摘要
[目的/意义]旨在对网络舆论进行正确积极的引导,提高舆情监测效果。[方法/过程]构建基于LDA的公共安全事件微博舆情主题知识图谱过程模型,利用困惑度指标确定微博舆情的最优主题数;利用Wasserstein距离对微博舆情的主题进行相似度度量,并将Wasserstein距离作为边权重构建主题知识图谱。以新浪微博“山西农村饭店坍塌事件”为例,进行舆情主题和意见领袖的分析讨论。[结果/结论]基于LDA的微博舆情主题知识图谱不仅可以有效识别出主题,还可以准确定位每个主题下的意见领袖,对舆情监管部门识别某一话题中的不同主题和意见领袖有一定的指导意义。
[Purpose/significance]The paper is to conduct correct and positive guidance to the online public opinion and improve the effect of public opinion monitoring.[Method/process]The paper constructs process model of thematic knowledge graph of Weibo public opinion of public security events based on the Latent Dirichlet Allocation(LDA)and the optimal number of topics of Weibo public opinion is determined by using the degree of confusion evaluation index.Wasserstein distance is used to measure the theme similarity of Weibo public opinion and is used as edge weight to construct the thematic knowledge graph.Taking Shanxi rural hotel collapse event on Sina Weibo for example it discusses the topic identification and user opinion leader identification.[Result/conclusion]The thematic knowledge graph of public opinion based on LDA can not only effectively identify topics but also accurately locate opinion leaders under each topic which has some guiding significance for the public opinion supervision department to identify different themes and opinion leaders in a topic.
作者
韩佳伶
余天池
Han Jialing;Yu Tianchi(School of Management Science and Information Engineering Jilin University of Finance and Economics,Changchun Jilin 130117)
出处
《情报探索》
2021年第9期85-93,共9页
Information Research