基于浅层句法分析的中文语义角色标注的研究.pdfVIP

基于浅层句法分析的中文语义角色标注研究 1 * 王鑫 ，孙薇薇，穗志方 北京大学 计算语言学研究所，北京，100871 wang-xin@ weiwsun@ szf@ 摘 要：语义角色标注是获取语义信息的一种重要手段。许多现有的语义角色标注都是在完全句法分析的 基础上进行的，但由于现阶段中文完全句法分析器性能比较低，基于自动的完全句法分析的中文语义角色 标注效果并不理想.因此本文将中文语义角色标注建立在了浅层句法分析的基础上。在句法分析阶段，利 用构词法获得词语的语素信息，有效提高了句法分析的性能，F 值达到了 93.61 。在角色标注阶段，利用 动词的语素信息，并提出了句子的“粗框架”特征，使角色标注阶段的 F 值达到了 74.1，比前人(71.9)有 较为显著的提升，从而证明了笔者的方法是有效的。 关键词：语义角色标注；浅层句法分析；语素；构词法 The Research on Chinese Semantic Role Labeling Based on Shallow Parsing WANG Xin, SUN Weiwei, Sui Zhifang Institute of Computational Linguistics,Peking University, Beijing, 100871 wang-xin@ weiwsun@ szf@ Abstract: Semantic role labeling is an important way to get semantic information. Many existing systems for Semantic Role Labeling (SRL) make use of full syntactic parses, but due to the low performance of the existing Chinese parser, the performance of labeling based on the full syntactic parses is still not satisfactory. So this paper realized SRL methods based on the shallow parsing. In shallow parsing stage, this paper made use of word formation to get morpheme information, the performance of the parser was improved, achieving F-score of 93.61. In the stage of semantic role labeling, this paper applied morpheme information of the verb and made use of a coarse frame feature, these methods have greatly improved the performance of semantic role labeling, F-score of the system has reached 74.1.It has achieved significant improvements over the best reported SRL performance(71.9) in the literature. So it has proved that our method is effective. Keywords: semantic role labeling; shallow syntactic analysis; morpheme; word formation 1 引言 近些年来，多种语言上的语义角色标注逐渐成为研究的热点。现有的