Deep Neural Networks for Multilingual Machine Translation

Abstract

With the emergence of information and communication technologies and the democratization of the Internet content production, we are facing an informational tsunami of different languages. Newspapers and news websites contribute fully to this content, making the amount of daily circulating data on the web or the printed one hard to grasp. Admittedly, it is very difficult to manage and translate manually and even conventional IT tools struggle to deliver results in all languages. To overcome this challenge, we propose to explore deep learning for multilingual machine translation.

This paper presents our newly developed neural text?to?text translation system and its evaluation. The realization of this system has consisted on the corpora elaboration and two deep neural processing modules for machine translation. The system allows users to get the corresponding translated sentences to the sentences that they fed to it, displayed on an ergonomic interface.