【摘要】 本文展示一种认知语言学研究的计算范式与技术路线:首先基于语言使用是一种以知识为基础的认知过程的假设,以计算机模拟人类理解自然语言的意义为技术背景,用情境网络对特定领域的有关语言表达及其意义进行认知建模;再用逻辑表达式刻画关键性的情境要素之间的语义关系,并用语义公理来表示有关逻辑表达式之间的推导关系,从而形成概念层面的知识推理和语义计算的知识单元;然后,为有关语言现象建立词汇句法知识库,从词汇、语义和句法分布等方面对跟有关情境相对应的词汇和句式进行详细的描写,并且在情境网络、词汇—句法知识库和真实文本中的有关描述实例之间建立双向联系;最后,形成完整的词汇语义计算的知识平台,再应用到有关的自然语言处理系统中。文章还通过"领属转移"这个实例,对情境网络和词汇句法知识库的开发过程进行简要的说明;通过"亲属关系的自动推理"这个专家系统的研究和开发过程,对认知模型和情境网络、逻辑表示和语义公理、词汇—语义—句法知识库的建立、乃至算法设计和程序实现的全过程进行例示。更多还原

【Abstract】 In this paper we propose a computational conceptualization of cognitive linguistics and construct a semantic computational platform for natural language understanding.Its technological approach mainly consists of three steps.Firstly,we form the conception unit of knowledge for logic inference and semantic computation by means of following procedures:(1)using situation networks to build the cognitive model for the meaning of the related linguistic expressions with respect to a special domain,(2)using logic formulas to describe the semantic relation among the key situation elements,and(3)using semantic axioms to represent the inference relation among the related semantic logic formulas.Secondly,we construct a lexical-syntactic knowledge base which describes the lexical,semantic,morphological and syntactic knowledge of the related lexicons with respect to the corresponding situations comprehensively.Thirdly,we build a lexical-semantic computational infrastructure by setting the bi-directional linkages among the situation networks,the lexical-syntactic knowledge base and the real text situation instances.To demonstrate the feasibility of the above computational paradigm and technological approach,we employ two cases of the development of a semantic situation network for possession-transferring domain and an expert system for Chinese kinship automatic inference.更多还原