基于随机行走N步的汉语复述短语获取方法.PDFVIP

基于随机行走N 步的汉语复述短语获取方法 马军，张玉洁，徐金安，陈钰枫 （北京交通大学，北京100044 ） 摘要：在利用大规模双语语料获取复述知识中，传统的基于“枢轴”方法只能考虑两步以内的复述现象。 本文针对已有方法的局限性，对不同语言之间互为翻译的短语对，构建基于图的复述获取模型，提出基于 随机行走N 步的复述获取算法，改进已有方法以获取更多潜在的复述知识。本文描述了以汉英短语翻译表 为基础的图模型、基于N 步的随机行走算法和基于期望步数的复述短语可信度计算方法。同时，我们在图 模型基础上提出基于多语言对扩展的方法。我们在NTCIR 汉英、英日双语平行语料上进行了实验与评测， 并与已有方法进行了对比。实验结果表明本文所提出的方法能够获取更多的复述知识，而且扩展语言对的 图模型能够有效获取更多潜在的复述知识。 关键词：复述获取；随机行走；图模型 中图分类号：TP391 文献标识码：A Chinese Paraphrases Acquiring Based on Random Walk N Steps Jun Ma, Yujie Zhang, Jinan Xu, Yufeng Chen (Beijing Jiaotong University, Beijing, 100044, China ) Abstract: Conventional “pivot” approach of acquiring paraphrases from bilingual corpus has limitations, where only candidated paraphrases within two steps are considered. In this paper, we propose a graph based model of acquiring paraphrases from phrases translation table. First, we describe a graph model based on Chinese-English phrases translation table, a random walk algorithm based on N number of steps and a confidence metric for the obtained paraphrases phrases. Furthermore, we augment the model to be able to integrate more language pairs, for instance, English-Japanese phrases translation table aiming at finding more potential Chinese paraphrases. We performed experiments on NTCIR Chinese-English and English-Japanese bilingual corpora and compared with the conventional method. The experimental results show that the proposed model acquired more paraphrases, and the performance was improved further after English-Japanese phrases translation was added into the graph model. Key words: Paraphrases acquisition; Random walk; Graph model 1 引言 [1] 复述（Paraphrases）是指具有相同语义的不同表达 ，是自然语言中的普遍现象，体现 了语言的多样性与复杂性。近年，复述处理在自然语言处理领域日益受到关注，在机器翻译 [2]