The development of artificial intelligence to change the automatic translation is progressing steadily. Several trends exist, which provide a service that is far from what a human service can achieve. However, automated models explore the slopes, to provide more efficient tools.

Neuronal translation

The Chinese translation services company, YBD Translations has been working for ten years with translation tools. Its operation is based primarily on the translation of complex content, as part of its terminological corpus research.

Automatic translation can be classified into three approaches:

Rule-based Translation

It is a system based on the application of rules at various levels of linguistic analysis (lexical, syntax, and grammar). It integrates the management of a very large number of special cases and exceptions. It produces coherent texts, but ultimately unsuitable for too specialized elements.

Statistical translation

This model is based on a statistical analysis of a large volume of examples already translated. It identifies changes in groups of words from the source language to the target, to reproduce those most likely translations for the new phrases you input. It is finally adapted to specific content, but may lack fluency.

Neuronal translation

System based on neural algorithms. The engine uses artificial intelligence and learning, always from a large volume of examples already translated through a neural network. This is the approach which has the most possibilities from the Chinese translation company YBD Translations’ point of view.

How YBD’s Business Works?

Innovation is part of the genes of our company. The founder, Andy Tsai, created YBD following his thesis on sub-phrasal analysis. This technology was innovative at the time. Next, the statistical translation has grown and today neuronal translation made a huge leap forward. At YBD, we strive to keep this DNA steeped in research applied to industrial problems.

Why are we interested in neuronal translation?

Neural networks have been one of the major trends of research for about thirty years. Long out of fashion, they are now coming back on center stage with the dramatic increase in computing power, and masses of data become accessible under the name Deep Learning (learning by using networks neurons with a large number of elements).

What is the point of this technology for a translation project?

We saw early on the value of this approach. With our complementary technologies, Automatic Translation performance is improving as it is fed with new content and the associated background terminology is expanding. The more translations done, the quality and fluency of the results approach the quality of a human translation.

Neuronal translation, added to our other tools, allows our clients to understand technical texts in a given trade, while remaining within the corporate firewall. These are substantial savings in cost and time: a professional translator can work up to two to three times faster while maintaining high translation quality.

Categories: Translation

0 Comments

Leave a Reply

Your email address will not be published. Required fields are marked *