基于BP神经网络的机器翻译自动化评判模型设计  

Design of automatic evaluation model of machine translation based on BP neural network

作　　者：王蕊[1] WANG Rui(Xi’an Si Yuan University,Xi’an 710038,China)

机构地区：[1]西安思源学院,西安710038

出　　处：《自动化与仪器仪表》2023年第4期15-19,共5页Automation & Instrumentation

基　　金：西安思源学院校级重点课程建设项目《英语听力》;西安思源学院横向课题合作项目《陕西开瑞建设工程项目管理有限公司高层领导队伍跨文化交际与商务英语会话能力提升的策略研究》(20211011)。

摘　　要：机器翻译评价模型内评价体系的不同会导致评价结果出现较大差异,为保证评判的精度,基于BP神经网络设计机器翻译自动化评判模型。计算词句翻译相似度,获取词向量与共现矩阵的关系式,对权重函数分段表达,得到相似度结果。基于BP神经网络构建翻译结果评价体系,通过神经网络模型的输入层、隐含层以及输出层节点,建立神经网络的向量指标,并对连接权数与阈值的输出值进行计算。设计自动化评价算法,判断“输入层-隐含层”、“隐含层-输出层”两个节点是否成功传输,得到机器翻译的自动化评判模型。结合四种不同语料库中的词句进行机器翻译的评价,在不同的评判模型中,BP神经网络模型的评判一致性程度均大于其他三种模型,可见该方法的评判精度最高,评判结果最准确。In order to ensure the accuracy of evaluation,an automatic evaluation model of machine translation is designed based on BP neural network.Calculate the similarity of word and sentence translation,obtain the relationship between word vector and co-occurrence matrix,and segment the weight function to obtain the similarity result.The evaluation system of translation results is constructed based on BP neural network.The vector index of neural network is established through the nodes of input layer,hidden layer and output layer of neural network model,and the output value of connecting weight and threshold is calculated.The automatic evaluation algorithm is designed to judge whether the two nodes of“input layer hidden layer”and“hidden layer output layer”are successfully transmitted,and the automatic evaluation model of machine translation is obtained.The evaluation of machine translation is carried out by combining the words and sentences in four different corpora.In different evaluation models,the evaluation consistency of BP neural network model is greater than that of the other three models.It can be seen that this method has the highest evaluation accuracy and the most accurate evaluation results.

关 键 词：BP神经网络 机器翻译 自动化评判模型 一致性程度 优化算法 翻译评价 

分 类 号：TP202[自动化与计算机技术—检测技术与自动化装置] TN06[自动化与计算机技术—控制科学与工程]