利用Markov网络抽取复述增强机器译文自动评价方法Enhance.PDF

利用Markov 网络抽取复述增强机器译文自动评价方法∗ 翁贞，李茂西，王明文 （江西师范大学 计算机信息工程学院，南昌 330022） 摘要: 在机器译文自动评价中，匹配具有相同语义、不同表达方式的词或短语是其中一个很大的挑战。许多研究工作提出 从双语平行语料或可比语料中抽取复述来增强机器译文和人工译文的匹配。然而双语平行语料或可比语料不仅构建成本高， 而且对少数语言对难以大量获取。我们提出通过构建词的 Markov 网络，从目标语言的单语文本中抽取复述的方法,并利用 该复述提高机器译文自动评价方法与人工评价方法的相关性。在WMT14 Metrics task 上的实验结果表明，我们从单语文本 中提取复述方法的性能与从双语平行语料中提取复述方法的性能具有很强的可比性。因此，本文提出的方法可在保证复述质 量的同时，降低了复述抽取的成本。 关键词：复述；机器译文自动评价；Markov 网络；相关性 中图分类号：TP391 文献标识码：A Enhance Automatic Evaluation of Machine Translation by Markov Network-Extracted Paraphrases Zhen Weng, Maoxi Li, Mingwen Wang (School of Computer Information Engineering, Jiangxi Normal University, Nanchang 330022, China) Abstract: It is a great challenge to match words or phrases which have the same meanings but different expressions in the automatic evaluation of machine translation. Many researchers proposed to enhance the matches between the words in machine translation and in human references by extracting paraphrases from bilingual parallel corpus or comparable corpus. However, the cost of constructing the bilingual parallel corpus or the comparable corpus is high; furthermore, it is difficult to obtain some language pairs on a large scale by using corpus. In this paper, the paraphrases were extracted from the monolingual texts in the target language by constructing the Markov networks of words, and applied to improve the correlation between the results of automatic evaluation and the human judgments of machine translation. The experimental results on WMT14 Metrics task showed that the performances of the approach of extracting paraphrase from monolingual text we proposed and that of extracting paraphrase from bilingual parallel corpus have a strong