基于改进BP网络的网络论坛热点主题挖掘

摘要

针对BP网络收敛速度慢的问题,提出将改善学习速率和增加动量项结合起来的方法对BP网络进行改进,通过Java编程开发一个以改进BP网络为核心的热点主题挖掘原型系统.实验结果表明,该系统可以快速并准确地获取符合现实舆情的热点主题分类结果.研究成果可以帮助民众及时了解网络热点,帮助政府和企事业单位及时跟踪民意并做出相应决策.%To solve the problem of slow convergence speed of BP network, a method is proposed to improve BP network by combining ameliorating learning rate and adding momentum factor. This paper developed a prototype system with improved BP network as the core of hot topics mining by Java programming language. Experimental results show that the system could acquire the classification results of hot topics according with the real public opinions quickly and exactly. The research result can help inform residents with instant hot spots as well as track public opinions so as to respond accordingly for governments and public enterprises.