对答结构的标注与应用研究——以汉语(二语)教学会话体语料为例  

A tagging frame of the question-answer structure and its application:A study of the TCSL conversation corpus

作　　者：杨丽姣[1] 熊文[2] 徐丽芳[3] 

机构地区：[1]北京师范大学中文信息处理研究所,北京100875 [2]中国专利信息中心,北京100088 [3]同方知网(北京)技术有限公司,北京100192

出　　处：《云南师范大学学报（对外汉语教学与研究版）》2015年第3期45-52,共8页Journal of Yunnan Normal University(Teaching & Studying Chinese as a Foreign Language Edition)

基　　金：国家高技术研究发展计划(863计划)"海量文本多层次知识表示及中文文本理解应用系统研制"(2012AA011104)

摘　　要：研究针对汉语(二语)教学会话体语料语义功能的检索需求,基于汉语国际教育动态语料库,探讨了一种语料库语言信息标注框架。围绕日常口语交际的主要目的以及教学范围,提出19类对答结构,刻画了引发语与应答语的基本形式。以此为基础,开展对答结构的自动识别算法研究,选取问候、感谢、祝贺、赞扬、介绍5个类别进行试验,实验测试在准确率和召回率上均取得较好的成绩。对答结构的标注框架对于会话体语料相关表达式的抽取具有较好的适应性,自动识别算法可用于语言信息的自动抽取以及语料库扩展应用软件研发等。With a consideration of the retrieval requirement of the semantic functions for the conver- sation style in the dynamic corpus of the international Chinese education, this paper explores a tagging frame for the corpus-based language information. Focusing on the major goals in daily conversations and the relevant teaching domains, this frame brings up nineteen categories of question-answer struc- tures and describes the basic forms of questions and answers. With this, it discusses an automatic rec- ognition algorithm for the question-answer structures by selecting five categories such as greeting, thanking, congratulating, praising, and introduction as the research objects. The results show that it is fairly satisfactory in terms of accuracy and recall rate as well as applicable in the automatic retrieval of the language information and related studies.

关 键 词：汉语二语教学 标注框架 对答结构 语义功能 自动识别 

分 类 号：H195[语言文字—汉语]