer to have a suitable question.

  2. Telling people their words are spam when they are not in English is rude.

As best I can determine the URL for the question is: "Contratar Produtora de Vídeo em São José dos Campos, o que devo me preocupar?" by André Carletti - it's 404 to me. Perhaps higher level users can see 404s.

I'm not overly concerned about failing one audit, I'm concerned about the fairness and an audit that may be against policy. A link to an automated translation would speed the auditing process, reduce errors and permit us to helpfully suggest migration of useful questions. If the question was not useful it could more easily be determined if deletion was the fairest course of action.


Foreign Language isn't SPAM

  • 11
    Regardless of if it is spam, or not, you failed because you did not do what you are supposed to do with non-English content. How do I deal with non-English content? is quite clear that it should be closed ("Unsalvageable") as "Unclear what you're asking". – Makyen Sep 8 '18 at 2:31
  • @Makyen - Not a good example, the chosen answer is prior to the new CoC and the next highest votes doesn't say that. --- This is tagged "Bug" and "Discussion" so I appreciate your opinion as to whether it's a bug or not. --- Should we discuss and decide it's not a bug I can pull those tags and make a feature request for (partial?) translation of foreign languages so we can clearly see if we should suggest it be migrated or if it should simply be deleted. – Rob Sep 8 '18 at 2:50
  • 2
    You are welcome to post another question about translating non-English content (this question has an answer and shouldn't be morphed into something else), but such a question will almost certainly be closed as a duplicate. You appear to be claiming that you shouldn't have failed the audit when you've obviously gone against the current and long-term established consensus. If you want the consensus to change, then try to change it before you act against it. The new CoC didn't actually change anything wrt. dealing with non-English content, nor almost all policies/consensus. – Makyen Sep 8 '18 at 3:52
  • 2
    As to characterizing the post I linked as "Not a good example", my opinion is that you are wrong, and your stated reasoning is flawed. You mention the next higher voted answer ( < 1/4 the score), but that answer says nothing in support of users translating posts for OPs. The only thing that answer does is to provide some additional information in the form of some links. Some links are to sites where non-English content is OK. The other links are to the FAQ here on MSE, which says English only, and the 2009 Blog, which also says English only. – Makyen Sep 8 '18 at 3:52
12

No, you're supposed to reject posts in a wrong language to the site. Requires editing means "The post is almost OK, it just need some formatting improvement". It should NOT be used when the post needs substantial editing from its author, including but not limited to "need more information / clarification" or "need translation". Editing (from others) is supposed to make a post look better, while maintaining the integrity of the post.

While you're not required to understand any other language than English, you're expected to ensure that you know what you're seeing when reviewing. At the very least, a proper action for this post would be Unsalvageable → should be closed → unclear what you're asking, if you failed to identify that it's spam (which an average reviewer could, given so many links to one single website).


For a case-by-case analysis, here's the revision of that post:

If you look carefully, all the three links goes to the same website. Two with bare domain (like http://example.com) and the other with a tail (like http://example.com/foo-bar). Furthermore, if you Ctrl-C/V the post body into Google Translate, the result would be (linked text marked in bold).

I need a video producer to do everything from video recording to editing all the material in Adobe Premiere software, as well as managing these videos on YouTube, Instagram and other social networks that have video. What is the best software that the producer should use? The Premiere itself, or Vegas, or another?

The producer must be from São José dos Campos, or from São Paulo, to have proximity and get to move when needed, to record.

Which software is suitable for short videos that I use on YouTube and Instagram? I will also use IGTV videos from Instagram, so it needs to be adapted for each channel after ...

Thank you!

Consider this in English. All three links goes to the same website, with linked text being irrelevant to the question. This is just a typical pattern of spam.

  • Now that I see the translated version it seems off topic. SPAM would be offering to do the work. Trying to hire someone is one of the services that StackExchange offers to employers. This person might have heard that there are talented people at the SE sites and you can hire them, they tried to ask in their own language and the reply is: "Get lost SPAMmer, we're not interested." Real nice. – Rob Sep 8 '18 at 2:41
  • 2
    @Rob By definition, any unsolicited promotion is spam. Spam intrinsically is a wide concept. It includes random advertisement, excessive or malicious link to another website. This question wouldn't be spam if it hadn't had those links, but as it's actually linked, it can't be an honest job offering ad, but direct spam of the linked website. – Meta Bug Wizard Sep 8 '18 at 2:49
  • 3
    @Rob Soliciting to hire someone (i.e. a job offer) on the main Q&A is spam. See: Are job offers spam? – Makyen Sep 8 '18 at 3:52
  • 1
    @Makyen - It could be suggested that they visit StackExchange's Developer Blog for information on hiring and they could be asked to delete their own question. It's a friendlier approach when it seems like they didn't intend to do harm. – Rob Sep 8 '18 at 4:04
2

I think the OP is overthinking here.

