面向机器翻译的假朋友语义自动校对

摘要

Analyze the semantic problem of false friends in statistical machine translation,in order to promote the accuracy of False Friends translation.It provides the technology of false friends recognition based on XML rules,which can correct the false friends translation errors in the existing translation results based on constructing the rule repository.The correction is automatic powered by the translation software developed by the authors.The test results of four mainstream online machine translation tools show that it can greatly promote the accuracy of Chinese-English false friends translation.It only tests the Chinese-English false friends translation,without involving other languages.Rule based method can effectively complements the static machine translation in regard to the semantic problem of false friends.It's suggested that the machine translation tools can improve the word and rule repository of false friends.%对统计机器翻译的假朋友语义问题进行分析,力求提高假朋友翻译的准确率.提出基于XML规则的假朋友识别技术,在构建规则库基础上对现有机器翻译结果中涉及假朋友的错误翻译进行语义校对,并开发了相应的机器翻译软件实现自动校对.对4个主流在线机器翻译工具进行实验测试结果表明,该方法能大幅度提高中英假朋友的翻译准确率.基于规则的方法能对统计机器翻译在假朋友语义问题上提供有效补充,建议机器翻译工具在假朋友词典和规则方面进行改进.