Word-based Domain Adaptation for Neural Machine Translation

Shen Yan, Leonard Dahlmann, Pavel Petrushkov, Sanjika Hewavitharana, Shahram Khadivi


Abstract
In this paper, we empirically investigate applying word-level weights to adapt neural machine translation to e-commerce domains, where small e-commerce datasets and large out-of-domain datasets are available. In order to mine in-domain like words in the out-of-domain datasets, we compute word weights by using a domain-specific and a non-domain-specific language model followed by smoothing and binary quantization. The baseline model is trained on mixed in-domain and out-of-domain datasets. Experimental results on En → Zh e-commerce domain translation show that compared to continuing training without word weights, it improves MT quality by up to 3.11% BLEU absolute and 1.59% TER. We have also trained models using fine-tuning on the in-domain data. Pre-training a model with word weights improves fine-tuning up to 1.24% BLEU absolute and 1.64% TER, respectively.
Anthology ID:
2018.iwslt-1.5
Volume:
Proceedings of the 15th International Conference on Spoken Language Translation
Month:
October 29-30
Year:
2018
Address:
Brussels
Venues:
EMNLP | IWSLT
SIG:
SIGSLT
Publisher:
International Conference on Spoken Language Translation
Note:
Pages:
31–38
Language:
URL:
https://aclanthology.org/2018.iwslt-1.5
DOI:
Bibkey:
Cite (ACL):
Shen Yan, Leonard Dahlmann, Pavel Petrushkov, Sanjika Hewavitharana, and Shahram Khadivi. 2018. Word-based Domain Adaptation for Neural Machine Translation. In Proceedings of the 15th International Conference on Spoken Language Translation, pages 31–38, Brussels. International Conference on Spoken Language Translation.
Cite (Informal):
Word-based Domain Adaptation for Neural Machine Translation (Yan et al., IWSLT 2018)
Copy Citation:
PDF:
https://aclanthology.org/2018.iwslt-1.5.pdf