From a reviewer's perspective:

  • if you are unsure, skip that review.
  • if the systems detects a wrong action, and you disagree ... the system has its flaws here. sometimes we get called out for making the correct decision. simply remember the corresponding issue, and in case it happens again, and you get banned, then contact a moderator, and explain the problem to him, and the ban might be lifted, or reduced.

From the perspective of the person asking in the wrong language: see my last paragraph.

Of course, user @Rob is correct: in the end, it is the responsibility of the questioner to ask a question within the scope of corresponding community.

For a reviewer, translating questions is a no go. You simply don't know whether the OP will be able to understand the answers given in English for example. Translating the question itself helps nobody, agreed.

If you happen to know the language used in the question, it make sense for you to give a "other" close reason and to provide a helpful comment in both English and the other language used by the OP. And maybe suggest to migrate there, or using a moderator flag.

Otherwise, just close as unclear. When you encounter in a review queue, and you are not sure: simply skip.

Beyond that:

No matter how good that proposed auto translation would work: I have seen enough questions supposedly written in nice English that simply didn't make any sense. Chances are that many automatically translated question end up in ways that still let you baffled.

At the same time, such a mechanism would be pretty expensive to implement. To result in unknown outcome? I think this would be a very unwise investment to make. The underlying problem is way too rare to spend so many resources on it.

Then: if stackexchange spends resources on language issues: why not improve "bad language detection"? Prevent people from using the wrong language, instead of trying to (slightly slightly) improve the job of reviewing. Doing a good review is great, but heck: not needing to do it, because the site prevented a invalid question from being posted tops that easily!

Finally, honestly: the people posting in the wrong language are the worst. They ignored all initial information. They ignored all help center content. They ignore the code of conduct. And, in practical reality: they almost always ignore any of the feedback they receive upon dumping their mess into our communities. There is zero reason to be extra welcoming such people. They messed up. Bigly. The reasonable answer is to politely show them the way out.

They are welcome to come back and do better, but at least for me, on SO: there are hundreds of newbies each day, and most of them show much more effort to ask a good question. I rather prefer that our limited resources get used for the newbies making small mistakes. Instead of worrying how we improve the lives of people ignoring everything they were asked to consider prior posting.

  • It's only for review, to migrate or delete. – Rob Sep 8 '18 at 16:25
  • 2
    That is simply not correct. Given the required privileges, you can close vote with specific migration targets, and you can also vote to delete on the question directly. I do that all the time (except for migration votes, as they are pretty rare). – GhostCat Sep 8 '18 at 16:44
  • @Rob Huh? A) your comment says: "It's only for review, to migrate or delete." is factually incorrect. I am not complaining about your request (question), but that single comment here. B) when answering to other comments using the mobile, I made the experience that sometimes, the other person isn't notified. Now that I have real keyboard, and I know that @ does work, I simply ensured to get to you. C) And honestly: I tried my best to answer your questions, and I don't see why you go snappy here. If you don't like to receive responses/feedback on your input, ... – GhostCat Sep 8 '18 at 18:50
  • 2
    ... sorry, then you are in the wrong place. There are other sites that allow you to block users, or prevent comments on your content, but well: the idea here is that we are all grown ups and able to agree to disagree. – GhostCat Sep 8 '18 at 18:51
0

Since there are only a couple of UpVotes, and a majority of the votes and comments favor deletion without warning or interaction, a solution that reduces our efforts seems better than continuing to suffer with the problem.

On StackOverflow if there is a problem detected with a question or answer it is not possible to post until it is fixed. One example is:

Your post appears to contain code that is not properly formatted as code. Please indent all code by 4 spaces using the code toolbar button or the CTRL+K keyboard shortcut. For more editing help, click the [?] toolbar icon.

The are various Natural Language Processing (NLP) libraries available that can detect foreign languages and the percentage of each within the text. We could use the result of the test to refuse to accept the post. Exceptions could be made on a per site basis (such as Law.SE, where some Latin is permitted).

This avoids us having this problem in the first place and saves the trouble of dealing with it.

A list of language identifier programs is offered on the: "100 Best GitHub: Language Identification" webpage, but through some research I've discovered that the Apache OpenNLP seems to detect the most (103) languages. One example project using that library is Ruthwik's "Language Detection in Java using OpenNLP".

A side benefit of the library is that it can be used as a SPAM detector, as explained on the webpage: "Stanford NLP and Java 9: Creating an Email Spam Filter", and it could probably assist the FireAlarm prioritizing its calls.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .