Introduction

With the continuous modernization of democratic politics and the rapid development of science and technology, people are paying more attention to natural and social phenomena and expressing their views and opinions on major social media platforms. When a topic has received more attention, frequent searches form online public opinion (Su and Wang, 2024), which fluctuates with the changing popularity of discussions. In general, sudden reversal events can increase the risk of public opinion transmission (Wu and Liu, 2023). On the one hand, public engagement in Internet topic discussions prompts the government to intensify its focus on and facilitate the resolution of incidents. On the other hand, due to information asymmetry, online public opinion often trends negatively, necessitating prompt government intervention to sustain a healthy online discourse environment. Contemporary society is now in a high-risk period due to the rise of sudden natural disasters, such as earthquakes. This situation has prompted the public to express their views and emotions on various social media platforms (Gao et al. 2021). In addition to comments and likes of the post, the forwarding directly increased the dissemination of public opinion (Lv, 2022). The spread of online public opinion contains various entities, including ordinary citizens, the government, the media, etc. These diverse actors intertwine and interact, complicating online public opinion (Hou et al. 2022). Related stakeholders with distinct viewpoints engage in vigorous discussions on a topic, attempting to persuade one another, reflecting a trend of evolutionary game theory (Qiu et al. 2023).

China has frequent earthquakes, which have caused more online public opinion. When an earthquake occurs, it captures the attention of the whole community. Not only do major media outlets report on it, but many social media users also express their concerns through platforms such as Weibo and Zhihu, collectively creating a comprehensive online network (Lian et al. 2017). When public opinion is constantly going towards orgasm, fear and anxiety will spread in society. In addition, individuals in disaster-stricken areas may resort to passive or aggressive actions to express their frustration with the government’s disaster relief efforts. This discontent can subsequently affect the efficiency of earthquake relief operations (Huang et al. 2019). Studying the mechanisms underlying the evolution of earthquake-related online public opinion is beneficial for the government in improving public opinion management and prevention strategies. Additionally, it can facilitate the efficient progression of earthquake relief efforts. However, online public opinion regarding earthquakes is a complex phenomenon. It emerges from the earthquake, prompting netizens, media, and government entities to engage in online discussions. If any of these variables change, the entire landscape of online public opinion related to earthquakes shifts rather than merely following a chronological progression. Nevertheless, the government needs to emphasize the management of earthquakes online public opinion. In the aftermath of an earthquake, the optimal disaster relief plan is still unclear. Moreover, no dedicated early warning system monitors earthquake online public opinion. In most cases, public opinion is managed reactively after it emerges, which leaves the government in a relatively passive position (Jiang et al. 2017). Nevertheless, earthquake online public opinion exists as an objective reality, posing challenges and difficulties for public opinion management in various countries. Especially when the government cannot quickly cope with the earthquake online public opinion, online public opinion will develop in a worse direction. Therefore, studying the mechanisms underlying the evolution of online public opinion on earthquakes is essential. Only after understanding its cause can the government effectively control the spread and dissemination of earthquake online public opinion, thus mitigating associated risks.

At present, more research focuses on the public opinion of emergencies, and there is less research on earthquake online public opinion (Zhang et al. 2023). Earthquake online public opinion, as an online discourse form of an emergency, has both typical characteristics of earthquake events and unique attributes. Therefore, this paper focuses on public opinion of earthquakes and places them in a wide range of sudden natural disasters. (Zhang et al. 2024). This study examines public reactions after disasters and provides recommendations for managing online discourse about sudden natural events. This paper utilizes the life cycle theory, the crisis management theory, and system dynamics methodologies to build a model for the evolution of earthquake online public opinion. This model analyzes the potential variables affecting online public opinion about earthquakes from physical and social perspectives, emphasizing the evolution of public opinion. Simultaneously, the paper uses the M5.7 earthquake in Yibin, China, as a case study, drawing on earthquake-related data from Sina Weibo to investigate the impacts of earthquakes, netizens, media, and government on earthquake online public opinion. Finally, this article studies the factors affecting the evolution of online public opinion by adjusting the variables in the model. The primary objective of this work is to prevent and control the development of online public opinion about earthquakes in a poor direction.

Literature review

Currently, there is a paucity of research specifically concentrated on earthquake online public opinion (Yuan et al. 2024), with a more significant number of studies addressing emergency online public opinion (F. Yin et al. 2024a). Emergency online public opinion is deeply intertwined with the availability of online platforms. The use of social platforms is highly private, and people can participate in topic discussions on the platforms anytime and anywhere. In other words, the platform’s accessibility allows users to fully express their feelings, emotions, and opinions (Ahanin and Ismail, 2022). Topics discussed on social platforms are primarily emotional (Liu et al. 2019), and content with positive emotions is more likely to be shared than negative emotions (Weismueller et al. 2022). Earthquake online public opinion is characterized by the uncertainty of the outbreak, the wide range of subjects, the depth of topics, and the continuity of dissemination. Major media outlets report on the event when an earthquake occurs, promoting public attention. The result is that netizens express their emotions through platforms such as Weibo, Zhihu, and Baidu. Researchers have utilized an accelerated genetic algorithm to improve the BP neural network. They have also evaluated the risks associated with earthquake online public opinion and investigated its evolutionary mechanisms (Huang et al. 2019). Users across the two platforms present differing perspectives on the same earthquake and showcase a variety of thematic interests (Ruan et al. 2022). In the evolution of emergencies, online public opinion usually catalyzes the escalation and spread of emergencies and even affects the transformation of public opinion (Y. X. Zhang et al. 2021b). Consequently, some scholars have studied the structural complex network modeling of online public opinion from the perspective of complex network evolution (Wang et al. 2018). In addition, hot events are randomly selected from Weibo as a data set to study the online public opinion of emergencies from topic feature extraction, sentiment analysis, and temporal variations (Liu et al. 2020).

In the in-depth mining of online public opinion, a causal model is constructed based on system dynamics to explore the trend of online public opinion on emergencies (Gao et al. 2019). An individual’s perception of the value of information is affected by their knowledge level and the configuration of the online environment (Sohn, 2014). Opinions in public opinion incidents are often fragmented into distinct camps despite the complexity of many viewpoints. Through comprehensive analysis of the polarization phenomenon, researchers have created a public opinion polarization model that considers individual heterogeneity and dynamic consistency (Chen et al. 2019). Information dissemination and opinion evolution are often intertwined, and agents’ opinions affect their diffusion actions (Xiong et al. 2014). Based on agent-based simulations, mass media are amid noisy opinions. Opinion heterogeneity is preserved due to weak assimilation strength and errors accompanying opinion modification (Takesue, 2021). In this context of mass media influence, it is valuable to explore how the dissemination of minority opinions develops across different simulated scenarios (Alvarez-Galvez, 2016). In the comprehensive research on online public opinion, numerous scholars have collected data from relevant social media platforms and employed text, image, audio, and video mining techniques to improve the evaluation of online public opinion. A novel function extension method for short-text stream classification was proposed for short-text data based on an efficient incremental combination classification model (Li et al. 2018). Similarly, a classification method for short text streams was created using the online Biterm Topic Model (BTM) (Hu et al. 2018).

Furthermore, topic modeling has been utilized for news topic detection to examine the comparative agendas of newspapers and public interest (Pinto et al. 2019). In the context of public opinion, paying attention to the fluctuations in social emotions that underlie significant public events is essential. Scholars have proposed various methods for analyzing text opinions, including probabilistic latent semantic analysis models, graph convolutional networks, dependency parsing models, and the creation of sentiment dictionaries. Clustering machine learning algorithms are employed to conduct sentiment analysis of Twitter data (Jacob and Vijayakumar, 2021). The government must enhance its capacity to guide mainstream sentiments, as these sentiments more accurately reflect the positions and attitudes of bloggers on social media. Moreover, this approach can also be utilized to explore political preferences through social media sentiments (Ceron et al. 2014).

