Standard neural machine translation (NMT) is on the assumption that the document-level context is independent. Most existing document-level NMT approaches are satisfied with a smattering sense of global document-level information, while this work focuses on exploiting detailed document-level context in terms of a memory network. The capacity of the memory network that detecting the most relevant part of the current sentence from memory renders a natural solution to model the rich document-level context. In this work, the proposed document-aware memory network is implemented to enhance the Transformer NMT baseline. Experiments on several tasks show that the proposed method significantly improves the NMT performance over strong Transformer baselines and other related studies.
Shu JIANG
Shanghai Jiao Tong University
Rui WANG
Shanghai Jiao Tong University
Zuchao LI
Shanghai Jiao Tong University
Masao UTIYAMA
National Institute of Information and Communications Technology
Kehai CHEN
National Institute of Information and Communications Technology
Eiichiro SUMITA
National Institute of Information and Communications Technology
Hai ZHAO
Shanghai Jiao Tong University
Bao-liang LU
Shanghai Jiao Tong University
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Shu JIANG, Rui WANG, Zuchao LI, Masao UTIYAMA, Kehai CHEN, Eiichiro SUMITA, Hai ZHAO, Bao-liang LU, "Document-Level Neural Machine Translation with Associated Memory Network" in IEICE TRANSACTIONS on Information,
vol. E104-D, no. 10, pp. 1712-1723, October 2021, doi: 10.1587/transinf.2020EDP7244.
Abstract: Standard neural machine translation (NMT) is on the assumption that the document-level context is independent. Most existing document-level NMT approaches are satisfied with a smattering sense of global document-level information, while this work focuses on exploiting detailed document-level context in terms of a memory network. The capacity of the memory network that detecting the most relevant part of the current sentence from memory renders a natural solution to model the rich document-level context. In this work, the proposed document-aware memory network is implemented to enhance the Transformer NMT baseline. Experiments on several tasks show that the proposed method significantly improves the NMT performance over strong Transformer baselines and other related studies.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2020EDP7244/_p
Copy
@ARTICLE{e104-d_10_1712,
author={Shu JIANG, Rui WANG, Zuchao LI, Masao UTIYAMA, Kehai CHEN, Eiichiro SUMITA, Hai ZHAO, Bao-liang LU, },
journal={IEICE TRANSACTIONS on Information},
title={Document-Level Neural Machine Translation with Associated Memory Network},
year={2021},
volume={E104-D},
number={10},
pages={1712-1723},
abstract={Standard neural machine translation (NMT) is on the assumption that the document-level context is independent. Most existing document-level NMT approaches are satisfied with a smattering sense of global document-level information, while this work focuses on exploiting detailed document-level context in terms of a memory network. The capacity of the memory network that detecting the most relevant part of the current sentence from memory renders a natural solution to model the rich document-level context. In this work, the proposed document-aware memory network is implemented to enhance the Transformer NMT baseline. Experiments on several tasks show that the proposed method significantly improves the NMT performance over strong Transformer baselines and other related studies.},
keywords={},
doi={10.1587/transinf.2020EDP7244},
ISSN={1745-1361},
month={October},}
Copy
TY - JOUR
TI - Document-Level Neural Machine Translation with Associated Memory Network
T2 - IEICE TRANSACTIONS on Information
SP - 1712
EP - 1723
AU - Shu JIANG
AU - Rui WANG
AU - Zuchao LI
AU - Masao UTIYAMA
AU - Kehai CHEN
AU - Eiichiro SUMITA
AU - Hai ZHAO
AU - Bao-liang LU
PY - 2021
DO - 10.1587/transinf.2020EDP7244
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E104-D
IS - 10
JA - IEICE TRANSACTIONS on Information
Y1 - October 2021
AB - Standard neural machine translation (NMT) is on the assumption that the document-level context is independent. Most existing document-level NMT approaches are satisfied with a smattering sense of global document-level information, while this work focuses on exploiting detailed document-level context in terms of a memory network. The capacity of the memory network that detecting the most relevant part of the current sentence from memory renders a natural solution to model the rich document-level context. In this work, the proposed document-aware memory network is implemented to enhance the Transformer NMT baseline. Experiments on several tasks show that the proposed method significantly improves the NMT performance over strong Transformer baselines and other related studies.
ER -