基于交互式机器翻译的译文查询行文预测技术

Research on the Technology of Text Query Based on Interactive Machine Translation

随着信息交流的密切,人工翻译工作量大,且收益较低,矛盾凸显。文中基于此背景以交互式机器翻译技术(IMT)为核心,针对各类用户翻译过程中查询调用行为出现的频繁鼠标、键盘切换问题,提出了一种交互机器替代的智能预测模型。此预测模型采用翻译条件选择机制,搭配对齐模型、翻译模型、语言模型等进行全面语义分析,在较大程度上保证了查词行为预测的可行性。经过测试,在人工双语对齐类语料中,预测准确性达64.99%,尤其在各类语义明确的名词类句段预测时,精确度可达72.28%。基于此种执行效率,机器交互翻译系统虽无法完全替代人工翻译过程,但可大幅减少重复、底层的劳动行为,使人从机械的操作中解放出来。在改善用户交互翻译体验的同时,大幅提升工作效率。

Nowadays,the exchange of information is very close,and the translation and inquiry of documents become the high frequency behavior in the process of research and study. However,the workload of manual translation is too large,and the income is low. Based on the background of interactive technology to Machine Translation（IMT） as the core,to solve the problem of frequent switching behavior of mouse and keyboard call query all kinds of users in the translation process,put forward a prediction model of intelligent interactive machine replacement. This prediction model is based on the translation condition selection mechanism,with the alignment model,the translation model and the language model. After testing,the accuracy of prediction is 64. 99%,which is more than 72. 28%,especially in all kinds of semantic NOUN class. Based on the efficiency of the system,the system can not completely replace the manual translation process,but it can greatly reduce duplication,the bottom of the labor behavior,so that people from the mechanical operation of the liberation. While improving the user interaction translation experience and greatly improve work efficiency.