不同语义认知视角下交互式智能翻译方法研究

Research on Interactive Intelligent Translation Methods from Different Semantic Cognitive Perspectives

为提升翻译方法的准确率,改进交互式智能翻译策略,提出不同语义认知视角下交互式智能翻译方法。利用翻译模型从资料库中获取待翻译信息,根据待翻译信息语义信息列举语义单元,获取句子中不同单词的语义类型与写法;采用人工构建语义语法树方式,获取语义语法模式,得出所需翻译句子的表达形式;采用伯克利对准器对齐词语,实现单词翻译。通过计算翻译模型翻译完成单词的翻译概率,利用词汇化加权公式计算该翻译概率获取短语相似度,生成新的特征,加入解码器,输出翻译结果。实验证明:所研究方法翻译单词时的平均准确率高达97.9%,平均召回率高达96.4%,翻译句子时准确率超过96%,具备较高的翻译准确率,为实际智能翻译方法提供参考。

In order to improve the accuracy of translation methods and interactive intelligent translation,this paper proposes interactive intelligent translation methods from different semantic cognitive perspectives.The translation model is used to obtain the information to be translated from the database,and the semantic units are listed according to the semantic information of the information to be translated,and the semantic types and writing methods of different words in the sentence are obtained;the semantic grammar tree is constructed manually to obtain the semantic grammar mode,and the expression form of the required translation sentence is obtained;the Berkeley aligner is used to align words to realize word translation.By calculating the translation probability of words in translation mo-del,the translation probability is calculated by lexicalization weighting formula to obtain phrase similarity,generate new features,add decoder,and output translation results.The experimental results show that:the average accuracy rate of the research method is 97.9%,the average recall rate is 96.4%,and the translation accuracy rate of sentence translation is more than 96%.The high translation accuracy provides a reference for the practical intelligent translation method.