Automatic Evaluation of Machine Translation Based on Linguistic Knowledge

Abstract

On the study of machine translation, translation system evaluation method is one of the most important links, which plays a very important role in the study of machine translation. Translation quality can be evaluated both manually and by computer. Among them, manual evaluation is a more commonly used means, because it has many advantages. However, there are two problems in the manual evaluation: first, the evaluation takes a long time; second, the manual evaluation will be subjective. In addition, due to the characteristics of machine translation itself, manual evaluation cannot fully meet the needs of machine translators. The automatic evaluation method comes into being, which just fills the defect of manual evaluation. At present, the research on automatic evaluation of machine translation is not deep enough, and the existing automatic evaluation system of machine translation is mostly aimed at a certain type of text test, cannot fully reflect the degree of difference between different texts and the same text in various fields. This paper proposes a method for automatic evaluation of machine translation based on linguistic characteristics. Through analyzing the defects of common automatic evaluation methods such as BLEU and NIST, this method is improved by using regression model of support vector machine, and based on linguistic knowledge, A system for classifying linguistic phenomena or fragments of a sentence from a linguistic perspective. Finally, the experimental results show that the recall rate and accuracy of the automatic evaluation method of machine translation in this paper are 90% and 96%, respectively. Compared with BLEU and NIST two traditional automatic evaluation methods, the automatic evaluation method of machine translation in this paper has a better result.

Related Papers

• → Design and Testing of Automatic Machine Translation System Based on Chinese-English Phrase Translation(2021)7 cited
• → Productivity and quality when editing machine translation and translation memory outputs: an empirical analysis of English to Welsh translation(2017)15 cited
• A Hybrid Approach to Example based Machine Translation for Indian Languages(2007)
• → Automatic Post-Editing Method Using Translation Knowledge Based on Intuitive Common Parts Continuum for Statistical Machine Translation(2014)
• Основные факторы улучшения машинного перевода(2015)