【作者】 张涛；

【导师】 廖乐健；

【作者基本信息】 北京理工大学 ， 计算机科学与技术， 2010， 硕士

【摘要】 自然语言处理是人工智能领域的分支学科。随着计算机技术的不断成熟,自然语言处理技术获得了飞速的发展,现阶段已经应用到机器翻译、文本分类等多个领域。现阶段,实现句子的语义分析逐渐成为自然语言处理的重点和难点。但传统的自然语言处理理论严格遵循“词法——语法——语义”的处理模式,对于句法规则灵活的汉语来说并不适用,我们迫切需要一种以语义深层结构为基础的自然语言处理理论。HNC理论,即概念层次网络理论,是关于自然语言处理的创新理论体系。其目标是以概念联想脉络为主线,建立一种模拟大脑语言感知过程的自然语言表述模式和计算机处理模式,从而使计算机获得消解模糊的能力。HNC以抽象概念的语义网络、语义块、句类等知识作为其理论支持,重在表达词语、语义块之间的语义关系。本文阐述了HNC理论的两类联想脉络——局部联想脉络和全局联想脉络,深入研究了语义块和句类知识,设计了理解自然语言语句的句类分析算法,并实现了句类分析系统。文中详细介绍了该系统的几个主要模块——语义块感知模块、句类假设检验模块、语义块构成分析模块,并建立了简单的HNC知识库。经实验证明,本文设计的句类分析系统已经实现了简单语句的理解,对输入的自然语句序列经过句类分析后,能够得到正确的句类表示式。更多还原

【Abstract】 NLP(Natural Language Processing) is one of the branches of artificial intelligence. With the rapid development of the computer science technology, NLP has been applied in the fields of machine translation and text categorization. The semantic analysis of sentences becomes more and more important nowadays. The conventional NLP theories which follow the processing pattern of“lexical analysis——syntax analysis——semantic analysis”are not fit for Chinese which has flexible syntax rules. We need a theory which emphasizes the semantic structure.HNC(Hierarchical Network of Concepts) theory is an innovation theory about NLP. It focus on the association of concepts, aims at creating a natural language processing system which simulates the operation of neurons in the brain and enable the computer to eliminate ambiguity. HNC is based on the concepts such as semantic networks, semantic chunks and sentence category.This paper introduces the knowledge of semantic chunk and sentence category, designs an algorithm for natural language understanding and implements a sentence category analysis system. Several important modules such as semantic chunk perception, sentence category assumption test and semantic chunk composition are discussed in detail. Proved by the experiment, the sentence category analysis system takes the simple sentences as the input and outputs the correct sentence category representation.更多还原