Actualités, CAT, Langues, Traductions

What is Neural Machine Translation (NMT)

Last year professionals had talked much about NMT. What is NMT?

Neural machine translation (NMT) is a machine translation approach that uses a large artificial neural network to predict the likelihood of a sequence of words, typically modeling entire sentences in a single integrated model.

All of the machine translation products (websites or apps) were based on algorithms using statistical methods to try to guess the best possible translation for a given word. This technology is called statistical machine translation.

However, one of the limitations of statistical machine translation is that it only translates words within the context of a few words before and after the translated word. For small sentences, it works pretty well. For longer ones, the translation quality could vary.

Now we have a new machine learning technology called deep learning or deep neural networks, one that tries to mimic how the human brain works (at least partially).

At a high-level, neural network translation works with in two stages:

— A 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.

— A second stage then translates this word model (not the word itself but the model the neural network has built of it), within the context of the sentence, into the other language.

One way to think about neural network-based translation could be to think of a fluent speaker in another language that would see a word, say “dog”. This would create the image of a dog in his or her brain, then this image would be associated to, for instance “le chien” in French. The neural network would intrinsically know that the word “chien” is masculine in French (“le” not “la”). But, if the sentence were to be “the dog just gave birth to six puppies” , it would picture the same dog with puppies nursing and would then automatically use “la chienne” (female form of “le chien”) when translating the sentence.

Because of this approach, sentences that are generated from a neural network based machine translation are usually better than statistical machine ones but also sound more fluent and natural, as if a human had translated them and not a machine.

Source: Microsoft

Publicités
That's interesting !, Traductions

Some interesting facts about translation history – 2

The word “translation”comes from a Latin term meaning « to bring or carry across ». The Ancient Greek term is ‘metaphrasis’ (« to speak across ») and this gives us the term ‘metaphrase’ (a « literal or word-for-word translation ») – as contrasted with ‘paraphrase’ (« a saying in other words »).  The first known translations are those of the Sumerian epic Gilgamesh into Asian languages from the second millennium BC. Later Buddhist monks translated Indian sutras into Chinese and Roman poets adapted Greek texts.

Translation undertaken by Arabs could be said to have kept Greek wisdom and learning alive. Having conquered the Greek world, they made Arabic versions of its philosophical and scientific works. During the Middle Ages, translations of these Arabic versions were made into Latin – mainly at the school in Spain. These Latin translations of Greek and original Arab works of learning helped underpin Renaissance scholarship. Religious texts have played a great role in the history of translation. One of the first recorded instances of translation in the West was the rendering of the Old Testament into Greek in the 3rd century BC.  Saint Jerome, the patron saint of translation, produced a Latin Bible in the 4th century AD that was the preferred text for the Roman Catholic Church for many years to come. Martin Luther himself is credited with being the first European to propose that one translates satisfactorily only toward his own language: a statement that is just as true in modern translation theory.

Translation today

In the modern world, translation is as important – if not more so – as it was several millennia ago. Officially, there are about 6,800 languages spoken around the world, of which a significant portion have unique scripts and many have shared scripts based on the origins of the language in question. These challenges are compounded by the fact that nearly every culture in the world has interactions with every other culture. This means that there are an incalculable number of translation requirements every second of every minute of every day around the world. It’s no wonder, then, that translation is a dominant part of intercultural interaction.

The slow speed of manual translation has led to technology stepping in. Thus, machine translation (MT) was born. With the dawn of the technological age, the application of software to the field of translation became an interesting subject that was, and continues to be. Although more fallible than purely human translation, machine translation is a useful tool that has found several applications. For example, MT is regularly used for weather reports and other speciality areas where linguistic variables are limited. It is sometimes used for written government or legal communication, too, albeit with a modicum of human intervention. Though currently limited in application, it is a useful tool in the repertoire of any professional translator – if only to make the job a little bit easier or quicker. In its most advanced form, MT may give satisfactory output for unrestricted texts, but it is still best used when domains and variables (such as disambiguation or named entities) are controlled in some way. There is no doubt that the need for human translators will remain, and that even the best MT software can only go so far where sensitive or specialised translation is required. For results of the highest quality and integrity with respect to the source and target material, there is still no adequate substitute for human translator.