Abstract
The accurate evaluation of machine translation quality is the main basis of machine translation systems research and development. Based on the analysis of three translation quality assessment factors, i.e. vocabulary, grammar and discourse, this paper built twelve indicators, which classified into five levels of evaluation index system. Meanwhile questionnaire was designed as well. The people with different age, gender, occupation and cultural level were selected as respondents. Then, the data for machine translation evaluation was obtained from questionnaire. The analytic hierarchy process was applied to determine the weighted vector of machine translation evaluation factors. In addition, on this basis, fuzzy mathematics theory was used to establish the comprehensive evaluation vector of machine translation quality. As a result, the quantity comprehensive evaluation for the translation text quality was realized.
Highlights
Machine translation is exactly the key to the cultural exchange and information searching among different languages
An accurate evaluation of translation quality is the main basis for the system development
Evaluation of the translation’s similarity between references and machine translation is one of the main methods[1].For instance, the automatic evaluation method developed by IBM named BLEU[2];the developed by Yokoyama based on bidirectional machine translation[3], the method of Yasuda, Akiba and Papineni based on Ngram language models for computing sentence similarity[4][5]
Summary
Machine translation is exactly the key to the cultural exchange and information searching among different languages. Evaluation of the translation’s similarity between references and machine translation is one of the main methods[1].For instance, the automatic evaluation method developed by IBM named BLEU[2];the developed by Yokoyama based on bidirectional machine translation[3], the method of Yasuda, Akiba and Papineni based on Ngram language models for computing sentence similarity[4][5] These evaluation methods have two problems: first of all, they need the help of a third partyan artificial translation as reference. The language is flexible, because of its lexical, grammatical, syntactic and contextual changes, the meaning will be very different These evaluation methods are limited to a micro level and lack of analysis of the article from a macro point of view. All of the evaluation factors are compared in pairs based on the principle of analytic hierarchy process to construct the comparative judgment matrix, and we examined the consistency of the comparative judgment matrix, we got the weight of each evaluation index and each evaluation level
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
