摘要
[研究目的]系统探讨微博舆情事件的演化特征,以提出针对性的对策建议,避免网络舆情扩散所可能引发的不利影响。[研究方法]为实现该目的,提出了基于CNN-BiLSTM-Attention的微博舆情多维特征演化分析框架,以深入剖析网络舆情的形成机制,进而优化对网络舆情的应对和处理策略。[研究结论]根据选取的事件从新浪微博获取数据,基于TF-IDF模型和K-Means聚类算法对微博舆情事件进行了维度划分,通过组合模型CNN-BiLSTM-Attention进行情感分类,并验证其准确性。最后,根据维度划分和情感分类的结果,结合舆情生命周期理论,从舆情热度、主题和情感三个方面研究了微博舆情事件的演化情况,并从生命周期和主题情感两方面得出网络舆情应对策略。
[Research purpose]This study aims to systematically explore the evolution characteristics of public opinion events in microblog,so as to put forward targeted countermeasures and suggestions to avoid the possible adverse effects caused by the diffusion of network public opinion.[Research method]A multi-dimensional feature evolution analysis framework based on CNN-BiLSTM-Attention is proposed to deeply analyze the formation mechanism of online public opinion,and then optimize the response and processing strategy of online public opinion.[Research conclusion]In this study,data are obtained from Sina Weibo according to the selected events,and the microblog public opinion events are divided into dimensions based on the TF-IDF model and K-Means clustering algorithm.The combined model CNN-BiLSTM-Attention is used for sentiment classification,and its accuracy is verified.Finally,according to the results of dimension division and sentiment classification,combined with the life cycle theory of public opinion,the evolution of public opinion events is studied from three aspects of public opinion heat,topic and sentiment,and the coping strategies of network public opinion are obtained from two aspects of life cycle and topic sentiment.
作者
王虎
吴浩伟
江长斌
Wang Hu;Wu Haowei;Jiang Changbin(School of Management,Wuhan University of Technology,Wuhan 430070)
出处
《情报杂志》
CSSCI
北大核心
2024年第11期144-151,128,共9页
Journal of Intelligence
基金
国家社会科学基金项目“大数据视域下‘隐性’政治舆情演化规律及治理路径研究”(编号:19BSH013)研究成果。
关键词
网络舆情
舆情演化
情感分析
神经网络
聚类算法
文本分析
微博
public opinion
evolution of public opinion
sentiment analysis
neural network
clustering algorithm
text analysis
Weibo
