back to article Google admits 'garbage in, garbage out' translation problem

Google's ever-so-clever Google Translate service may be falling foul of a problem known to grizzled engineers across the globe: garbage in, garbage out. The problem was discussed by Google's director of research, Peter Norvig at the Nasa Innovative Advanced Concepts conference at Stanford, California on Wednesday, in response …

COMMENTS

This topic is closed for new posts.
  1. veti Silver badge

    Data retention, it's all the rage

    How about Google Translate "remembers" every translation it ever makes, and refrains from using those same values for future self-training?

    It's not as if Google has some sort of ethical problem with hanging on to its customer data.

    1. JeffyPoooh Silver badge
      Pint

      Re: Data retention, it's all the rage

      If they run out of disk space (LOL), they could use a hash. Obviously.

      Another option might be to include some unprintable characters as a type of 'Google Translate wuz here' watermark. If people are cutting and pasting, then they might not notice the watermark.

      Meanwhile, how come nobody has seeded the 'net with Monty Python style nipple-fondling mistranslations?

    2. big_D Silver badge

      Re: Data retention, it's all the rage

      Maybe it should remember what it has translated and promise never to use those translations again!

      The Translate functionality is pretty dreadful. I've yet to see a halfway usable translation for most of the texts I've tried.

    3. Tom 13

      Re: Data retention, it's all the rage

      Better yet, since click through agreements are all the rage for software and websites, put one on the Google Translate service. In it you require anyone using the service for a website translate to put in a translate_robots.txt tag to indicate the translation came from Google translate. Or maybe get really fancy and allow it to specify how the translation was created.

    4. JaitcH
      Thumb Down

      Re: Data retention, it's all the rage

      @ veti:

      If you are so wound up about Google Translate and privacy - simply don't use it!

      1. veti Silver badge

        Re: Data retention, it's all the rage

        @JaitcH: Who's wound up? I'm quite sincerely wondering why this is an issue, when the solution looks so simple.

  2. Gray Ham
    Joke

    Ne dérangez pas le chat qui dort, s’il vous plait. J’ai d’autres chats à fouetter.

    (Interestingly, while Bing sort of manages to translate the first part, it totally misses the second. Google is the reverse).

    1. frank ly Silver badge

      re. chat(s)

      Google translates the entire thing as, "Do not disturb the sleeping cat, please. I have other fish to fry."

      If you remove the first sentence, Google then translates the second sentence correctly. I could understand it 'correcting' the word 'cat' to 'fish' if it was using some kind of recognition of common expression algorithm, but not the behaviour as seen.

      Update: If you have anything as a first sentence, it gives 'fish' in the second, instead of 'cats'. Then after more messing around, it always gives 'fish', no matter how you arrange it.

      1. Khaptain Silver badge
        Coat

        Re: re. chat(s)

        They have not yet mastered the use of the Babel Fish......

      2. J.G.Harston Silver badge

        Re: re. chat(s)

        "...I have other fish to fry" is the correct English idiom that matches the French idiom.

      3. M man

        Re: re. chat(s)

        let sleeping dogs lie, I have other fish to fry. > referencial translation

        do not disturb the cat, I am whipping other cats. > literal translation

        1. Phil O'Sophical Silver badge

          Re: re. chat(s)

          And to add to the fun, I think the reference to "having other cats to whip" in the French refers to the "cat o' nine tails", and not a feline.

          1. Irony Deficient

            cat o’ nine tails

            Phil O’Sophical, I think that the proper French for that is martinet (“swift”, of the avian variety).

  3. psychonaut

    www.myhovercraftisfullofeels.com

    Text

    1. Michael H.F. Wilkinson Silver badge

      Re: www.myhovercraftisfullofeels.com

      I will not buy this record; it is scratched!!

  4. xyz

    IIRC

    >This post-modern problem means that Google's machines may be training themselves on data >generated by Google's machines...

    When we tried feeding cows and pigs to cows and pigs we ended up with mad cow disease and foot and mouth, so by about half past eleven tomorrow Google could be exhibiting some very twitchy Skynet type behaviour otherwise known as having gone Ballmer.

  5. JaitcH
    Happy

    WHATEVER Google Translates shortcomings ...

    many people appreciate the fact it is available.

    The Cong An (People's Police) in VietNam have equipped all their sleeping quarters (aka police stations') with computers so the can communicate with Foreigners.

    The same has happened in Cambodia/Kampuchea and Laos.

    I am able to quickly scan e-mail and web sites in languages foreign to me and at least get the gist of what it is about. Damn site cheaper than paying USD$5/A4 sheet of print for a professional translator.

    See, Google does do good!

This topic is closed for new posts.

Biting the hand that feeds IT © 1998–2019