CASIA OpenIR  > 模式识别国家重点实验室  > 自然语言处理
神经网络语言模型在统计机器翻译中的应用
张家俊; 宗成庆
Source Publication情报工程
2017-03
Issue3Pages:21-28
Abstract近两年来,神经机器翻译(Neural Machine Translation, NMT)模型主导了机器翻译的研究,但是统计机器翻译(Statistical Machine Translation, SMT)在很多应用场合(尤其是专业领域)仍有较强的竞争力。如何利用深度学习技术提升现有统计机器翻译的水平成为研究者们关注的主要问题。由于语言模型是统计机器翻译中最核心的模块之一,本文主要从语言模型的角度入手,探索神经网络语言模型在统计机器翻译中的应用。本文分别探讨了基于词和基于短语的神经网络语言模型,在汉语到英语和汉语到日语的翻译实验表明神经网络语言模型能够显著改善统计机器翻译的译文质量。
Other AbstractNeural Machine Translation (NMT) dominates the research of machine translation in recent two years. However, Statistical Machine Translation (SMT) is very competitive in many scenarios such as some specific domains. It becomes a key issue how to apply deep learning technology to improve SMT performance. As language model is one of the most crucial modules in SMT, this paper investigates the usage of neural language model in statistical machine translation. We explore respectively the word-based and phrase-based neural language model, and evaluate the models on both Chinese-to-English and Chinese-to-Japanese translation tasks. The extensive experiments demonstrate that the neural language models can significantly improve the translation performance of statistical machine translation.
Keyword统计机器翻译 神经网络语言模型
Document Type期刊论文
Identifierhttp://ir.ia.ac.cn/handle/173211/20675
Collection模式识别国家重点实验室_自然语言处理
Corresponding Author张家俊
Recommended Citation
GB/T 7714
张家俊,宗成庆. 神经网络语言模型在统计机器翻译中的应用[J]. 情报工程,2017(3):21-28.
APA 张家俊,&宗成庆.(2017).神经网络语言模型在统计机器翻译中的应用.情报工程(3),21-28.
MLA 张家俊,et al."神经网络语言模型在统计机器翻译中的应用".情报工程 .3(2017):21-28.
Files in This Item:
There are no files associated with this item.
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[张家俊]'s Articles
[宗成庆]'s Articles
Baidu academic
Similar articles in Baidu academic
[张家俊]'s Articles
[宗成庆]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[张家俊]'s Articles
[宗成庆]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.