1. Andrjus Frantskjavitsius Avatar Andrjus Frantskjavitsius
  2. Untitled project
  3. minoritytranslate
  4. Issues

Mark up all published translations as MT mediated?

Issue #77 resolved
Peeter Tinits created an issue

Form should be: "Edited with MinorityTranslate using articles ["link to version used" "en:Ocean acidification"]

Comments (41)

  6. Peeter Tinits reporter

    Ei tea, kas saaks või oleks mõtet kohe "märgisena" seda märkida? Seda teab aga ainult @Kruusamagi kommenteerida.

  7. Peeter Tinits reporter

    There's a section on How to translate here: https://en.wikipedia.org/wiki/Wikipedia:Translation.

    It says

    The new, translated article must credit the source article. The recommended way to do this is to do both of the following: (a) Have the first edit summary of the target article include an interlanguage link to the source article. Example: Content in this edit is translated from the existing French Wikipedia article at [[:fr:Exact name of French article]]; see its history for attribution. (b) Place the template {{Translated page}} on the target article's talk page.

    @Kruusamagi knows best how much this is used, or whether we should use it.

    Including the from languages in the edit summary sounds like a good idea though.

    So, a proposed format for the edit summary would be: This edit is composed with Minority Translate from [[:fr:Exact name of French article|French]], [[:en:Exact name of English article|English]], [[:et:Exact name of Estonian article|Estonian]]; see source article history for attribution.

    Simpler could be: This edit is composed with Minority Translate based on [[:fr:Exact name of French article|French]], [[:en:Exact name of English article|English]], [[:et:Exact name of Estonian article|Estonian]] articles.

    Including the info on source articles seems generally a good idea in keeping in mind the recommended practice of Wikipedia translations. If it is technically easy.

  8. Andrjus Frantskjavitsius repo owner

    This seems like wiki rules that we can't enforce.

    @peeter_t2 , for some reason @Kruusamagi didnt work for you. Im doing it for you.

  9. Peeter Tinits reporter

    Yea, I noticed the @Kruusamägi @kruusamagi issue as well. Thanks! It didn't seem to work for you either actually.

    I'm not sure on the position of enforcement. I think most translators would consider it a useful help (as they might be too lazy to be correct as well). This is just one way to mark edits, enforcing an MT marking seems smart for us. This could be optional too. Having an MT marking means that we could mine usage statistics even if people don't want this in the corpus.

    Maybe @keeleleek has some opinions on this? Content translation used basically the same structure: "(Loodud lehekülje "Ocean acidification" tõlkimisel) (Märgis: Sisutõlge)" Both article name & sisutõlge link to their respective destinations. In principle, we could use exactly the same structure, maybe even ask for their translations or find out where they got them, and use a tag for Minority Translate.

    Edit: As per wikipedia recommendations Content translation even links to the used article version not just the article name. This would be the most desirable option.

  10. Peeter Tinits reporter

    Märgis = https://en.wikipedia.org/wiki/Wikipedia:Tags ; https://en.wikipedia.org/wiki/Special:Tags . I'm not sure how easy adding new ones is. Ivo should know this for sure. Having a tag would mean that you can see things like this automatically generated: https://en.wikipedia.org/w/index.php?title=Special:RecentChanges&tagfilter=contenttranslation

    Edit: I think anyone with administrator privileges in a Wikipedia can add them. Ivo has such privileges. Not sure how this would work across languages though...

  11. Ivo Kruusamägi

    I don't think there is need for adding anything to the talk page. Showing MT and the source should be more than enough.

    Special:Tags system isn't known to me. I don't think I have any right to edit something in regard of that.

  13. Andrjus Frantskjavitsius repo owner

    Finding the source (language) is a problem. We can't guarantee that the user has the correct language selected when he marks the article for uplaoad.

  14. Peeter Tinits reporter

    The same problem also exists for the source language data in the "corpus". Can you think of any solution?

    The probability of the correct marking seems very high though, if this is done while the article is marked for upload (and not on the upload moment itself).

  15. Peeter Tinits reporter

    On "tagging": I found this on the talk page https://en.wikipedia.org/wiki/Wikipedia_talk:Tags

    "At the moment, the only way to add a tag from your tool would be for the tool to submit the edit using OAuth, which will automatically tag every edit. Ability to specify tags for edits when submitting the edit via the API is an interesting idea, but it would probably need to be limited in some manner to avoid spamming via tags and would probably need to come with a method for deleting inappropriate tags from edits. Anomie⚔ 13:36, 14 August 2014 (UTC)"

    And the manual for tags is available here: https://www.mediawiki.org/wiki/Manual:Tags These are mainly meant for extensions. I'm not sure how difficult it is for us, but it would be a great feature.

    Other relevant links for this are: https://en.wikipedia.org/wiki/Help:Edit_summary https://en.wikipedia.org/wiki/Wikipedia:Edit_filter

  16. Kristian K

    I think it is possible to see MT as an extension and thus use tags system, but this is something for the future.

    I think for now the edit summary could be simply Edited with MinorityTranslate using [language list] Wikipedia articles. where [language list] is the expanded language names, e.g "english, french and estonian".

