Research on BN of network public opinion crisis risk caused by short videos of rainstorm disaster

doi:10.16265/j.cnki.issn1003-3033.2022.11.0077

Abstract

Abstract:

In order to reduce the risk of network public opinion crisis caused by short videos, taking the "7·20" Henan rainstorm incident as an example, this paper constructed a prediction index system for the risk of public opinion crisis based on short-form online videos from five aspects:the degree of disaster, response behavior, video attributes, public opinion sentiment tendency and public opinion spread diffusion of the unexpected incident. BN was used to build a prediction model to discuss the accuracy of public opinion forecast. The validation results indicate that the model can effectively predict the risk level of network public opinion crisis, and the public's response behavior in the face of emergencies has an important influence on the risk level of public opinion crisis, sentiment tendency and the spread of public opinion. When the disaster is severe and the public's behavior is relatively negative, the main reason for the outbreak of public opinion is the public's emotional tendency and the spread of public opinion. Short videos with a high number of comments are more likely to trigger network public opinion crises than retweets and likes.

Key words: rainstorm disaster, online short video, public opinion crisis risk, Bayesian network (BN), public opinion forecast

LYU Wei, ZHOU Wennan, CHEN Wentao, HAN Yefan, FANG Zhiming. Research on BN of network public opinion crisis risk caused by short videos of rainstorm disaster[J]. China Safety Science Journal, 2022, 32(11): 192-199.

Figures/Tables 8

Tab.1

Fig.1

Tab.2

BN edge probabilities for predicting risk level of online public opinion crisis

序号	S₀	S₁₁	S₁₂	$…$	S₅₄
视频1	低	高	高	$…$	高
视频2	中	高	高	$…$	高
视频3	中	低	高	$…$	高
$︙$	$︙$	$︙$	$︙$		$︙$
视频224	中	中	高	$…$	低

Tab.2

Tab.3

Fig.2

Tab.4

Tab.5

Fig.3

References 15

[1]	毕翔, 唐存琛, 肖俏. 短视频社交媒介舆情监测与危机防范机制研究[J]. 情报理论与实践, 2019, 42(10):102-108.
	BI Xiang, TANG Cunchen, XIAO Qiao. Research on public opinion monitoring and crisis prevention mechanism of short video[J]. Information Studies:Theory & Application, 2019, 42(10):102-108.
[2]	刘继, 武梦娇. 基于贝叶斯网络的网络舆情态势评估分析:以“新冠肺炎疫情”事件为例[J]. 情报杂志, 2021, 40(3): 187-192,103.
	LIU Ji, WU Mengjiao. Assessment and analysis of network public opinion situation based on Bayesian network:COVID-19 as an example[J]. Journal of Intelligence, 2021, 40(3):187-192,103.
[3]	XING Ziyao, SU Xiaohui, LIU Junming, et al. Spatiotemporal change analysis of earthquake emergency information based on microblog data:a case study of the "8.8" Jiuzhaigou earthquake[J]. ISPRS International Journal of Geo-Information, 2019, 8(8):359-359. doi: 10.3390/ijgi8080359
[4]	周义棋, 田向亮, 钟茂华. 基于微博网络爬虫的巴黎圣母院大火舆情分析[J]. 武汉理工大学学报:信息与管理工程版, 2019, 41(5):461-466.
	ZHOU Yiqi, TIAN Xiangliang, ZHONG Maohua. The network public opinion analysis of Notre-Dame fire based on weibo and web crawler[J]. Journal of Wuhan University of Technology:Information & Management Engineering, 2019, 41(5):461-466.
[5]	陈璐. 论短视频新闻在突发事件中的舆论导向作用:以“重庆万州公交车坠江”事件为例[J]. 西部广播电视, 2019(2): 26-27.
[6]	王微, 王晰巍, 娄正卿, 等. 信息生态视角下移动短视频UGC网络舆情传播行为影响因素研究[J]. 情报理论与实践, 2020, 43(3):24-30.
	WANG Wei, WANG Xiwei, LOU Zhengqing, et al. Research on the influencing factors of network public opinion dissemination behavior in User-Generated Content of mobile video clips from the perspective of information ecology[J]. Information Studies:Theory & Application, 2020, 43(3):24-30.
[7]	王英杰, 胡漠, 张津赫, 等. 信息疫情下短视频网络舆情预警指标体系构建研究[J]. 情报科学, 2021, 39(11):38-44.
	WANG Yingjie, HU Mo, ZHANG Jinhe, et al. Early warning index system of short video network public opinion under infodemic[J]. Information Science, 2021, 39(11):38-44.
[8]	MAO Yan, LI Zuning, LI Yongjian, et al. Emotion-based diversity crowd behavior simulation in public emergency[J]. The Visual Computer, 2019, 35(12):1725-1739. doi: 10.1007/s00371-018-1568-9
[9]	SHIWAKOTI N, TAY R, STASINOPOULOS P, et al. Passengers' perceived ability to get out safely from an underground train station in an emergency situation[J]. Cognition Technology and Work, 2018, 20(3):367-375. doi: 10.1007/s10111-018-0473-3
[10]	李琰, 毛钰, 张燕. 建筑工人应急行为影响机制[J]. 中国安全科学学报, 2021, 31(10):1-7. doi: 10.16265/j.cnki.issn1003-3033.2021.10.001
	LI Yan, MAO Yu, ZHANG Yan. Influence mechanism of construction workers' emergency behavior[J]. China Safety ScienceJournal, 2021, 31(10):1-7.
[11]	杨静, 朱莉萨, 朱镇远, 等. 基于贝叶斯网络的网络舆情案例匹配模型[J]. 现代情报, 2019, 39(10):94-101. doi: 10.3969/j.issn.1008-0821.2019.10.011
	YANG Jing, ZHU Lisa, ZHU Zhenyuan, et al. A model of case matching for internet public sentiment based on belief network[J]. Modern Information, 2019, 39(10):94-101.
[12]	周昕, 邱长波, 李瑞. 基于贝叶斯网络的网络舆情危机节点诊断研究[J]. 现代情报, 2018, 38(11):59-65. doi: 10.3969/j.issn.1008-0821.2018.11.011
	ZHOU Xin, QIU Changbo, LI Rui. Study on crisis diagnosis of network public opinion based on Bayesian network[J]. Modern Information, 2018, 38(11):59-65.
[13]	付业勤. 旅游危机事件网络舆情研究:构成、机理与管控[D]. 泉州: 华侨大学, 2014.
	FU Yeqin. Research on the network opinion of tourism crisis:compositions, mechanism, and management & control[D]. Quanzhou: Huaqiao University, 2014.
[14]	闵晨, 王国华, 陈强. 视频在热点事件议程设置与舆论引导中的作用研究[J]. 情报杂志, 2018, 37(7):80-87.
	MIN Chen, WANG Guohua, CHEN Qiang. The video's role in agenda setting and public opinion guidance of hot issues[J]. Journal of Intelligence, 2018, 37(7):80-87.
[15]	PEARL B J. Probabilistic reasoning in intelligent systems:network of plausible inference[M]. San Francisco: Morgan Kaufmann Publishers, Inc, 1988:45-50.

