基于改进神经网络的韩语机器翻译系统设计

Design of Korean Machine Translation System Based on Improved Neural Network

韩语机器翻译受到韵律格式影响,导致翻译的可靠性不好,提出基于改进神经网络的韩语机器翻译系统设计方法。采用语料库模型建立韩语机器翻译系统的双语平行语料库,建立以深度学习为基础的神经网络翻译语义控制模型,采用规则和模板的匹配方法,采用统计机器翻译的方法,构建以语义综合评估为翻译结构参数模型的韩语机器翻译自然语言处理模型,采用无监督的神经网络学习方法,建立韩语机器翻译的语义对照模型,实现迭代反向翻译和回译。系统构建中,设计了数据库模块、语义推荐模块、数据访问层模块和翻译生成模块,在改进神经网络模型下实现机器翻译系统设计。测试结果表明,该方法进行韩语机器翻译的可靠性较好,具有很好的语义连贯性、翻译一致性和翻译流利性,翻译输出的准确度较高。

Korean machine translation is influenced by prosodic format,which leads to poor reliability of translation.This paper puts forward a design method of Korean machine translation system based on improved neural network.The bilingual parallel corpus of Korean machine translation system is established by corpus model,and the semantic control model of neural network translation based on deep learning is established.The natural language processing model of Korean machine translation with semantic comprehensive evaluation as the translation structure parameter model is constructed by the matching method of rules and templates and statistical machine translation method.The semantic comparison model of Korean machine translation is established by unsupervised neural network learning method,and iterative reverse translation and back translation are realized.In the system construction,the database module,semantic recommendation module,data access layer module and translation generation module are designed,and the machine translation system is designed under the improved neural network model.The test results show that this method is reliable in Korean machine translation,with good semantic coherence,translation consistency and translation fluency,and the accuracy of translation output is high.