  19. Peeter Tinits reporter

    it is; see for example https://et.wikipedia.org/w/index.php?title=Ookeanide_hapestumine&action=history

    the instructions how are here: https://en.wikipedia.org/wiki/Help:Edit_summary

    Wikilinks always rendered; other wikitext coding ignored. Edit summaries render internal links, including piped links, and interwiki links, even when enclosed within <nowiki> and </nowiki>. Therefore, copying wikitext in the edit summary box may be preferable to copying text from the preview, except when one wants to save space.

    also see: https://en.wikipedia.org/wiki/Wikipedia:Translation. as linked above

  22. Peeter Tinits reporter

    Great! I tested it now, nice to have it, but not yet perfectly functional.

    1) double brackets are required for links [[...]] ( more important). My mistake in the last recommendation.

    2) is it possible to link to revision instead of article name?

    Thanks!

  23. Peeter Tinits reporter

    For revision this should work (I tested it too), en naturally varies with language:

    [[en:Special:Redirect/revision/684778272 | en:Name_of_page]]

    The link should have "en:Name_of_page" (without quotes) as visible, and link to the specific revision used. Thanks!

  25. Andrjus Frantskjavitsius repo owner

    I misunderstood the syntax:

    ["link to version used" "en:Ocean acidification"]

  26. Kristian K

    Since the names of the articles can take quite much room, and we're short of room, I suggest not to use names. Otherwise some fallbacks must be devised when the message is too long.

    All is doable, if message_not_too_long then post_with_names else post_message_without_names

    but probably a third state should also be captured -- when also the message without names is to be truncated....

  27. Peeter Tinits reporter

    Ok, the edit summary box can hold one line of 250 characters. The way I wrote it, is wikipedia recommended policy (and incidentally also how content translate does it.

    So, add a check for 250 characters and cut it off? - I think it will be quite rare to get to this. I would need about 10 languages? I'm not really doing math properly though.

    Let's analyze first how often 250 would happen before we cut the whole thing. I think not often.

  28. Kristian K

    The following string is 57 characters: [en:Special:Redirect/revision/684778272 en:Name_of_page]

    250 divided by 57 and floored is the number 4. Thus four links.

    05.02.2016 11:47 peeter_t2 kirjutas:

  29. Peeter Tinits reporter

    hm, ok, i don't know how to solve this. Another way to think about this, is that using the mapping to revisions, adds 38 extra characters per language.

    Base the strategy on the number of languages?

    Hard cap at 3, "and others" ?

    Count the number of characters based on designed post, and optimize for the character count?

    No idea actually.

  31. Peeter Tinits reporter

    Or lets use three, and just cut it off if there are more (probably rare). Being able to consult multiple languages seems to be a basic feature here.

  32. Peeter Tinits reporter

    this is still unfixed. double brackets needed [[ , how about we limit this to three articles max then, and add "and other languages" if more.

    Edited with MinorityTranslate using the article [[en:Special:Redirect/revision/684778272|en:Name_of_page]].

    Edited with MinorityTranslate using articles [[en:Special:Redirect/revision/684778272|en:Name_of_page]] and [[en:Special:Redirect/revision/684778272|en:Name_of_page]].

    Edited with MinorityTranslate using articles [[en:Name_of_page]], [[en:Name_of_page]], and [[en:Name_of_page]].

    Edited with MinorityTranslate using articles [[en:Name_of_page]], [[en:Name_of_page]], [[en:Name_of_page]], and [[en:Name_of_page]].

    Edited with MinorityTranslate using articles [[en:Name_of_page]], [[en:Name_of_page]], [[en:Name_of_page]], [[en:Name_of_page]], and same articles in other languages..

    This should never fail. Possible to just catch it and reduce by one article name?

    First thing though: add double-brackets!

  33. Andrjus Frantskjavitsius repo owner

    Will have a look.

    I think I will go with "Edited with MinorityTranslate using [[en:Special:Redirect/revision/684778272|en:Name_of_page]]." .

    Remember that it is alway possible to look up the source articles from the statistics page.

  34. Peeter Tinits reporter

    Yes, it will be a solution. The reason for adding info on multiple articles is just Wikipedia policy, not anything to do with corpus, where the links are there to attribute the sources. However this policy is not totally clear, so just one link also fills the minimal requirements. (Though one of the features of the software is that we can use multiple source articles and combine them, so why not capitalize on that with up to three links.)

    A simple if check for number of articles could be a very easy solution. But anyway choice is up to you.

  38. Andrjus Frantskjavitsius repo owner

    Will use the one language code version right now. This is because I don't have time to mess with the application right now.

    I don't think that just dumping all languages in the summary is a good idea. Later I can add something that measures the amount of text changed.

  39. Andrjus Frantskjavitsius repo owner

    Additional languages will be a seperate issue. A smarter war of adding languages is needed.

  40. Peeter Tinits reporter

    Did you fix the brackets? Need double brackets instead of single ones for links to work.

  41. Peeter Tinits reporter

    Also, I thought I gave multiple solutions. A no-brainer is also to add a link to revision, not article (see above). Of course this depends on the availability of time. Best of luck with whatever you're working on! ;)

  42. Log in to comment
Assignee
–
Type
enhancement
Priority
major
Status
resolved
Votes
0
Watchers
1
Jira: the preferred issue tracker for Bitbucket. Join the team!