Neural Machine Translation System of Indic Languages -- An Attention based Approach

  • 2020-02-02 07:15:18
  • Parth Shah, Vishvajit Bakrola
  • 0

Abstract

Neural machine translation (NMT) is a recent and effective technique whichled to remarkable improvements in comparison of conventional machinetranslation techniques. Proposed neural machine translation model developed forthe Gujarati language contains encoder-decoder with attention mechanism. InIndia, almost all the languages are originated from their ancestral language -Sanskrit. They are having inevitable similarities including lexical and namedentity similarity. Translating into Indic languages is always be a challengingtask. In this paper, we have presented the neural machine translation system(NMT) that can efficiently translate Indic languages like Hindi and Gujaratithat together covers more than 58.49 percentage of total speakers in thecountry. We have compared the performance of our NMT model with automaticevaluation matrices such as BLEU, perplexity and TER matrix. The comparison ofour network with Google translate is also presented where it outperformed witha margin of 6 BLEU score on English-Gujarati translation.

 

Quick Read (beta)

loading the full paper ...