Interactive approach in machine translation systems
Zhang Sen, Chen Zhaoxiong, Huang Heyan · 2002
To get high quality output is the main task of machine translation (MT) research and development. Because of the weakness of the grammar method in representing natural language and the limitation of modern computer architecture based on Boolean logic, the MT results are not up to most users' expectations, and the translation module needs to have access to the knowledge and the reasoning capabilities which is so natural and easy for human beings and still so difficult for computers to solve many ambiguous problems. Therefore, the idea of combining man and computer to achieve the task of natural language translation is reasonable and necessary. The solution is to use man as an important external knowledge source, and through the interactive approach (IA), the translation module may consult its user to solve difficult problems when it faces them. The principle of an interactive approach is to improve the quality of MT systems, but not to reduce the efficiency of the batch process. We propose some key points in designing and implementing the interactive approach in an intelligent machine translation system for English-Chinese.