微博网络舆情建模及发展趋势预测研究

发布时间：2018-03-11 12:30

本文选题：微博网络舆情　切入点：RBF神经网络　出处：《兰州交通大学》2017年硕士论文　论文类型：学位论文

【摘要】：互联网技术日新月异,微博社交平台的出现,改变了人们的生活方式,对传统的信息传播媒体,产生了巨大的冲击。微博平台上的舆情作为社会舆情的延伸,发挥着举足轻重的作用。由于在微博上发布消息具有及时性、任意性,并且信息传播很迅速,传播的范围广泛,造成微博上的信息既有真实的,也有虚假的。一些人将微博当作传播谣言的工具,对社会生态环境、人民的正常生活造成很大的负面影响。因此,研究微博平台上舆情的趋势具有重要的现实意义。在用算法实现预测微博平台上热点话题的趋势时,需要采集相关的舆情数据进行实验。本文面向新浪微博获取数据,从获得的数据中,聚类发现热点话题,将选取的话题表示为单位时间的数据序列。步骤如下:(1)选取新浪微博为研究对象,采集相关的微博数据;(2)对获取的微博文本进行预处理,采用聚类算法得到热点舆情话题;(3)根据微博的特点,提取能表示微博舆情趋势的特征,构成舆情预测实验数据。目前,RBF神经网络是应用最广泛和成功的神经网络之一。它结构简单、可塑性强,并且具有全局最优逼近能力与良好的推广能力。但是,在进行舆情预测的时候,RBF神经网络预测结果的准确度与径向基函数的中心,方差(宽度)以及隐含层到输出层的权值之间有着紧密的联系。因此,传统的RBF神经网络中,参数的选择限制了它在网络舆情预测中的应用。本文对RBF神经网络的参数进行优化,并采用优化过的模型进行预测实验。通过对3个模型的实验结果进行对比发现,用改进的引力搜索算法优化的RBF神经网络模型对微博舆情的预测精度及速度最好。因此,本文的算法模型可以更好的预测网络话题的发展趋势,预测结果有利于政府对舆情信息的监控和引导,也有利于社会的和谐稳定。
[Abstract]:With the rapid development of Internet technology and the emergence of Weibo's social platform, it has changed people's way of life and had a huge impact on the traditional information dissemination media. Because of the timeliness and arbitrariness of publishing information on Weibo, and the rapid spread of information and the wide range of dissemination, the information on Weibo has both true information. There are also false ones. Some people regard Weibo as a tool for spreading rumors, which has a great negative impact on the social ecological environment and the normal life of the people. Therefore, It is of great practical significance to study the trend of public opinion on Weibo platform. When we use algorithms to predict the trend of hot topics on Weibo platform, we need to collect relevant public opinion data for experiment. From the data obtained, the hot topics were found by clustering, and the selected topics were expressed as a data sequence per unit time. The steps are as follows: 1. Select Sina Weibo as the research object, collect the relevant Weibo data. Using clustering algorithm to get hot topic of public opinion.) according to Weibo's characteristics, we can extract the features that can represent the trend of public opinion of Weibo. At present, RBF neural network is one of the most widely used and successful neural networks. It has simple structure, strong plasticity, global optimal approximation ability and good generalization ability. In the prediction of public opinion, there is a close relationship between the accuracy of the prediction results and the center of the radial basis function, the variance (width) and the weights from the hidden layer to the output layer. Therefore, in the traditional RBF neural network, the accuracy of the prediction results is closely related to the center of the radial basis function, the width and the weight of the hidden layer to the output layer. The selection of parameters limits its application in network public opinion prediction. In this paper, the parameters of RBF neural network are optimized, and the optimized model is used for prediction experiment. The RBF neural network model optimized by the improved gravitational search algorithm has the best prediction accuracy and speed for Weibo's public opinion. Therefore, the algorithm model in this paper can better predict the development trend of the network topic. The prediction results are beneficial to the government's monitoring and guidance of public opinion information, and to social harmony and stability.
【学位授予单位】：兰州交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP183;TP391.1

【参考文献】