When do Contrastive Word Alignments Improve Many-to-many Neural Machine Translation?
Zhuoyuan Mao, Chenhui Chu, Raj Dabre, Haiyue Song, Zhen Wan, Sadao Kurohashi
Abstract
Word alignment has proven to benefit many-to-many neural machine translation (NMT). However, high-quality ground-truth bilingual dictionaries were used for pre-editing in previous methods, which are unavailable for most language pairs. Meanwhile, the contrastive objective can implicitly utilize automatically learned word alignment, which has not been explored in many-to-many NMT. This work proposes a word-level contrastive objective to leverage word alignments for many-to-many NMT. Empirical results show that this leads to 0.8 BLEU gains for several language pairs. Analyses reveal that in many-to-many NMT, the encoder’s sentence retrieval performance highly correlates with the translation quality, which explains when the proposed method impacts translation. This motivates future exploration for many-to-many NMT to improve the encoder’s sentence retrieval performance.
BibTeX
@inproceedings{mao-etal-2022-contrastive,
title = "When do Contrastive Word Alignments Improve Many-to-many Neural Machine Translation?",
author = "Mao, Zhuoyuan and
Chu, Chenhui and
Dabre, Raj and
Song, Haiyue and
Wan, Zhen and
Kurohashi, Sadao",
editor = "Carpuat, Marine and
de Marneffe, Marie-Catherine and
Meza Ruiz, Ivan Vladimir",
booktitle = "Findings of the Association for Computational Linguistics: NAACL 2022",
month = jul,
year = "2022",
address = "Seattle, United States",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2022.findings-naacl.134/",
doi = "10.18653/v1/2022.findings-naacl.134",
pages = "1766--1775"
}