Neural Machine Translation for English-Malayalam

Neural Machine Translation systems produce state-of-art translation for high resource languages. It is yet a challenge in low-resource and morphologically rich languages. In this paper, we have discussed the existing techniques in handling the morphologically rich and low-resource languages and presented our experiments on developing English-Malayalam NMT system where we have processed the data using different techniques namely word segmentation using morphological analyser and applying Byte pair Encoding (BPE) technique. The results show a significant improvement by implementing the word segmentation using morphological analyser.