English

Linguistic Input Features Improve Neural Machine Translation

Computation and Language 2016-06-29 v2

Abstract

Neural machine translation has recently achieved impressive results, while using little in the way of external linguistic information. In this paper we show that the strong learning capability of neural MT models does not make linguistic features redundant; they can be easily incorporated to provide further improvements in performance. We generalize the embedding layer of the encoder in the attentional encoder--decoder architecture to support the inclusion of arbitrary features, in addition to the baseline word feature. We add morphological features, part-of-speech tags, and syntactic dependency labels as input features to English<->German, and English->Romanian neural machine translation systems. In experiments on WMT16 training and test sets, we find that linguistic input features improve model quality according to three metrics: perplexity, BLEU and CHRF3. An open-source implementation of our neural MT system is available, as are sample files and configurations.

Keywords

Cite

@article{arxiv.1606.02892,
  title  = {Linguistic Input Features Improve Neural Machine Translation},
  author = {Rico Sennrich and Barry Haddow},
  journal= {arXiv preprint arXiv:1606.02892},
  year   = {2016}
}

Comments

WMT16 final version; new EN-RO results

R2 v1 2026-06-22T14:21:33.837Z