摘要
简要介绍了统计机器翻译的国内外现状,对当前统计机器翻译的主流方法进行了理论上的探讨。首先讨论了对数线性模型,然后进一步描述了对数线性模型的特征选择、参数训练和搜索算法,最后在此基础上进行了初步的实验。实验结果表明:基于短语的统计机器翻译方法能够取得较好的翻译效果,值得做更深一步的研究。
This paper briefly introduces the research status of statistical machine translation, and mainly focused on the log - linear model at the basis of the current dominating research methods. We describes the feature selection, parameter training, decoding algorithm , and makes some primary experiments on the log -linear model , the results show that this kind of model can achieve good performance in translation task.
出处
《沈阳航空工业学院学报》
2007年第2期32-34,共3页
Journal of Shenyang Institute of Aeronautical Engineering
关键词
统计机器翻译
短语
语言模型
翻译模型
statistical machine translation
phrase
language model
translation model