To carry out earthquake relief more effectively, the government should fully use the content presented by online public opinion and formulate appropriate countermeasures. The government must race against time to rescue trapped people in the face of sudden natural disasters. In sudden natural disasters, the government must act swiftly to rescue those trapped and the MCDM method has been applied to emergency response decision-making (Nassereddine et al. 2019). Government departments must swiftly respond to emergencies and conduct rescue operations. If the government fails to respond promptly or omits earthquake rescue work, it will be criticized online, directly triggering public opinion. The risk of online public opinion brings many challenges and pressures to the government and even triggers mass incidents, seriously affecting the government’s credibility and image (Y. X. Zhang et al. 2021b). Therefore, the government must quickly solve practical problems in rescue and respond to the needs of the general public. When solving online public opinion about emergencies, comprehensive methods are often used to analyze natural disasters, artificial accidents, public health, and social security to provide solutions quickly (Huang et al. 2022). Concurrently, it is essential to enhance oversight of news dissemination, considering its characteristics and challenges, particularly in eliminating misinformation and other related issues (Han, 2017). Scholars use deep learning to predict public opinion sentiment to improve the early warning of online public opinion on emergencies. The principle is to use style measurement and adversarial learning to eliminate topic bias, thereby strengthening emergency management of emergencies (Li et al. 2022). In addition, a hybrid sentiment entity recognition model has been developed to classify sentiment datasets in online public opinion using machine learning techniques (Wang et al. 2016). Many researchers focus on capturing and classifying the comment content related to trending microblog events while standardizing and integrating public opinion on emergencies (M. Zhang et al. 2021a). Furthermore, by analyzing the impact on network controllability, positive guidance technology is employed to establish a new model that facilitates external control and prevents and manages online public opinion during emergencies (Zhang et al. 2018). Online public opinion represents social conditions and public opinion to a certain extent. Text mining (Han et al. 2020) or topological evolution methods (Lian et al. 2017) are often used to mine the content of public opinion. The diffusion model method is applicable in studying the information dissemination process in social media (Li et al. 2018). The evolution of online public opinion is mainly a process of complex psychological and behavioral dynamic interactions among participants (Fan et al. 2015). To further understand the evolution of online public opinion, it is necessary to construct an evolution model based on information entropy to simulate the dynamics of public opinion (Huang et al. 2014). Thus, this study uses life cycle, crisis management theory, and system dynamics methods to construct an earthquake online public opinion evolution model. Then, this paper simulates public opinion, aiming to provide suggestions and insights for the government to manage and mitigate online public opinion.

In summary, the earthquake online public opinion research has revealed some deficiencies and areas that need further exploration. (1) Earthquake online public opinion has not received enough attention, and most studies focus on online public opinion related to general emergencies. Consequently, analyzing the underlying causes for its emergence remains rather vague. (2) The mechanisms underlying the evolution of earthquake online public opinion are not yet well established. Most studies primarily analyze the influencing variables of public opinion without exploring the simulation of earthquake online public opinion in depth. (3) Insufficient attention is paid to analyzing the roles of various stakeholders in the context of earthquake online public opinion. In different stages of online public opinion development, the roles of other stakeholders are different. However, the stage of analysis is often neglected in current research.

The potential contributions of this paper are as follows: First, the evolution model of earthquake online public opinion is constructed around four key entities: earthquake events, media, netizens, and the government. This model can consider various variables that affect the size of online public opinion and analyze its trend from a macro perspective. Second, a stage analysis method is adopted to accurately grasp the evolution process of online public opinion on earthquakes. Finally, this study creatively uses the system dynamics method and simulation to reflect each actor’s influence on online public opinion.

Materials and Methods

Compared to other online public opinions, earthquake online public opinion is unique. Most other types of public opinion have a long brewing period. Through media celebrities reporting on the incident, the public has a general understanding of the outline of the incident. Generally speaking, events that are reversed or inconsistent with common sense will arouse greater social attention, accompanied by anger and compassion from netizens. Camps with different opinions will engage in heated discussions, thus climaxing public opinion (Bordogna and Albano, 2007). In the evolution of online public opinion regarding the earthquake, the earthquake outbreak immediately attracted society’s attention. Online public opinion will continue to grow with the disclosure of earthquake disaster information. From the perspective of regional distribution, the participants in the topic discussion include people in the earthquake-stricken areas and people from other parts of the country. From the standpoint of age, it consists of the elderly but also middle-aged and young students. Thus, the subjects participating in the earthquake online public opinion are complex.

Research framework

This paper constructs an index for disseminating online public opinion about earthquakes based on two primary components: physical and social attributes. The physical attributes focus on the earthquake disaster and are explored through two dimensions. The first dimension includes magnitude, outbreak time, and distance from the epicenter to the city center. The second dimension is the secondary disasters caused by earthquakes and the affected population in disaster areas. In addition, regarding social attributes, earthquake online public opinion is influenced by three main variables: media communication, netizen participation, and government governance. These variables will be discussed in detail in the following sections, and a system dynamics approach will be employed to illustrate the impact of each influencing variable.

