CORC  > 自动化研究所  > 中国科学院自动化研究所  > 毕业生  > 硕士学位论文
题名自然语言时间语义信息处理
作者杜津
学位类别工学硕士
答辩日期2005-04-30
授予单位中国科学院研究生院
授予地点中国科学院自动化研究所
导师杨一平
关键词自然语言理解 概念网络 本体论 时间语义 时间模型 NLU ConNet Ontology temporal semantic structure time model
其他题名Temporal Semantic Information Processing In Natural Language
学位专业计算机应用技术
中文摘要在自然语言当中,所描述的语义总是发生在一定的时空之中。时间信息已经成为自然语言表达中必不可少的一种语义信息。如何有效的计算和模拟自然语言中的时间语义,在人工智能自然语言理解领域具备重要的意义。本文以本体论框架——概念网络为基础,深入分析自然语言时间信息表达特点,构建了一种比较完全的自然语言时间表达模型,并对现代汉语中时间语义信息的提取和计算进行了初步的研究和尝试,取得了一定的成果。 本文首先分析了自然语言中时间语义表达的特点,为其两种主要的表达方式,绝对时间表达和相对时间关系表达分别建立对应模型,并将这两种模型通过时间历史有机的结合成为一个整体。绝对时间模型以时间单位系统作为度量的基础,以日历系统为核心,可以有效地量化时间段,为时间段的计算打下了基础;相对时间关系以十三种基本时间段关系为表达基础,可以有效地表达和推理时间段之间的两两关系。时间历史以绝对时间段为核心组织相对时间关系,为自然语言的时间语义提供了统一的模型。 概念网络提供了对现代汉语语义的分割,以概念为基本单元,并对现代汉语文本提供了分词以及初步语义分析功能。本文以概念为基础,分析了汉语中各种时间相关概念的特征以及语法功能,将概念分为两大类型:实词概念与虚词概念,分别针对这两种不同的时间概念建立了对应的时间对象和时间关系映射规则集。在此基础之上,进一步深入分析经过概念网络处理过的现代汉语句子语义结构,结合各个时间概念的时间语义映射模型,生成统一的时间历史结构,实现了现代汉语时间语义的有效处理。 这一工作充实了概念网络本体论框架,扩展了概念网络模型的具体语义,实现了其时间语义模型。为概念网络进一步构建其他的背景知识,充实和丰富概念的其他具体语义,提供了借鉴和初步的研究实践经验。 随着时间概念语义的不断充实,各种时间关系映射规则集和时间对象进一步丰富和完善,将为概念网络提供深入的时间语义处理能力。而随着概念网络具体语义处理能力的逐步丰富和强大,包括时间语义处理能力,空间语义处理能力等的扩展,将为自然语言语义处理的发展打下坚实的基础。
英文摘要In natural language, the meanings of sentences are all located in some positions of space-time. Temporal information has been one kind of necessary semantic information in natural language expressions. It’s significant for Natural Language Understanding to effectively process temporal semantic structures. This article analyzes the characteristics of temporal information expressions in natural language on the basis of Ontology framework---ConNet, builds a uniform time representation model and tries to extract and process the temporal semantic structures in modern Chinese sentences. By now we have made some achievements. At first we discuss the characteristics of temporal expressions in natural language, classified them into two classes: absolute time expressions and relative time expressions. Two different models for the two kinds of expressions have been built and organized into a united semantic structure. The absolute time model based on time granularities system and calendar system can effectively measure time; the relative time model has thirteen basic relationships for intervals, and is capable of expressing and reasoning the relationships between intervals. The time history organizes intervals’ relationships according to absolute time intervals, forms the uniform model for temporal semantic structures. ConNet (concept network) provides the segmentation of semantic meanings. It uses the concept as the basic semantic unit, has the abilities of segmentation function and preliminary semantic analysis functions for Chinese sentences. We analyze the characteristics of temporal concepts and their grammar functions, classified them into two classes: substantives and functional words. Substantives are mapped into temporal objects in time model. Functional words are mapped into rule sets for the construction of time relations. After detailed analysis of the preliminary semantic structures of sentences, combined with respective time structures of temporal concepts, the uniform time histories can be created and processed. Our work enriches the knowledge of ConNet’s framework, extends its domain semantic meanings, and realizes its time semantic model. A method for ConNet to construct background and domain knowledge is proposed and set as an example for reference. ConNet will possess powerful ability to process temporal semantic meanings with the enrichment of temporal concepts’ semantic structures and extensions for time objects and mapping rule sets. The increasing abilities of processing domain semantic meanings, including time and space semantic meanings processing abilities for ConNet will form solid foundations for advanced natural language semantic processing.
语种中文
其他标识符200228014603542
内容类型学位论文
源URL[http://ir.ia.ac.cn/handle/173211/6877]  
专题毕业生_硕士学位论文
推荐引用方式
GB/T 7714
杜津. 自然语言时间语义信息处理[D]. 中国科学院自动化研究所. 中国科学院研究生院. 2005.
个性服务
查看访问统计
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。


©版权所有 ©2017 CSpace - Powered by CSpace