基于规则的汉语复句层次关系自动识别研究  

A Study on the Rule-based Automatic Identification of Hierarchy Relations in Chinese Compound Sentences

作　　者：吴锋文[1] 胡金柱[1,2] 肖明[1,2] 肖升[1] 舒江波[1] 

机构地区：[1]华中师范大学语言与语言教育研究中心,湖北武汉430079 [2]华中师范大学计算机科学系,湖北武汉430079

出　　处：《华文教学与研究》2010年第1期85-92,共8页TCSOL Studies

基　　金：教育部社科基金重大项目"面向中文信息处理的现代汉语有标复句联结机制及分析策略"(07jjd740063);华中师范大学"211工程"项目"小句中枢说在汉语与临界领域研究中的应用与验证"子课题

摘　　要：汉语复句层次关系自动识别是中文信息处理中很有意义而极具挑战性的一项课题。为解决由于关系词省略和关系词不能标示分句间层次组合顺序的局限,考察了1-2型和2-1型三分句复句关联模式的关系词隐现形式,发现12种关系词隐现形式的复句有望实现层次关系识别,这些隐现形式的关系词句法特征与分句组合顺序间存在4种制约规则。结合这些隐现形式的特征信息和分句层次组合制约规则,提出了一种基于规则的层次关系识别流程,以期为实现复句层次关系自动识别打下基础。Automatic identification of hierarchy and semantic relations in Chinese compound sentences is a significant and challenging research project in field of Chinese information processing. Relational words play an index role in identifying hierarchy and semantic relations in Chinese compound sentences automatically. However, relational word is often omitted or can＇t indicate the limitations of order combination of hierarchy among subordinate sentences. So it is necessary to examine the appearance and omission mode of relational words in clauses and combination orders of clauses. For Chinese compound sentences composed of 3 clauses, 12 kinds of appearance and omission mode of relational words can be identified automatically, and there are 4 restricting rules between syntactical features of relational words and combination orders of clauses. Based on these rules, an automatic iden- tification procure of hierarchy and semantic relations in Chinese compound sentences is put forward, which will lay the foundation for automatic identification of hierarchy and semantic relations in Chinese compound sentences.

关 键 词：规则 复句 层次 语义关系 自动识别 

分 类 号：H146.3[语言文字—汉语] TP391.1[自动化与计算机技术—计算机应用技术]