基于概念基元的句群情景框架抽取研究  

Extracting the Situation Framework of Sentence Group Based on Concept Primitives

作　　者：韦向峰[1] 缪建明[1] 张全[1] 池毓焕[1] 

机构地区：[1]中国科学院声学研究所,北京100190

出　　处：《微计算机应用》2010年第4期21-24,共4页Microcomputer Applications

基　　金：国家973项目"自然语言理解的交互引擎研究"(2004CB318104);国家科技支撑计划课题"搜索引擎中的语言翻译基础研究"(2007BAH05B02-05);中科院声学所知识创新工程项目"句群理解处理理论及其应用"(O654091431);中国科学院声学研究所"所长择优基金"(GS13SJJ04);中国科学院青年人才领域前沿项目(O754021432)的资助

摘　　要：句群的构成要素是领域、情景和背景,领域是核心。领域关联的情景框架主要依据领域句类知识库的描述,从句群语句的语义块中抽取。本文介绍了在语言概念空间中对句群情景框架的概念符号描述,通过匹配整合句群情景框架中语义块的概念约束和句类分析结果中的语义块,给出了从自然语言空间的句群中抽取情景框架要素的方法,并根据实验结果总结了句群情景框架抽取中需要加强和完善的地方。Domain, situation and background are three basic elements of a sentence group. Extracting situation frame from semantic chunks of sentences in a sentence group should base on the knowledge description which saved in domainal sentence category knowledge base. This paper introduces the description of situation frame with conceptual symbols in linguistic conceptual space. By matching the semantic chunks in the result of analyzing sentence semantic categories with the conceptual restriction of semantic chunks in the frame- work of situation, it also shows how to extract the elements of situation frame from sentence group in natural language. Finally, some aspects to be improved are shown through an experiment.

关 键 词：句群 情景框架 领域句类 语义块 句类分析 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]