When it comes to translating sentences, Google Translate’s accuracy varies from sorta correct to wildly off the mark. Google’s first attempt at rectifying this is to roll out its Neural Machine Translation (NMT) system, a system based on machine learning that the company claims would result in more accurate translations. First tested out on Mandarin Chinese, Google has announced that the system will be rolled out to additional languages as well.
According to Google’s Keyword blog, NMT will be expanding to English, French, German, Korean, Japanese, Portuguese and Turkish. Google mentions that the languages listed above covers one-third of the world’s population and more than 35% of Google Translation’s queries. Because of the way NMT works, Google believes that by applying to system to the languages listed above, Google Translate will be able to show results that are far more accurate, particularly over time.
Expanding NMT to these seven languages is just the next step in Google’s plans for Translate. The company is looking to rollout NMT to cover all 103 active languages that are in use today. In the meantime, Google is looking to open up its NMT tech to other businesses so they can use it as they see fit.
Source: The Keyword