Machine Translation Using Deep Learning: A Comparison

Abstract

In recent days, machine translation is rapidly evolving. Today one can find several machine translation systems that provide reasonable translations, although they are not perfect. The main objective of machine translation is to provide interaction among the people speaking two different languages. Machine translation, being an important task of natural language processing, leads to the development of different approaches, namely, rule-based machine translation, statistical machine translation, and neural machine translation for the translation process. The recently proposed method is the neural machine translation which improves the quality of translation between natural languages through neural networks. Neural machine translation led to remarkable improvements in the translation process by retaining the contextual information. End-to-end neural machine translation uses RNN Encoder-Decoder mechanism to train the neural translation model with bilingual corpora which is bilingual parallel sentence pairs, an important resource of machine translation. NMT has a reasonable BLEU score which is the evaluation metrics for machine translation. In this paper, we present a survey on the different kinds of machine translation approaches with their strengths and limitations and the various evaluation metrics to measure the accuracy of the translation.