突发事件网络舆情反转的PCA-LDA-LSSVM预测模型  

PCA-LDA-LSSVM model for predicting network public opinion reversal of emergencies

作　　者：赵琳琳 温国锋[1] 杨永清[1] ZHAO Linlin;WEN Guofeng;YANG Yongqing(School of Management Science and Engineering,Shandong Technology and Business University,Yantai Shandong 264005,China)

机构地区：[1]山东工商学院管理科学与工程学院,山东烟台264005

出　　处：《中国安全生产科学技术》2023年第8期186-190,共5页Journal of Safety Science and Technology

基　　金：山东省社科规划研究项目(20DGLJO14);国家社会科学基金项目(20BSH151)。

摘　　要：为有效引导与控制突发事件网络舆情,建立科学的预警机制,提出突发事件网络舆情反转的主成分分析(PCA)-线性判别分析(LDA)-最小二乘支持向量机(LSSVM)预测模型。利用PCA提取具有相关性的影响因素主成分,利用LDA方法分析相互独立的影响因素和主成分对突发事件网络舆情反转的影响,并将LDA分析后的影响因素作为LSSVM的输入向量,预测突发事件网络舆情反转,通过选取33组突发事件网络舆情数据进行试验研究。研究结果表明:影响因素重要性由大到小依次为网民情感正倾向、网民情感负倾向、舆情事件性质、舆情传播形式、舆情首发主体权威性;当网民情感正倾向明显减少、网民情感负倾向明显增加时,应采取措施引导舆情发展。In order to guide and control the network public opinion in emergencies effectively,and establish the scientific early warning mechanism,a model for predicting the network public opinion reversal of emergencies based on PCA-LDA-LSSVM was built.Firstly,the PCA was used to extract the principal components of influencing factors with correlation.Then,the LDA was used to analyze the influence of the unrelated influencing factors and principal components on the network public opinion reversal of emergency,and the influencing factors after LDA analysis were taken as the input vector of LSSVM to predict the network public opinion reversal of emergency.Finally,according to 33 sets of emergency network public opinion data,a series of comparative experiments were designed and carried out.The results show that the order of importance of influencing factors is the positive emotional tendency of netizens,the negative emotional tendency of netizens,the nature of public opinion events,the form of public opinion dissemination,and the authority of the first subject of public opinion.The measures should be taken to guide the development of public opinion when the positive emotional tendency of netizens is reduced and the negative emotional tendency of netizens is increased significantly.

关 键 词：突发事件 网络舆情 主成分分析(PCA) 线性判别分析(LDA) 最小二乘支持向量机(LSSVM) 

分 类 号：X913.4[环境科学与工程—安全科学]