基于动态主题情感分析的虚假新闻网络舆情研究  

Fake News Network Public Opinion Based on Dynamic Theme Sentiment Analysis

作　　者：于凯 刘迪 YU Kai;LIU Di(School of Public Administration,Xinjiang University of Finance&Economics,Urumqi 830001,China;School of Information Management,Xinjiang University of Finance&Economics,Urumqi 830001,China)

机构地区：[1]新疆财经大学公共管理学院,新疆乌鲁木齐830001 [2]新疆财经大学信息管理学院,新疆乌鲁木齐830001

出　　处：《长春师范大学学报》2021年第4期32-37,共6页Journal of Changchun Normal University

基　　金：新疆维吾尔自治区自然科学基金面上项目“基于多层网络模型的信息传播源头定位研究”(2019D01A22);新疆维吾尔自治区天山青年计划优秀青年科技人才项目(2018Q027)。

摘　　要：目前,关于虚假新闻的研究主要从辨别、影响和治理等角度出发,对虚假新闻中的公众情感分析多为定性研究,缺少客观定量地表示公众情感变化,为舆情治理提供有效的决策支持方面较薄弱。鉴于此,本文提出一种基于动态主题情感模型的分析方法,运用改进的TF-IDF和SO-PMI相结合的方法构建专属情感词典,对比真实新闻与虚假新闻的主题与情感变化,运用主成分分析法构建多主体虚假新闻网络舆情治理指标体系和模型,最后进行实证分析。通过实证研究表明,该方法可以有效地分析公共安全事件中公众情感、行为等动态变化特点,可为进一步制定治理虚假新闻网络舆情的措施提供有效理论依据。At present,the research on fake news is mainly from the perspective of identification,influence and governance.The analysis of public emotion in fake news is mostly qualitative research,which can not objectively and quantitatively express the changes of public emotion and provide effective decision support for public opinion governance.In view of this,the paper proposes an analysis method based on the dynamic theme emotion model,uses the improved TF-IDF and SO-PMI method to build an exclusive emotion dictionary,compares the theme and emotion changes of real news and fake news,uses the principal component analysis method to build a multi-agent fake news network public opinion governance index system and model,and finally makes an empirical analysis.The experimental results show that this method can effectively analyze the dynamic characteristics of public emotion and behavior in public security incidents,and provide an effective basis for further formulating measures to control the fake news network public opinion.

关 键 词：虚假新闻 TF-IDF SO-PMI 主成分分析 动态主题情感分析 

分 类 号：G201[文化科学—传播学]