基于现代智能识别技术的英语机器翻译模型
2018-08-21张帆
张帆



摘 要: 基于句法分析的英语机器翻译方法无法解决智能识别技术中海量英语语言的部分结构歧义导致机器翻译准确度低的问题。因此,在分析智能机器翻译工具辅助英语翻译应用的基础上,设计与实现基于现代智能识别技术的英语机器翻译模型。依据英文句子产生中文句子和对齐过程,基于英文句子获取中文句子长度、首个中文词串的链接位置,获取总体中文句子以及机器翻译对在句子中的次数。采用基于最大熵的统计机器翻译方法,通过直接最大熵模型训练得到相关参数,获取不同英语语言特征间的最佳组合方式,解决海量英语语言中的部分结构歧义问题,提高英语机器翻译的准确度。实验结果表明,所设计的英语机器翻译模型,具有较高的翻译准确度和稳定性。
关键词: 智能识别技术; 英语翻译; 机器翻译模型; 结构歧义; 最大熵; 翻译准确度
中图分类号: TN915?34; H319.3 文献标识码: A 文章编号: 1004?373X(2018)16?0151?04
Abstract: The English machine translation method based on syntactic analysis cannot resolve the problem existing in intelligent recognition technology for part of structural ambiguity in the massive English language, resulting in low accuracy of machine translation. Therefore, an English machine translation model based on modern intelligent recognition technology is designed and implemented on the basis of application analysis of intelligent machine translation tool assisted English translation. Chinese sentences and alignment process are generated based on English sentences, Chinese sentence lengths and the link position of the first Chinese word string are obtained based on English sentences, and the whole Chinese sentences and the number of machine translated sentences are obtained. The statistical machine translation method based on the maximum entropy is adopted. The best combination mode of different English language features is obtained after obtaining related parameters by means of direct maximum entropy model training, so as to resolve the problem of part of structural ambiguity in the massive English language, and improve the accuracy of English machine translation. The experimental results show that the designed English machine translation model has high translation accuracy and stability.
Keywords: intelligent recognition technology; English translation; machine translation model; structural ambiguity; maximum entropy; translation accuracy
由于全球化的高速發展,不同国家间的信息流动呈现高速性,英语成为当前国际间沟通的主要语言。当前智能识别技术在不同领域中的应用价值不断提升,基于现代智能识别技术的英语机器翻译模型,能够提高英语机器翻译效率和准确度,实现无障碍交流[1?2]。而传统基于句法分析的英语机器翻译方法,无法解决智能识别技术中的海量英语语言中的部分结构歧义问题,存在机器翻译准确度低的问题。因此,本文设计基于现代智能识别技术的英语机器翻译模型,通过直接最大熵模型,获取复杂英语句子中不同特征间的最佳组合方式,消除部分结构歧义,提高英语机器翻译的准确度[3?4]。……
