摘要
基于模型学习的统计机器翻译存在过于依赖给定的平行训练句子、错误难以预测等不足,重新构建基于核岭回归技术的汉英机器翻译模型,将翻译过程看作源语言字符串到目标语言字符串之间的一个映射,输入和输出字符串都被嵌入到各自的再生核希尔伯特空间,在输入端的特征空间和输出端的特征空间之间使用岭回归的方法得到两者之间高维的映射关系,获取给定并行语料库的特征,并使用这些映射来生成机器翻译输出。经实验验证表明,基于核岭回归技术的汉英机器翻译模型能够形成相对应的准确率更高的翻译内容。
For the present,statistical machine translation based on model learning still has some shortcomings,such as over-reliance on given parallel training sentences and unpredictability of errors.To this end,a Chinese-English machine translation model based on kernel ridge regression is constructed:regarding the translation process as a mapping between source language strings and target language strings,embedding both input and output strings into their respective regenerated kernel Hilbert spaces,using ridge regression to obtain a high-dimensional mapping relationship between the feature space at the input side and the feature space at the output side,obtaining features of a given parallel corpus,and using these mappings to generate machine translation outputs.The experimental results show that the Chinese-English machine translation model based on kernel ridge regression will make for more accurate translation contents.
作者
谢庚全
何俊霖
Xie Gengquan;He Junlin(College of Foreign Languages,Hainan University,Haikou 570228,China;School of Computer Science and Technology,University of Chinese Academy of Sciences,Beijing 100049,China)
出处
《黑河学院学报》
2022年第12期181-183,共3页
Journal of Heihe University
基金
国家社会科学基金项目“人类命运共同体视角下中国与新加坡在‘南海行为准则’案文磋商中的双赢博弈模式研究”(18XGJ018)
海南省基础与应用基础研究计划(自然科学领域)高层次人才项目“海南旅游文本汉英网上平行语料库创建中基于岭回归技术的机器翻译模型研究”(2019RC107)。
关键词
机器学习
统计机器翻译
核方法
岭回归
machine learning
statistical machine translation
kernel methods
ridge regression
