语义信息处理及神经网模型

Semantic Information Processing and Model of Neural Network

本文从语言信息的角度全面分析了当前语义形式化的复杂性和困难性。在此基础上系统地提出了语言信息度量化的概念,进而阐明了抽象数学模型的语义分析方法和模型构造原则。在构造了一种语义处理神经网模型基础上,应用其并行、非线性、非逻辑的处理机制,分析处理了两个基于语用知识的自动分词难题。本文对一些新的概念“隐式知识表达”、“语法信息”、“语义信息”、“语用信息”等进行了定义和描述。神经网机理和仿真结果表明这种理论具有许多新的特点和长处。

The paper thoroughly analyses the complications and difficuties of current semantic -formulation based OB nature language information processing, proposes the conception of magnitude in the nature language information and furthermore demonstrates the semantic analysing methods on abstract mathematical model and principles of framing mathematical model. After having constructed a neural network of semantic processing, the article has analysed and processed the two difficult problems of automatically segregating chinese words which can heen done correctly only based on semantic knowledge. The paper represents some new conceptions which are .'hidden knowledge representation', 'gra-mmar information' 'pargmatic information' and gives exact definitions of them. The problem of semantic formulation is explored initiatelly. The mechanism of neural network and it's simulation denotes that the new theory has many new advantages and features,