目标变量	一级指标	二级指标	描述	取值
网络舆情危机风险S₀	灾情程度S₁	水位S₁₁	膝盖以下、膝盖以上臀部以下、臀部以上	低、中、高
	灾情程度S₁	受灾人数S₁₂	(0,5)、[5,20]、(20,+∞)	低、中、高
	应对行为S₂	被困者积极自救S₂₁	隶属、不隶属	是、否
		参与营救他人S₂₂	隶属、不隶属	是、否
		捐赠物资S₂₃	隶属、不隶属	是、否
		感恩救援人员S₂₄	隶属、不隶属	是、否
		发布救援信息S₂₅	隶属、不隶属	是、否
		服从转移安置S₂₆	隶属、不隶属	是、否
		非理性行为S₂₇	隶属、不隶属	是、否
	视频属性S₃	互助救助类S₃₁	隶属、不隶属	是、否
		寻求帮助类S₃₂	隶属、不隶属	是、否
		受灾现场类S₃₃	隶属、不隶属	是、否
		非理性行为类S₃₄	隶属、不隶属	是、否
	舆情情感倾向S₄	微博原文情感倾向S₄₁	情感值大于0、情感值小于等于0	积极、消极
	舆情情感倾向S₄	微博评论情感倾向S₄₂	积极情感占比大于50%、积极情感占比小于等于50%	积极、消极
	舆情传播扩散度S₅	信息发布主体S₅₁	官方媒体、自媒体	官媒、自媒体
		微博评论总数S₅₂	(0,300]、(300,700]、(700,1000]	低、中、高
		微博转发总数S₅₃	(0,300]、(300,700]、(700,1000]	低、中、高
		微博点赞总数S₅₄	(0,300]、(300,700]、(700,1000]	低、中、高

节点	排名前5 的节点	相关度指标	占比	相关度指标变异系数
S₀	S₂	0.187 85	15.3	0.048 490 9
	S₅	0.085 05	6.95	0.023 714 9
	S₄₂	0.044 08	3.6	0.003 486 3
	S₁	0.041 63	3.4	0.005 809 1
	S₄₁	0.030 14	2.46	0.002 423 1
S₄	S₂	0.347 34	35.5	0.109 097 7
	S₁	0.015 74	1.61	0.005 196 1
	S₄₂	0.015 33	1.57	0.005 237 8
	S₄₁	0.011 13	1.14	0.003 805 0
	S₂₅	0.007 67	0.783	0.002 513 3
S₅	S₀	0.085 05	5.38	0.012 373 7
	S₅₂	0.031 15	1.97	0.004 661 0
	S₂	0.014 39	0.911	0.002 248 7
	S₅₃	0.014 27	0.904	0.001 956 7
	S₅₄	0.007 32	0.463	0.001 084 9

节点	状态	后验推理1			后验推理2
节点	状态	预测模型 S₀=低 (p=56.1)	后验概率1 S₀=高 (p=100)	变动趋势	后验概率1 S₀=高 (p=100)	后验概率2 S₀=高、S₁=高、S₂=消极 (p=100)	变动趋势
S₁	低	22.3	30.7	37.67	30.7	—	—
	中	29.0	52.2	80.00	52.2	—	—
	高	48.6	17.0	-65.02	17.0	—	—
S₂	消极	37.3	21.0	-43.70	21.0	—	—
	中立	29.5	29.8	1.02	29.8	—	—
	积极	33.1	49.3	48.94	49.3	—	—
S₃	消极	26.7	35.1	31.46	35.1	38.3	9.12
	中立	35.5	31.8	-10.42	31.8	30.9	-2.83
	积极	37.7	33.1	-12.20	33.1	30.8	-6.95
S₄	消极	58.6	38.7	-33.96	38.7	100	158.40
S₄	积极	41.4	61.3	48.07	61.3	0	-100
S₅	低	36.2	36.2	0.00	36.2	0	-100
	中	29.5	23.4	-20.68	23.4	18.9	-23.81
	高	34.3	40.4	17.78	40.4	81.1	100.74