大学英语翻译考试人工辅助计算机评分初探

Research on the Human-aided Auto-assessment for Translation Tests in College English

翻译考试一直被看成是一种很需要人工和智力投入的考试，传统上大学英语中的翻译考试都是由人工进行评阅的，基于大学英语考试参加人数多、教师工作量大等特点，本文从研究机器翻译自动评价的现有成果入手，探讨了大学英语翻译考试在人工辅助下经由计算机自动评价的可能性与相关实现方法。本文的研究对象是北京大学大学英语教研室在本科大学英语教学中进行的期末考试中翻译部分的评分工作，针对评卷教师反映评卷工作量大、后期由于疲劳准确度下降等问题而开展的。就大学英语考试的特点，本文提出利用相关的软件、工具和算法，在已有研究的基础上实现大学英语翻译考试网络化和自动评分。

Translation tests have been regarded as a process which intelligence. Traditionally, tests as such are conducted by human work, requires an investment of both human labor and which involves both organization and evaluation. With a large student population and huge workload of teacher＇ s marking, there ought to be some way to change the current situation with the help of computer technology. This paper conducts a research on the automatic assessment methodology of translation tests in College English based on the theories on auto-assessment of machine translation. It also proposes related algorithms, tools and softwares in order to realize the human-aided automatic assessment of translation tests.