A deep source-context feature for lexical selection in statistical machine translation

被引：6

作者：

Gupta, Parth ^{[1
]}

Costa-Jussa, Marta R. ^{[2
,3
]}

Rosso, Paolo ^{[1
]}

Banchs, Rafael E. ^{[4
]}

机构：

[1] Univ Politecn Valencia, PRHLT Res Ctr, Camino Vera S-N, E-46022 Valencia, Spain

[2] Univ Politecn Cataluna, TALP Res Ctr, ES-08034 Barcelona, Spain

[3] Inst Politecn Nacl, Ctr Invest Comp, Av San Juan Dios Batiz, Mexico City 07738, DF, Mexico

[4] Inst Infocomm Res, Human Language Technol, 1 Fusionopolis Way, Singapore 138632, Singapore

来源：

PATTERN RECOGNITION LETTERS | 2016年 / 75卷

关键词：

Natural language processing; Neural nets and related approaches; Semantics;

D O I：

10.1016/j.patrec.2016.02.014

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

This paper presents a methodology to address lexical disambiguation in a standard phrase-based statistical machine translation system. Similarity among source contexts is used to select appropriate translation units. The information is introduced as a novel feature of the phrase-based model and it is used to select the translation units extracted from the training sentence more similar to the sentence to translate. The similarity is computed through a deep autoencoder representation, which allows to obtain effective low-dimensional embedding of data and statistically significant BLEU score improvements on two different tasks (English-to-Spanish and English-to-Hindi). (C) 2016 Elsevier B.V. All rights reserved.

引用

页码：24 / 29

页数：6