Syntax-Based Context Representation for Statistical Machine Translation

Kehai CHEN; Tiejun ZHAO; Muyun YANG

doi:10.1587/transinf.2018EDP7209

Syntax-Based Context Representation for Statistical Machine Translation

Kehai CHEN, Tiejun ZHAO, Muyun YANG

Full Text Views

0

Cite this

Summary :

Learning semantic representation for translation context is beneficial to statistical machine translation (SMT). Previous efforts have focused on implicitly encoding syntactic and semantic knowledge in translation context by neural networks, which are weak in capturing explicit structural syntax information. In this paper, we propose a new neural network with a tree-based convolutional architecture to explicitly learn structural syntax information in translation context, thus improving translation prediction. Specifically, we first convert parallel sentences with source parse trees into syntax-based linear sequences based on a minimum syntax subtree algorithm, and then define a tree-based convolutional network over the linear sequences to learn syntax-based context representation and translation prediction jointly. To verify the effectiveness, the proposed model is integrated into phrase-based SMT. Experiments on large-scale Chinese-to-English and German-to-English translation tasks show that the proposed approach can achieve a substantial and significant improvement over several baseline systems.

Publication: IEICE TRANSACTIONS on Information Vol.E101-D No.12 pp.3226-3237

Publication Date: 2018/12/01

Publicized: 2018/08/24

Online ISSN: 1745-1361

DOI: 10.1587/transinf.2018EDP7209

Type of Manuscript: PAPER

Category: Natural Language Processing

Authors

Kehai CHEN
  Harbin Institute of Technology
Tiejun ZHAO
  Harbin Institute of Technology
Muyun YANG
  Harbin Institute of Technology

Keyword

syntax context representation, tree-based neural network, translation prediction, statistical machine translation

Cite this

Copy

Kehai CHEN, Tiejun ZHAO, Muyun YANG, "Syntax-Based Context Representation for Statistical Machine Translation" in IEICE TRANSACTIONS on Information, vol. E101-D, no. 12, pp. 3226-3237, December 2018, doi: 10.1587/transinf.2018EDP7209.
Abstract: Learning semantic representation for translation context is beneficial to statistical machine translation (SMT). Previous efforts have focused on implicitly encoding syntactic and semantic knowledge in translation context by neural networks, which are weak in capturing explicit structural syntax information. In this paper, we propose a new neural network with a tree-based convolutional architecture to explicitly learn structural syntax information in translation context, thus improving translation prediction. Specifically, we first convert parallel sentences with source parse trees into syntax-based linear sequences based on a minimum syntax subtree algorithm, and then define a tree-based convolutional network over the linear sequences to learn syntax-based context representation and translation prediction jointly. To verify the effectiveness, the proposed model is integrated into phrase-based SMT. Experiments on large-scale Chinese-to-English and German-to-English translation tasks show that the proposed approach can achieve a substantial and significant improvement over several baseline systems.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2018EDP7209/_p

Copy

@ARTICLE{e101-d_12_3226,
author={Kehai CHEN, Tiejun ZHAO, Muyun YANG, },
journal={IEICE TRANSACTIONS on Information},
title={Syntax-Based Context Representation for Statistical Machine Translation},
year={2018},
volume={E101-D},
number={12},
pages={3226-3237},
abstract={Learning semantic representation for translation context is beneficial to statistical machine translation (SMT). Previous efforts have focused on implicitly encoding syntactic and semantic knowledge in translation context by neural networks, which are weak in capturing explicit structural syntax information. In this paper, we propose a new neural network with a tree-based convolutional architecture to explicitly learn structural syntax information in translation context, thus improving translation prediction. Specifically, we first convert parallel sentences with source parse trees into syntax-based linear sequences based on a minimum syntax subtree algorithm, and then define a tree-based convolutional network over the linear sequences to learn syntax-based context representation and translation prediction jointly. To verify the effectiveness, the proposed model is integrated into phrase-based SMT. Experiments on large-scale Chinese-to-English and German-to-English translation tasks show that the proposed approach can achieve a substantial and significant improvement over several baseline systems.},
keywords={},
doi={10.1587/transinf.2018EDP7209},
ISSN={1745-1361},
month={December},}

Copy

TY - JOUR
TI - Syntax-Based Context Representation for Statistical Machine Translation
T2 - IEICE TRANSACTIONS on Information
SP - 3226
EP - 3237
AU - Kehai CHEN
AU - Tiejun ZHAO
AU - Muyun YANG
PY - 2018
DO - 10.1587/transinf.2018EDP7209
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E101-D
IS - 12
JA - IEICE TRANSACTIONS on Information
Y1 - December 2018
AB - Learning semantic representation for translation context is beneficial to statistical machine translation (SMT). Previous efforts have focused on implicitly encoding syntactic and semantic knowledge in translation context by neural networks, which are weak in capturing explicit structural syntax information. In this paper, we propose a new neural network with a tree-based convolutional architecture to explicitly learn structural syntax information in translation context, thus improving translation prediction. Specifically, we first convert parallel sentences with source parse trees into syntax-based linear sequences based on a minimum syntax subtree algorithm, and then define a tree-based convolutional network over the linear sequences to learn syntax-based context representation and translation prediction jointly. To verify the effectiveness, the proposed model is integrated into phrase-based SMT. Experiments on large-scale Chinese-to-English and German-to-English translation tasks show that the proposed approach can achieve a substantial and significant improvement over several baseline systems.
ER -