基于多知识源的词汇消歧一体化处理  

AN INTEGRATING PROCESSING MECHANISM FOR WORD DISAMBIGUATION BASED ON MULTIPLE KNOWLEDGE SOURCES

作　　者：郭宏蕾[1] 姚天顺[2] 

机构地区：[1]北京航空航天大学计算机系 [2]东北大学计算机系

出　　处：《小型微型计算机系统》1999年第7期531-535,共5页Journal of Chinese Computer Systems

基　　金：国家自然科学基金;863高科技项目

摘　　要：词汇消歧是语言分析的基石．本文提出一种基于多知识源的词汇消歧一体化处理机制．该机制充分利用了知识库和文本结构的信息，以句法标签、词频、搭配、上下文语义、语义可选约束、句法线索等知识源为消歧指示器，实现各类词汇歧义的一体化处理．在词汇消歧一体化处理中，不依据固定有序的规则集进行消歧，而是根据反映信息指示特性的消歧度动态权衡各种指示器的有效性，择优求解．Lexical disambiguation is the basic of language analysis. This paper presents an integrating processing mechanism for lexical disambiguation based on multiple knowledge sources. This mechanism disambiguates the variety of lexical ambiguities integrally with all information on knowledge base and text constructs. The knowledge source such as syntax tags, word frequencies, collocation , context, semantic optional constraint, syntactic restricts serve as indicators for disambiguating. This disambiguation approach makes an optimize solving by dynamically detecting the effective of all indicators according to the disambiguation degree of the indicating information properties.

关 键 词：自然语言处理 词汇消歧 知识源 一体化处理 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]