基于神经网络的汉语文法分析专家系统的设计与实现-模式识别与智能系统专业论文.docxVIP

北京工业大学工学硕士学位论丈统能够快速的记忆和存储知识，进行有效的文法推理。 北京工业大学工学硕士学位论丈 统能够快速的记忆和存储知识，进行有效的文法推理。 6．论文研究了汉语文法分析中采用启发式搜索的方法，通过启发函数值的 性质引入了“概率语法”的概念，推导了概率语法中文法分析的方法及 相关性质。论文针对文法分析中的知识扩充问题，提出了确定编码的论 域和采用多个神经网络同时进行推理的方法。并进一步分析了模糊系统 和神经网络的相似之处，提出了进一步改进的方法。 7．论文设计、实现的汉语文法分析系统在运行中取得了较理想的效果，其 研究成果对汉语文法分析方法的研究具有参考价值。 关键词文法分析：专家系统；神经网络：自然语言 ABSlKAOIThe ABSlKAOI The Design and Implement of the Chinese Syntax Analysis Expert System Based on the Neural Network The thesis deVelops an Chinese syntaX analysis system based on neural network with expert system疳ame on the ground of Chinese syntax analysis，in which constructing t11e knowledge database for saVing a11d maIlaging s”ta)【k110wledge and developing the reasoning machine with the linguist imelligent behaVior and the ability of syntax analysis is me most important． The main research results are clarified as belo、M 1． The Chinese symax analysis expert system based on neural network is proposed．The thesis presents me、Ⅳhole system fr锄ework，and introduces the矗砌e a11d矗mction of each constnlct module in detail．Whars more，Ⅱ1e feasibility of the intelligent system for Cllinese synt“analysis，which is uIlitedby expen system and neural ne铆ork，is discussed in tlle mesis． 2．In the thesis，synta)【knowledge aIld鲥姗-rIlar nlles are absn∞ted from the modem Chinese gmmmar library，a11d the knowIedge express method by pmduct mle is anal sized systematically．In order to make hlowledge mallageable a11d exploit net、vork to save kIlowledge，the grallllllar mles wim product express are staIldardized，thaf is dualistic． 3． The method of saving synta)(knowledge and constnlcting knowIedge databaSe witll nellraI network is pmposed iIl thc t}lesis according to t}le special feature of knowledge description．The train samples come from tlle mle database w11ich is summarized也e by modem Chinese graIIlIIlar 1ibrary． The join weight Value and threshold Value matriX of neural network afkr training fo肌t11e knowledge datab蠲e of the expen system based on neural ．IlI． 北京工业大学工学硕士学位论文!!!!e!!!!