摘要
[目的/意义]随着信息技术的高速发展,网络舆情已经成为公共决策的重要参考,正确分析和引导突发舆情事件,重点在于舆情的情报感知,为突发舆情事件的应急管理提供情报支持。[方法/过程]文章剖析突发舆情事件的发展态势,从突发事件舆情中提取情报,分类为诉求型情报和支援型情报,建立全新的突发舆情事件情报语料库,基于朴素贝叶斯、支持向量机、K-近邻三种算法,构建突发舆情事件情报感知模型,搜集大量“新型冠状病毒肺炎疫情”的文本数据进行实证分析。[结果/结论]突发舆情事件情报感知模型中三类分类算法的准确率、召回率、F1值较高,结果验证了突发舆情事件情报语料库的科学性,以及突发舆情事件情报感知模型的合理性,其中运用支持向量机算法感知结果为:诉求型情报的F1值为0.873,支援型情报的F1值为0.894,相比其他算法的感知评价指数,支持向量机算法的感知结果更加准确。
[Purpose/significance]With the rapid development of information technology,internet public opinion has been a critical reference to public decisions.Analyzing and leading public opinion emergencies correctly lays on intelligence perception of public opinion,which will provide intelligence support for emergency management of public opinion emergencies.[Method/process]The paper analyzed the development trend of public opinion emergencies and extracted intelligence including appealing intelligence and supporting intelligence from public opinion of emergencies and constructed brand-new public opinion emergency information corpus.Based on Naive Bayes,support vector machine and K-nearest neighbor algorithms,the paper built intelligence perception model of public opinion emergencies and collected large amount of text data of“COVID-19”to conduct empirical analysis.[Result/conclusion]The accuracy,recall and F1 value of three classification algorithm in intelligence perception model of public opinion emergencies were higher and the result showed that public opinion emergency information corpus was scientific and intelligence perception model of public opinion emergencies was valid.The perception result obtained by support vector machine was:F1 value of appealing intelligence was 0.873 and F1 value of supporting intelligence was 0.894.Compared with perception evaluation index of other algorithms,the perception result of support vector machine algorithm was more accurate.
出处
《情报理论与实践》
CSSCI
北大核心
2021年第10期119-128,共10页
Information Studies:Theory & Application
基金
国家社会科学基金项目“基于大数据的网络舆情全息建模与决策情报感知研究”的成果之一,项目编号:20BXW074。
关键词
突发事件
舆情事件
情报感知模型
情报语料库
emergencies
public opinion event
information perceptual model
information corpus
