The history of machine translation has been introduced on the Internet. For example, Asia Pacific Association for Machine Translation (AAMT) introduces the history of machine translation in Japan from the beginning to present. The realization of translation without human intervention has been a dream since the advent of computers. This dream has been handed down to the present without ending.

What is machine translation technology?

There has been continuous progress in the machine translation technology.

Rule Based Machine Translation
(RBMT, or RMT)

The method is to interpret their structure to translate the sentences based on the rules that you have registered. It is said to be an old machine translation method, and it is not used much today. However, even without the corpus (past translation assets), there is a merit that you can translate to some extent or there are no omissions in translations.

Statistical Base Machine Translation
(SBMT, or SMT)

This is a method of registering a large amount of bilingual data (usually referred to as a “corpus”) and generating a translation by a statistical method. It is said that there is no need to develop rules and dictionaries, so it is difficult to be bound by the nature of both the original text and the translated sentences, so it is easy to use multilingual languages.

Neural Machine Translation

In the past, there were two main technologies and a combination of these technologies. Neural Machine Translation NMT)was then created by neural networks and deep learning. In November 2016, Google translation with NMT,which was introduced by Google, gave a tremendous impact to various fields because of the vastly improved quality of machine translation. In NMT, it is said that a translation is output with a structure similar to that in which a human understands the language. Depending on the engine, it is said that the translation of the TOEIC 900 or more can be generated.

Japanese is considered difficult to use in machine translation, and it is not yet perfect at this stage. However, the category “translation” is in the process of being absorbed by some category.