• vi-VNen-GB

Automatic translation machine

The idea of machine translation started from 1629. However, the machine translation has really developed since 1954. By the end of 1980s, there were more new methods in machine translation. Especially the methods of Statistical Machine Translation (SMT) and Example-based Machine Translation (EBMT). These two methods have begun to replace the rules of syntax and semantics by the procession on large text data blocks which is allowed by computing capacity.

Statistical Machine Translation has proven to be an effective method to develop machine translation systems of high quality when SMT machines have continued to win in machine translation contests annually. The methods to improve the quality of SMT are still being concerned by many researchers. The objective is to have processions appropriate to the characteristics of specific language pairs such as the interrelation of word order, word selection according to the context, and the combination with syntax processing methods before and after the procession.

Realizing many advantages, we have decided to build a translaton machine translation system on the foundation of SMT method and combined SMT, EBMT, and RBMT together with the hope to creat the difference in quality of machine translation.

Machine translation system is a disperal system, the machine translation is scattered in the network environment (LAN / Internet), are connected together by a distributed load (load balancer) and buffers for the purpose of increasing serving multiple users simultaneously while ensuring the quality of automatic translation.

Currently we have built translation machine in term of Law and Medicine, in which the quality of translation machine in term of law subject has proved its effectiveness. Translation machines in term of other subjects are being under construction; We will publish these translation machines according to subjects whenever their quality meet the standards set by VIEGRID.