Automatic assessment for Chinese-to-English translation questions in CET English

• Main Author: Huang, Ping
• Other Authors: Hui Siu Cheung
• Format: Final Year Project
• Language: English
• Published: 2015
• Subjects: DRNTU::Engineering::Computer science and engineering
• Online Access: http://hdl.handle.net/10356/62626
• Institution: Nanyang Technological University

The College English Test (CET) is a national English as a foreign language test in the University of People's Republic of China. In general, both undergraduate students and postgraduate students have to take CET for testing their English proficiency. There are two types of CET, CET-4 and CET-6. In both types, the Chinese-to-English (CE) translation question is an essential part of the paper. The most recent reform of CET-46 happened in the December of 2013, which brought a great difference to the traditional translation question. On top of it, the overall difficulty degree of the translation was increased as well. In order to excel in the translation question in both CET-4 and CET-6, having enough practices is essential. However, it is a time consuming and tedious process for teachers to mark students’ translations. Besides, e-learning becomes more and more popular in today’s society. Thus, an automatic translation assessment system will be very useful and helpful for both students and teachers. This project aims to build a complete computerized automatic translation assessment system, named as “CET Translation Master”, for providing a web-based environment for students to practice for CET translation questions. In addition, the different assessment techniques are investigated to give the instant feedback to user. Both semantic and grammar analysis need to be taken into consideration when doing the assessment. However, this project mainly focuses on the semantic assessment techniques for the CE translation. Moreover, various tests have been carried out to evaluate the different techniques’ performance. Although the final ensemble approach has only obtained 67% accuracy for the sentence translation, it is capable of achieving an average accuracy of 81% for the paragraph translation. Lastly, some suggestions are given for the future work.