Translate Google is probably the most widely used dictionaries in the world. There is certainly nothing surprising in order to support one of the largest corporations speaking for itself. This project has been developed for more than 10 years and can translate the text into 103 different languages. What is true of your obligations to fulfill, at least good, but this does not mean that any paste text get the perfect translation in another language. This has remained a long way to go yet, but Google engineers already have an idea of the translator to become more accurate. The new feature is already available, however, it is only one – a better translation will be available for now only nine more popular languages.
From now on the text entered will be considered in its entirety. This means that the translator must translate the whole sentence better, in order to understand its meaning and context.
So far, the program translated texts based on a translation phrasing algorithm, which consisted of capturing individual strings of words statistical methods. Now for the interpreter has been implemented Neural machine Translation technology because the text introduced will be taken into account in its entirety. This means that the translator must translate better to the whole sentence or longer text, in order to understand its meaning and context. But this is not the end – through the use of neural networks and machine learning technology, this system over time will exercise better and better, which – hopefully – will translate into more natural and satisfactory results in the future, Much of the work of professional translators.
This sounds very promising, but at this time Neural Machine Translation technology only works when translating between languages like English, French, German, Spanish, Portuguese, Chinese, Japanese, Korean and Turkish. The reason for their choice in the first place is not surprising – they represent 35% of all translations within the translator, from Google. On the implementation of others (including Polish) we wait a little longer, but the Mountain View giant declares that the new function of time will cover all 103 languages. We are looking forward to it!