摘要
针对目前分析复句中分句内成分间依存关系的方法多是从分析语法成分间的关系出发,并通过句子内词与词之间的依存关系来揭示句子的句法结构,但容易忽略语言结构的层次而导致在语义理解上有明显缺陷这一问题,在上述分析过程中引入综合分析法的思想,提出了一种新的汉语依存句法分析方法,并以汉语结构类型模板为桥梁,得到结构和语义兼顾的依存关系分析结果。实验结果表明该方法相较于传统方法在依存关系界定性能上有一定提高。
In view of the fact that most of the current methods of analyzing the dependence relations among the components ina compound sentence are based on the analysis of the relations between grammatical components and also reveal the sentence's syntactic structure through the dependence relations between words in the sentence,it is easy to neglect the hierarchy of linguistic structure,which leads to obvious defects in semantic understanding. In the process of the above analysis,the idea of comprehensive analysis is introduced. A new method of Chinese dependency parsing is proposed. Based on the template of Chinese structure type,theresults of dependency analysis with both structure and semantics are obtained. The experimental results show that the performance ofthis method is better than that of traditional methods.
引文
[1]刘伟权,王明会,钟义信.建立现代汉语依存关系的层次体系[J].中文信息学报,1996,10(2):32-46.LIU Weiquan,WANG Minghui, ZHONG Yixing. OnStudy of Hierarchy Structure of Dependency Relations inChinese[J]. Journal of Chinese Information Processing,1996,10(2):32-46.
[2]郑印.基于关系词搭配的汉语复句依存关系的层次体系的自动构建[D].武汉:华中师范大学,2016:12-13,39-40.ZHENG Yin.Automatic Establishment of the Hierarchiesof the Dependency Relation of Chinese Compound Sentence Based on Collocation of Relation Words[D]. Wuhan:Central China Normal University,2016:12-13,39-40.
[3]刁胜权.汉语复句中基于语料统计的决策式依存句法分析方法研究[D].武汉:华中师范大学,2016:34-42.DIAO Shengquan. Research on Decision-making Dependency Syntax Method Based on Corpus Statistics in Chinese Compound Sentence[D]. Wuhan:Central China Normal University,2016:34-42.
[4]Wanxiang Che,Zhenghua Li,Ting Liu. LTP:A ChineseLanguage Technology Platform[C]//In Proceedings of theColing 2010:Demonstrations,2010(8):13-16.
[5]姚文琳.汉语依存句法分析方法的研究与实现[D].青岛:中国海洋大学,2009:27-29.YAO Wenlin. Research and Implementation of ChineseDependency Parsing Method[D]. Qingdao:Ocean University of China,2009:27-29.
[6]房玉清.实用汉语语法[M].北京:北京语言大学出版社,2008:22-30.FANG Yuqing. A Practical Chinese Grammar[M]. Beijing:Beijing Language And Culture University Press,2008:22-30.
[7]袁文宜.依存语法概述[J].科技情报开发与经济,2010,20(18):152-154.YUAN Wenyi. Dependency Grammar Overview[J].Sci-Tech Information Development&Economy,2010,20(18):152-154.
[8]冯志伟.判断从属树合格性的五个条件[C]//全国语言文字应用学术研讨会,1998.FENG Zhiwei. Five Criteria for Judging the Eligibility ofDependent Trees[C]//National Language and LiteratureApplication Symposium,1998.
[9]舒江波.面向中文信息处理的复句关系词自动标识研究[D].武汉:华中师范大学,2011:97-148.SHU Jiangbo. Research on Automatic Identification ofCompound Sentence Related Words in Chinese Information Processing[D]. Wuhan:Central China Normal University,2011:97-148.
[10]胡金柱,沈威,杜超华.基于规则的复句中的关系词标注探讨[C]//学生计算语言学研讨会,2006.HU Jinzhu,SHEN Wei,DU Chaohua. A Study of Relation Word Tagging in Rule-based Compound Sentences[C]//student computational linguistics workshop,2006.
[11]向磊.基于决策树的汉语复句关系词自动识别中规则挖掘方法研究[D].武汉:华中师范大学,2014:30-32.XIANG Lei. Research on Rule Mining Method Based onDecision Tree for Automatic Recognition of Chinese Compound Sentences Related Words[D]. Wuhan:CentralChina Normal University,2014:30-32.
[12]李琼.利用标点符号自动识别分句[J].皖西学院学报,2011,27(1):108-110.LI Qiong. Automatic Recognition of Clauses by Punctuation Marks[J]. Journal of West Anhui University,2011,27(1):108-110.
[13]李琼.汉语复句书读前后语言片段的非分句识别[D].武汉:华中师范大学,2008:39-46.LI Qiong. Non-clausal Recognition of Linguistic Fragments before and after Chinese Compound Sentences[D]. Wuhan:Central China Normal University,2008:39-46.
[14]邢福义.汉语复句研究[M].北京:商务印书馆,2001:38-51.XING Fuyi. Study on Chinese Compound Sentence[M].Beijing:Commercial Press,2001:38-51.
[15]鲁松,白硕,李素建.汉语多重关系复句的关系层次分析[J].软件学报,2001,12(7):987-995.LU Song,BAI Shuo,LI Sujian. Relational HierarchyAnalysis of Chinese Multiple Relational Complex Sentences[J]. Journal of software,2001,12(7):987-995.
[16]肖升,胡金柱,姚双云.小句关联体研究[J].宁夏大学学报(人文社会科学版),2010,32(4):58-62.XIAO Sheng,HU Jinzhu,YAO Shuangyun. A Study ofClausal Association[J]. Journal of Ningxia University(Social Science Edition),2010,32(4):58-62.