一种多语言机制英语在线翻译辅助系统  

A multilingual English online translation assistant system

作　　者：杨虹 YANG Hong(School of Foreign Languages,Xianyang Normal University,Xianyang 712000,Shaanxi Province,China)

机构地区：[1]咸阳师范学院外国语学院,陕西咸阳712000

出　　处：《信息技术》2022年第2期25-29,35,共6页Information Technology

基　　金：咸阳师范学院科研项目2020年度课题(2020X-SYH168)。

摘　　要：文中通过对以词组为本的统计式机器翻译系统(Statistical machine translation system,SMT)进行词对应方法的优化和筛选来建立多语言机制的英语在线翻译辅助系统。通过M1模型来计算平行语料库中语言字与字之间可能的配对几率,以EM演算法求取未知资料状况下的参数最大值。通过设置条件值参数作为挑选对应句子的依据,采用双语新闻语料库进行实验对比。条件值参数在0.2至0.4之间为文中系统运作于平行语料库的最佳区间。In this paper,a multi-lingual online translation assistance system for English is established by optimizing and selecting word correspondence methods in a phrase based Statistical machine translation system(SMT).The M1 model is used to calculate the probability of possible pairing between words in parallel corpus,and the EM algorithm is used to obtain the maximum value of parameters in the case of unknown data.By setting the parameter of condition value as the basis of selecting corresponding sentences,the bilingual news corpus is used for experimental comparison.The best range for the system to operate in parallel corpus is when the condition value parameter is from 0.2 to 0.4.

关 键 词：统计式机器翻译 平行语料库 多语言机制 EM演算法 系统训练 

分 类 号：TP311.5[自动化与计算机技术—计算机软件与理论]