Tuesday 15 November 2016

Google Translate Enhanced By Neural Machine Translation Technology


Internet has made the world feel smaller, connecting people all over the Earth. The problem? Many of us speak different languages! Of course, some of you speak several languages, but for many of us - including your servant - one is all we know.

Fortunately, translation services like Google Translate, remove barriers, allowing people to communicate more easily, despite differences in language. Today, the research giant announced that its service has greatly improved thanks to the Neural Machine Translation technology.

But wait, what is Neural Machine Translation, anyway? Barak Turovsky, product manager for Google, explains the following.

At a high level, the system translates into complete sentences of neurons at a time, rather than piece by piece. Use this wider context to help you find the most relevant translation, which then rearranges and adjusted to be more like a human conversation with grammar. As it is easier to understand each sentence, paragraph and translated articles are much softer and easier to read. And all this is possible through an end-to-end learning system based on Neural Automatic Translation, which basically means that the system learns over time to create better quality and more natural translations.

Turovsky also said that "today we are launching neuronal translation machines with a total of eight languages ​​and towards English, French, German, Spanish, Portuguese, Chinese, Japanese, Korean and Turkish. These represent the mother tongue of about a third of the world's population, covering more than 35 percent of all Google Translate queries.


While many of the more popular Google Translate langauges are supported by this technology, the search giant still has a long way to go. However, as some people say, "do not leave the ideal in the best way." In other words, Google deserves congratulations for improving the service, although only a fraction of users experience the benefits today.

No comments:

Post a Comment