基于词间依存的汉语基本块依存关系识别  

Chinese Chunk Dependency Relationship Parsing Based on Words Dependency

作　　者：李丽[1] 赵文娟[1] 樊孝忠[1] 

机构地区：[1]北京理工大学计算机学院,北京100081

出　　处：《计算机科学》2013年第11A期259-262,共4页Computer Science

摘　　要：基本块的分析是句法分析中的重要技术,根据依存理论,提出了一种分析基本块之间的依存关系的方法。首先使用BIO标记来识别基本块,然后根据词之间的依存关系判别基本块之间的依存关系。实验表明,基本块识别的正确率和召回率分别为82.3%和78%,基本块之间依存关系识别的正确率和召回率分别为89%和90.5%。Chinese Chunking is an import technique in the Syntax parsing. This paper adopted the theoretical analysis of dependency to parse the Chinese Chunk Dependency Relationship. First of all, BIO tags are used to identify the Chinese chunk, then according to the dependency of words to distinguish the dependency of Chinese Chunk. Experimental results show that, the Chinese Chunking rates of accuracy and recall are respectively 82. 3 % and 78%, and the Chinese Chunk Dependency rates of accuracy and recall are respectively 89% and 90. 5% .

关 键 词：基本块 依存关系 词之间的依存关系 句法分析 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]