The term “neural” is really valuable in the sphere of artificial intelligence since the companies have formed the request for the further development in machine learning. New technologies are already implemented in some apps, and so Google has decided to use the “Neural Machine Translation” (NMT) in order to increase the level of the translations.
The NMT has helped Google to translate the sentences entirely instead of analyzing one word after another. According to the company’s blog, the system investigates the whole context for the most correct translation, which won’t be different from the phrases we use daily in terms of grammar.
At this moment Google Translate operates in 103 languages, but the “Neural Machine Translation” has been implemented only for eight of them:English, French, German, Spanish, Portuguese, Chinese, Japanese, Korean and Turkish. The company explains that the choice of the languages was due to the prevalence of them in one third of the world. Nevertheless, Google is planning to extend the NMT for all 103 languages.
The representatives of Google comment that after this update Google Translate has changed more than in the last ten years.
In addition to this, the “Neural Machine Translation” has become available for business with a help of the Google Cloud Translation API platform.