基于贝叶斯后验的网络舆论三角模糊数型危机识别  

Triangle Fuzzy Number Crisis Identification of Network Public Opinion Based on Bayesian Posterior

作　　者：魏玲[1] 郭新朋[1] Wei Ling, Guo Xinpeng(School of Management, Harbin University of Science and Technology, Harbin 150000, Chin)

机构地区：[1]哈尔滨理工大学管理学院

出　　处：《统计与决策》2018年第6期25-29,共5页Statistics & Decision

基　　金：国家自然科学基金资助项目(71272191);黑龙江省哲学社会科学研究规划项目(16GLD02)

摘　　要：为了解决专家经验模糊化问题,提高网络舆论危机识别效率,文章提出一种基于贝叶斯后验的网络舆论三角模糊数型危机识别方法。应用贝叶斯网络后验概率根据专家经验初步处理,再进行三角模糊数均值化、去模糊化和归一化改进处理,计算出各个变量的危险指数,并依照危险等级确定各变量的危险级别,实现危机识别的高效性与准确性。基于“魏则西”事件的网络舆论危机识别的实验结果表明,该方法能够提高影响因素危机识别的准确性。In order to solve the problem of expert experience fuzziness and improve the efficiency of network public opinion crisis identification, this paper proposes a network fuzzy triangle number crisis identification method based on Bayesian posterior. According to the posterior probability of Bayesian network, the paper uses the expert experience to make a preliminary dispose, and then to average triangular fuzzy number, perform defuzzification and normalization processing, and figure out the risk index of each variable. Finally the paper relies on the level of danger to determine the risk level of each variable and achieve the efficiency and accuracy of crisis identification. The result of the experiment on the network consensus crisis identification of ＂Wei Zexi＂ inci- dent shows that the proposed method improves the accuracy of the crisis identification of influencing factors.

关 键 词：网络舆论 贝叶斯网络 三角模糊数 危机识别 

分 类 号：G202[文化科学—传播学] TB114[理学—概率论与数理统计]