汉语复句中分句的语义关联特征  

Semantic Relatedness of Clauses in Chinese Compound Sentences

作　　者：胡金柱[1,2] 舒江波[1,2] 罗进军[2] 

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

出　　处：《语言文字应用》2010年第4期121-130,共10页Applied Linguistics

摘　　要：汉语复句层次关系的自动识别是中文信息处理的难点之一,而其中关系标记非充盈模式下无标分句的层次归属以及关系标记充盈模式下某些特殊因果类、并列类有标复句的分句关联的确定更是难中之难。文章提出利用分句语义关联理论来解决这类问题:1)从句法和语义两个角度,总结出标示分句语义关联的3大类14个小类;2)根据特征之间的相互约束规律和语料中分句特征的统计规律,确定特征分析的先后顺序;3)讨论分句间语义关联度的计算方法,根据分句间的关联度,确定分句的层次归属。The automatic recognition of hierarchy and relationship of Chinese compound sentences has been a knotty problem in Chinese information processing,and it is even knottier on two points,one is the hierarchy ascription of non-marked clauses,the other is the confirmation of relevancy of clauses in some causal and coordination marked compound sentences.It uses semantic relevancy theory to solve these problems in this paper.The discussion focuses on three parts： sum up the semantic features which indicate the semantic relatedness between clauses from the syntactic and semantic aspects;confirm the analysis order of features according to the restriction law among features and thestatistic result of the occurrence of clauses’features in corpus;discuss the computing method of the semantic relatedness degree between clauses,and then confirm the hierarchy ascription of clauses according to semantic relatedness degree.

关 键 词：复句 语义关联特征 层次归属 自动识别 

分 类 号：H146[语言文字—汉语]