摘要
针对传统翻译系统在单语语料库中易出现翻译语料丢失和翻译准确率低的问题,以单语语料库中的自动英诗汉译为研究对象,提出基于无监督学习的神经网络机器翻译方法,该方法将序列到序列模型Seq2Seq和注意力机制Attention相结合,构建Seq2Seq+Attention的单语语言机器翻译模型;在编码器中加入BiLSTM网络,通过回译策略对机器翻译模型进行反向训练和翻译,从而将无监督学习方法转换为有监督学习,以提升最终翻译结果准确率。实验结果表明,在单语语料库中,提出的基于无监督机器翻译方法在不同训练次数下BLEU值最高可达25。且通过人工评分发现,人工评分总分可达17.72分,总体分数较高。由此说明提出的方法可有效避免翻译语料丢失现象,提升翻译准确率。
The traditional translation system is prone to lost translation corpus and low translation accuracy in the mono-lingual corpus,The automatic Chinese translation of English poetry is studied in a monolingual corpus,A neural network ma-chine translation method based on unsupervised learning,This approach combines the sequence-to-sequence model Seq2Seq and the attention mechanism Attention,Build the monolingual language machine translation model of Seq2Seq+Attention;Adding the BiLSTM networks to the encoder,Reverse training and translation of machine translation models through a reback strategy,Thus transforming the unsupervised learning approach to supervised learning,To improve the accuracy of the final translation results.Experimental results show that the proposed unsupervised machine translation-based method achieves maxi-mum BLEU values 25 for different training times in monolingual corpora.Moreover,it was found that the total Score of manual Score could reach 16.68 points,and the overall Score was high.This shows that the proposed method can effectively avoid the loss of translation corpus and improve the translation accuracy.
作者
朱亚辉
ZHU Yahui(School of Foreign Languages,Changsha Normal University,Changsha 410100,China)
出处
《自动化与仪器仪表》
2022年第10期161-165,共5页
Automation & Instrumentation
基金
湖南省教育厅优秀青年项目《接受美学视角下儿童文学文化翻译研究——以曹文轩作品为例》(16B024)
湖南省教改课题《以成果为导向的一流专业翻译课程融合式教学模式构建与实践》(HNJG-2022-0376)。
