机器翻译研究综述

An Overview Study on Machine Translation

机器翻译历经几十年的发展变化,从"机械大脑"到基于规则的机器翻译、基于实例的机器翻译、基于统计的机器翻译和基于深度学习的神经网络机器翻译,在技术上不断推陈出新,算法上不断优化升级。机器翻译在不同的发展阶段呈现出不同的特点,厘清机器翻译系统各阶段的技术内涵与特征、发展状况、优势及局限,反思机器翻译发展面临的机遇与挑战,在机器翻译系统快速发展的现阶段显得尤为必要。

Over decades of evolution,machine translation(MT)has experienced developments from“mechanical brains”to rule-based MT,example-based MT,statistical MT,and deep learning neural MT.Each new development witnessed innovation in technology,as well as optimization and upgrading in metrics and algorithms.Machine translation has presented different characteristics at different stages of development.It is of great necessity to clarify the technical connotation and characteristics,history of development,advantages and limitations of each MT system,to discuss pressing issues regarding MT technologies,and to reflect on the opportunities and challenges faced by MT under current situation.