Intelligent modeling and simulation of online public opinion for major accidents based on proactive safety

doi:10.16265/j.cnki.issn1003-3033.2024.01.0741

Abstract

Abstract:

In case of major security accidents, information disseminated wantonly and difficult to discern in terms of authenticity can easily cause negative social sentiments. This poses several issues to emergency rescue. Sentiment analysis, topic calculation, and the SEIR model were used to investigate online opinion simulation and control strategies for major security accidents. The CDBN-TCN-CRF sentiment analysis model was proposed by coupling the Convolutional Deep Belief Networks (CDBN), temporal convolutional networks (TCN), and conditional random fields (CRF). Then, the T-distributed Wasserstein autoencoder (TWAE) topic computation model was used to discern sentiment polarity, topic categories, race sentiment trajectory, and public attention focal points within the network discourse. Furthermore, the proposed SEIR model was used to predict online public opinion tendency and analyze the dissemination dynamics and their influencing factors. The results indicate that the coupling of CDBN-TCN-CRF sentiment analysis, TWAE topic computation, and SEIR model has a better prediction performance on network discourse analysis and trend analysis.

Key words: proactive safety, major accident, online public opinion analysis, susceptible-exposed-infectious-removed (SEIR)model, sentiment analysis

CLC Number:

X928.03事故预防与预测

CHEN Xin, XIE Kefan. Intelligent modeling and simulation of online public opinion for major accidents based on proactive safety[J]. China Safety Science Journal, 2024, 34(1): 53-61.

Figures/Tables 9

Fig.1

Fig.2

Fig.3

Tab.1

Tab.2

Fig.4

Fig.5

Fig.6

Fig.7

References 15

[1]	WANG Cong, WANG Zhenpo, ZHANG Zhiqiang, et al. Integrated post-impact planning and active safety control for autonomous vehicles[J]. IEEE Transactions on Intelligent Vehicles, 2023, 8(3): 2062-2076. doi: 10.1109/TIV.2023.3236150
[2]	ALDIANSYAH M R, SASONGKO P S. Twitter sentiment analysis about public opinion on 4G smartfren network services using convolutional neural network[C]. 3^rd International Conference on Informatics and Computational Sciences, 2019: 1-6.
[3]	刘定一, 沈阳阳, 詹天明, 等. 融合微博热点分析和LSTM模型的网络舆情预测方法[J]. 江苏大学学报, 2021, 42(5): 546-553.
	LIU Dingyi, SHEN Yangyang, ZHAN Tianming, et al. Network public opinion forecasting method fusing microblog hotspot analysis and LSTM model[J]. Journal of Jiangsu University, 2021, 42(5): 546-553.
[4]	YAN Chun, LIU Jiahui, LIU Wei, et al. Research on public opinion sentiment classification based on attention parallel dual-channel deep learning hybrid model[J]. Engineering Applications of Artificial Intelligence, 2022,116:DOI:10.1016/j.engappai.2022.105448.
[5]	ZHANG Linghan. Design of early warning mechanism of network public opinion using deep learning[C]. 2022 IEEE 4^th Eurasia Conference on IOT, Communication and Engineering, 2022: 250-252.
[6]	ZHANG Baosheng, HUANG He. Research on evolution and control of network public opinion based on SEIR model with martial law mechanism[C]. 2020 International Conference on Information Science and Education, 2020: 272-276.
[7]	JIANG Changbin, ZHANG Yaoyuan, WANG Hu, et al. Study on coupled social network public opinion communication based on improved SEIR[C]. 2020 IEEE International Conference on Big Data and Cloud Computing, 2020: 1495-1500.
[8]	RAFFAELE D, PATRICIA D, GIUSEPPE G, et al. A modified SEIR model: stiffness analysis and application to the diffusion of fake news[C]. International Conference on Computational Science and Its Applications, 2022: 90-103.
[9]	GENG Lixiao, YANG Shuran, WANG Ke, et al. Modeling public opinion dissemination in a multilayer network with SEIR model based on real social networks[J]. Engineering Applications of Artificial Intelligence, 2023,125:DOI:10.1016/j.engappai.2023.106719.
[10]	LEE H, GROSSE R, RANGANATH R, et al. Unsupervised learning of hierarchical representations with convolutional deep belief networks[J]. Communications of the ACM, 2011, 54(10): 95-103.
[11]	LEA C, VIDAL R, REITER A, et al. Temporal convolutional networks: a unified approach to action segmentation[C]. European Conference on Computer Vision, 2016: 47-54.
[12]	TOLSTIKHIN I, BOUSQUET O, GELLY S, et al. Wasserstein auto-encoders[C]. 6^th International Conference on Learning Representations(ICLR 2018), 2018: 1-16.
[13]	方匡南, 任蕊, 朱建平, 等. 基于动态SEIR模型的传染性疾病预测和政策评估[J]. 管理科学学报, 2022, 25(10): 114-126.
	FANG Kuangnan, REN Rui, ZHU Jianping, et al. Prediction and policy assessment of infectious diseases transmission using dynamic SEIR model[J]. Journal of Management Sciences in China, 2022, 25(10): 114-126.
[14]	刘丽华. 自然灾害微博舆情的社会计算模型建构研究[D]. 武汉: 武汉大学, 2018.
	LIU Lihua. Research on social computing model construction of natural disasters weibo public opinion based on sentiment analysis and topic modeling[D]. Wuhan: Wuhan University, 2018.
[15]	竺佳芳. 顾及时空主题特征的分布式灾害舆情信息获取与分析方法研究[D]. 杭州: 浙江大学, 2019.
	ZHU Jiafang. Distributed acquisition and analysis methods of disaster-related public opinions based on spatiotemporal and theme features[D]. Zhejiang: Zhejiang University, 2019.

主题	关键词
主题1	天津	爆炸	天津港	炸药	危险品	安全	事故
主题2	遇难	火灾	瑞海	消防	公司	能量	消防员
主题3	天津	爆炸	火灾	响应	网友	机场	回家
主题4	天津	交通	爆炸	学校	照料	停课	关注
主题5	天津港	酒店	服务	安全	滞留	物资	部署
主题6	逆行	爆炸区	消防	居民区	仓库	危险品	滨海
主题7	安全	居家	爆炸	火灾	烧伤	消防	浓烟
主题8	天津港	睡觉	居民	爆炸	爆发	周边	火灾
主题9	爆炸	滞留	婴儿	公安	天津	帮助	救援
主题10	缺乏	事故	重大	安全	物资	消防	厂区
主题11	天津	爆炸	突发	滨海	事故	防火	浓烟
主题12	滨海	爆炸	应急	消防	报道	中国	媒体
主题13	危害	爆炸	火灾	爆炸区	食物	居家	消防
主题14	天津	火灾	天津港	瘫痪	交通	直播	记录
主题15	天津港	救援	伤员	烫伤	防护	维修	躲避

分类模型	精确率(Precidion)	召回率(Recall)	F₁值
LDA+SVM	0.74	0.79	0.78
VAE+SVM	0.82	0.83	0.85
WAE+SVM	0.89	0.84	0.87
TWAE+SVM	0.92	0.94	0.95