基于条件随机域的复杂最长名词短语识别  

Recognition of Complex Maximal Length Noun Phrase Using Conditional Random Fields

作　　者：冯冲[1,2] 陈肇雄[2] 黄河燕[2] 张亮[2,3] 王江伟[2,3] 

机构地区：[1]中国科学技术大学计算机科学与技术系,安徽合肥230027 [2]中国科学院计算机语言信息工程研究中心 [3]南京理工大学计算机系,北京100083

出　　处：《小型微型计算机系统》2006年第6期1134-1139,共6页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(60272088)资助;国家"八六三"基金项目(2002AA11401)资助.

摘　　要：识别句子中的最长名词短语是一个对机器翻译等任务具有重要实际价值的难题.为了克服传统方法在处理词之间的长程关联的不足和标注偏置等问题,本文采用条件随机域建立统计模型,有针对性的研究了复杂最长名词短语的识别,并给出了一种带置信度估计的解码算法,提高了本文工作的实用性.The recognition of Chinese maximal-length noun phrase is a difficult task, which is valuable for many applications such as machine translation. To overcome the deficiency in capturing the long distance relationship between words and label bias with the traditional methods, a statistical model based on conditional random field is constructed with the focus on the complex maximal length noun phrases. And a decoding algorithm with confidence estimation is given, which is proved to be effective for enhancing the practical usability.

关 键 词：最长名词短语 条件随机域 机器翻译 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]