{"id":5165,"date":"2017-03-14T12:42:47","date_gmt":"2017-03-14T19:42:47","guid":{"rendered":"https:\/\/blogs.msdn.microsoft.com\/translation\/?p=5165"},"modified":"2017-03-14T12:42:47","modified_gmt":"2017-03-14T19:42:47","slug":"korean-becomes-microsoft-translators-11th-neural-network-translation-language","status":"publish","type":"post","link":"https://www.microsoft.com\/en-us\/translator/blog\/2017\/03\/14\/korean-becomes-microsoft-translators-11th-neural-network-translation-language\/","title":{"rendered":"Korean Becomes Microsoft Translator\u2019s 11th Neural Network Translation Language"},"content":{"rendered":"

\"seoul_at_night\"<\/a><\/p>\n

 <\/p>\n

Last year Microsoft announced<\/a><\/strong> the release of its Neural Network based translation system<\/a><\/strong> for 10 languages: Arabic, Chinese, English, French, German, Italian, Japanese, Portuguese, Russian, and Spanish. Today, Korean is being added to the list.<\/p>\n

Neural Network translation uses the full context of a sentence to translate words based not only on a few words before and after it, but on the full sentence, generating more fluent and more human sounding translations. This new AI-powered technology delivers the most significant improvement in machine translation quality since statistical machine translation became the industry standard 10 years ago.<\/p>\n

Thanks to these improvements in quality and fluency, translations are the closest they have ever been to human generated ones.<\/p>\n

 <\/p>\n

HOW IT WORKS<\/strong><\/p>\n

\"how-it-works\"<\/a><\/p>\n

 <\/p>\n

At a high level, Neural Network translation works in two stages:<\/p>\n

    \n
  1. The first stage models the word that needs to be translated based on the context of this word (and its possible translations) within the full sentence, whether the sentence is 5 words or 20 words long.<\/li>\n
  2. The second stage then translates this word model (not the word itself but the model the neural network has built), within the context of the sentence, into the other language.<\/li>\n<\/ol>\n

    Neural Network translation uses models of word translations based on what it knows from both languages about a word and the sentence context to find the most appropriate word as well as the most suitable position for this translated word in the sentence.<\/p>\n

    One way to think about neural network-based translation is to think of a fluent English and French speaker that would read the word \u201cdog\u201d in a sentence: \u201cThe dog is happy\u201d. This would create in his or her brain the image of a dog. This image would be associated with \u201cle chien\u201d in French. The Neural Network would intrinsically know that the word \u201cchien\u201d is masculine in French (\u201cle\u201d not \u201cla\u201d). But, if the sentence were to be \u201cthe dog just gave birth to six puppies\u201d, it would picture the same dog with puppies nursing and then automatically use \u201cla chienne\u201d (female form of \u201cle chien\u201d) when translating the sentence.<\/p>\n

     <\/p>\n

    Here’s an example of the benefits of this new technology used in\u00a0the following sentence: (one of the randomly proposed on our try and compare site: http:\/\/translate.ai<\/a><\/strong>)<\/p>\n

    M277dw\uc5d0 \uc885\uc774 \ubb38\uc11c\ub97c \uc62c\ub824\ub193\uace0, \uc2a4\ub9c8\ud2b8\ud3f0\uc73c\ub85c \uc2a4\uce94 \uba85\ub839\uc744 \ub0b4\ub9b0 \ub4a4 \ud574\ub2f9 \ud30c\uc77c\uc744 \uc2a4\ub9c8\ud2b8\ud3f0\uc5d0 \uc989\uc2dc \uc800\uc7a5\ud560 \uc218 \uc788\ub2e4.<\/p>\n

    Traditional Statistical Machine Translation would offer this translation:<\/p>\n

    \u201cM277dw, point to the document, the paper off the file scan command Smartphone smartphones can store immediately.\u201d<\/em><\/p>\n

    Neural Network translation, in comparison, generates this clear and fluent sentence:<\/p>\n

    \u201cYou can place a paper document on M277DW, and then save the file to your smartphone immediately after the scan command.\u201d<\/em><\/p>\n

     <\/p>\n

    The Neural Network translation systems are available for you to use through many entry points:<\/p>\n