CASIA OpenIR  > 毕业生  > 博士学位论文
基于概念网络的汉语问答技术研究
其他题名Research on Chinese Question Answering Technologies based on Conceptual Network
任禾
学位类型工学博士
导师杨一平
2007-06-18
学位授予单位中国科学院研究生院
学位授予地点中国科学院自动化研究所
学位专业计算机应用技术
关键词概念网络 问答系统 汉语信息处理 语义理解 Conceptual Network Question Answering System Chinese Information Processing Semantic Understanding
摘要自1950年图灵测试提出以来,问答系统一直是人工智能领域研究的重要课题之一。相对于英文问答系统的迅速发展以及语义信息处理技术的广泛应用,目前语义分析的方法在汉语问答系统中的应用还不多。因此,本文主要从语义分析的角度,采用一种新的知识表达体系——概念网络,来对问答系统的各个环节展开研究。本文的工作主要有以下几部分: 第一,对概念语义复合的形式化做了进一步的工作。本文在前人工作的基础上,提出了串概念、λ表达式、语义模板和语义情境等语义表示方法。 第二,设计了层次概念复合算法,通过层层概念复合,最后用一个大的语义结构来表达句子的语义。 第三,通过构建数据描述概念,将问句的语义匹配过程和回答信息的抽取过程独立开来,从而使问句的语义匹配方法不仅仅适用于限定领域问答系统。 第四,设计了汉语疑问代词概念模型,并利用它来确定汉语特指问的疑问焦点。 最后,设计了基于概念网络的汉语回答的生成方法。我们主要借鉴前人的工作,采用管道型结构,分别对文本规划,句子规划和语法实现三个阶段建模,充分发挥概念网络语义表达的优势,最终将回答信息用比较自然的方式表达出来。
其他摘要Since turing testing came up in 1950, question answering system has always been one of the hot topics in artificial intelligence field. Compared with the rapid development of English question answering systems in which semantic processing has been largely used, Chinese question answering systems have much a little use of semantic processing. Therefore, this thesis is mainly from the perspective of semantic processing to solve the problems of question answering. A new kind of knowledge representation method, conceptual network, is adopted, based on which we carry out studies for each parts of the question answering system. The work mainly consists of the following aspects: First, further works has been done on the formalization of semantic compounding of concepts. Based on the previous works, several semantic representation methods are proposed, such as serious concept, λ expression, semantic pattern and semantic scene. Second, we have designed a concept compound algorithm to express the meaning of natural language sentences with a large semantic structure. Third, semantic matching and answer extraction are made independent by constructing data description concepts. This makes the semantic matching method can be used more widely. Fourth, Chinese interrogative concept model is built,and then an algorithm is designed to compute the references of interrogative concepts. Finally, an answer generation method based on conceptual network is developed. In virtue of predecessors’ work, we use the pipeline structure to model the text planning、sentence planning and grammar realizing, respectively. It takes good advantage of the conceptual network, and can express the answer in a natural way.
馆藏号XWLW1122
其他标识符200418014690004
语种中文
文献类型学位论文
条目标识符http://ir.ia.ac.cn/handle/173211/6022
专题毕业生_博士学位论文
推荐引用方式
GB/T 7714
任禾. 基于概念网络的汉语问答技术研究[D]. 中国科学院自动化研究所. 中国科学院研究生院,2007.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
CASIA_20041801469000(3225KB) 暂不开放CC BY-NC-SA请求全文
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[任禾]的文章
百度学术
百度学术中相似的文章
[任禾]的文章
必应学术
必应学术中相似的文章
[任禾]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。