关键词: 机器翻译, 机器翻译方法, 语料库法, 神经网络法

Abstract: First of all, machine translation methods are streamlined into three types based on their ways to knowledge processing, i.e. rule-based methods, corpus-based methods, and hybrid methods, and their merits and demerits are compared. Then, the latest study trends mainly comprise computational complexity reducing, words alignment enhancing, prior knowledge and constraints incorporating, and historical memory improving, attention-based neural networks, as the current mainstream methods of machine translation, are captured. Finally, the future development orientations are prospected, particularly, networks integration deepening, data processing parallelizing and training methods diversifying of machine translation methods.

Key words: machine translation, machine translation methods, corpus-based methods, neural network methods