Neural System Combination for Machine Translation

Abstract

Neural machine translation (NMT) becomes a new approach to machine translation and generates much more fluent results compared to statistical machine translation (SMT). However, SMT is usually better than NMT in translation adequacy. It is therefore a promising direction to combine the advantages of both NMT and SMT. In this paper, we propose a neural system combination framework leveraging multi-source NMT, which takes as input the outputs of NMT and SMT systems and produces the final translation. Extensive experiments on the Chinese-to-English translation task show that our model archives significant improvement by 5.3 BLEU points over the best single system output and 3.4 BLEU points over the state-of-the-art traditional system combination methods.

arXiv commentAccepted as a short paper by ACL-2017

References in corpus (4)

• Sequence to Sequence Learning with Neural Networks
• Modeling Coverage for Neural Machine Translation
• Pre-Translation for Neural Machine Translation
• Bridging Neural Machine Translation and Bilingual Dictionaries

Cited by in corpus (9)

• Asynchronous Bidirectional Decoding for Neural Machine Translation
• A Comprehensive Survey of Multilingual Neural Machine Translation
• Translating Phrases in Neural Machine Translation
• Towards Neural Machine Translation with Partially Aligned Corpora
• Learning to Discriminate Noises for Incorporating External Information in Neural Machine Translation
• Neural Machine Translation: A Review and Survey
• Word, Subword or Character? An Empirical Study of Granularity in Chinese-English NMT
• FlexNER: A Flexible LSTM-CNN Stack Framework for Named Entity Recognition
• Inquisitive Question Generation for High Level Text Comprehension