LSTM Neural Reordering Feature for Statistical Machine Translation

TL;DR

This paper introduces an LSTM-based neural reordering feature for statistical machine translation, leveraging long-context modeling to improve reordering accuracy and achieve significant performance gains.

Contribution

It presents a novel neural reordering model using LSTM networks that directly models word pairs and alignments for better reordering predictions.

Findings

Significant improvements over baseline systems in Arabic-English and Chinese-English translation tasks.

LSTM neural reordering feature is robust across different language pairs.

Longer context modeling enhances reordering accuracy.

Abstract

Artificial neural networks are powerful models, which have been widely applied into many aspects of machine translation, such as language modeling and translation modeling. Though notable improvements have been made in these areas, the reordering problem still remains a challenge in statistical machine translations. In this paper, we present a novel neural reordering model that directly models word pairs and alignment. By utilizing LSTM recurrent neural networks, much longer context could be learned for reordering prediction. Experimental results on NIST OpenMT12 Arabic-English and Chinese-English 1000-best rescoring task show that our LSTM neural reordering feature is robust and achieves significant improvements over various baseline systems.