汉语比较句识别研究精选.pdfVIP

22 5 Vol. 22, No . 5 2008 9 JOU RNAL OF CHINESE INFORMAT ION PROCESSIN G Sep., 2008 : 100 0077( 2008) 0500 009 , , , ( , 100871) : 比较是常见的表达方式, 提取事物之间的比较关系是一项新颖而有实用价值的研究识 自然语言中的 比较句, 是提取比较关系的一个重要步骤目前还没有针对汉语比较句的自动识 研究, 语言学上比较句的哪些 特征能够应用到自动识 上来是一个亟待研究的问题该文讨论了汉语比较句的范畴外延和特征, 定义了汉语 比较句识 的任务, 并提出用 SVM 分类器将汉语句子分为 比较 和 非比较 两类该文比较了比较句的语言学 特征和统计特征, 包括特征词序列模式等在分类中的作用实验结果表明: 基于类序列规则的SVM 分类器能够 有效地识 汉语比较句, 效果优于传统基于词的文本分类 : 计算机应用; 中文信息处理; 汉语比较句识 ; 比较挖掘; 文本分类; 序列模式 : T P 91 : A Learningto Identify Chinese ComparativeSentences HUANG Xiaojiang, WAN Xiaojun, YANG Jianwu, XIAO Jianguo ( Instit ute of Computer Science and Technology of Peking Universit y, Beijing 100871, China) Abstract: Comparison is a common kind of expression, and it is novel and subst antial research to extract comparative relat ions betw een object s. Identifying comparat ive sentences in natural language is an important step in extracting comparative relat ions. T o our knowledge, t here is no research on identifying Chinese comparative sentences automatically . This paper f irst defines the problem of Chinese comparat ive sentence identification, and then proposes to use SVM to classify a Chinese sentence into either comparative or not. Various linguistic and stat istical features have been explored, such as keyw ords and sequential patterns. Experimental result s demonstrate the effectiveness of the sequential patterns, i. e. the classifier w ith sequential patterns can significant ly out perform the tradit ional term based classifier. We also empirically investigate the important factors that affect classification performance. Keywords: comput er application; Chinese infor