The framework of this paper is illustrated in the diagram above (Fig. 1). This study employs system dynamics methodology to construct four distinct models: the impact degree model for earthquake disaster events, the impact degree model for netizens, the impact degree model for media, and the impact degree model for government. After constructing the sub-models, this paper established a comprehensive evolution model of earthquake online public opinion. Then, Python web crawling technology collected all relevant data on microblog discussions about earthquakes. Finally, the collected data was cleaned and processed for simulation and visualization.

Fig. 1
figure 1

Research framework diagram.

Full size image

Life-cycle approach and public opinion crisis management theory

Originating in biology, the life-cycle approach depicts the evolution of different entities in a way that is similar to an organism’s life cycle. It is shown as a parabola with an “n” shape that opens downward. This theory is becoming widely used in many fields (Zhang et al. 2024) and is becoming increasingly critical. The life cycle theory views the development of things as a regular evolutionary process from birth to extinction (Wei et al. 2023). This provides feasible reference ideas for exploring the development patterns of different things. The end of one cycle is accompanied by the beginning of another, forming a cyclical mechanism. Only by clearly understanding the operating mechanism of the development of things can we better deal with and solve practical problems. Fink proposed the life cycle theory from the crisis management perspective, which roughly divides crises and risks into four stages: incubation period, outbreak period, development period, and dissipation period (Wylie, 1987). To create a more complete life cycle cost model, Guinee et al. extended the traditional environmental life cycle assessment to life cycle sustainability analysis (Guinee et al. 2011). Based on the life cycle theory (Liu et al. 2024), this paper uses the system dynamics method to construct an evolution model of earthquake online public opinion, aiming to examine the complex mechanism behind the evolution of public opinion.

The public opinion crisis management theory applies traditional principles to online public discourse (Zhang et al. 2024). Pearson et al. developed a multidisciplinary framework for crisis management research, integrating insights from psychological, socio-political, and technological studies (Pearson and Clair, 1998). Comfort illustrated the necessity of enhancing emergency management under uncertain conditions through the example of sudden natural disasters, such as hurricanes (Comfort, 2007), and advocated for the incorporation of crisis management into a complex system by government entities (Capizzo et al. 2024). Disasters are often accompanied by more significant risks of online public opinion because disasters are often accompanied by rumor spreading and authority erosion (Alexander, 2014). Crisis management is critical, as it can identify and mitigate the risks brought about by emergencies, such as terrorist acts (Huang et al. 2024). Policy conflicts arise from physical objective risks and social structural risks (Lim et al. 2016).

To better study the evolution of online public opinion, this paper combines the two theories of life cycle and crisis management to construct a complete theoretical framework (Fig. 2). This theoretical framework divides earthquake online public opinion into three different stages. Each subject’s roles and degrees of influence in various stages of the crisis are different. From this, we can analyze how different stakeholders should take targeted measures. The integrated theoretical framework forms the basis of the evolution model of online public opinion on earthquakes, which will be further elaborated in the following chapters.

Fig. 2
figure 2

Framework of public opinion lifecycle crisis management theory.

Full size image

Online public opinion refers to the formation of a public opinion in the online space, mainly manifested as the attitudes and opinions of the public towards a specific topic on social platforms. Generally speaking, as the discussion of a particular topic deepens, a significant online public opinion will be formed. In other words, the stock of online public opinion reflects a specific topic’s level of discussion and popularity. The stock of online public opinion is mainly presented through the accumulation of information on social platforms, including the views, attitudes, and emotions of various subjects (netizens, media, and government) on the Internet. So, this paper has constructed a complete theoretical analysis framework, laying the foundation for building the earthquake online public opinion evolution model in the following text. Therefore, this study aims to explore the evolution mechanism of earthquake online public opinion and clarify the degree of influence of each subject in online public opinion. To better evaluate the risk variables of public opinion, and finally propose emergency management strategies for public opinion.

System dynamics

System dynamics fundamentally depend on the intricate interdependence between system behavior and internal mechanisms (Naugle et al. 2024). This is accomplished by developing and manipulating mathematical models, which gradually reveal the causal relationships and effects that drive changes (Fu et al. 2024). This dynamic process is referred to as structure (Forrester, 1992). System dynamics is expanding, encompassing nearly every field, and progressively evolving into a relatively mature new discipline (F. L. Yin et al. 2024b). Effective management of online public opinion is essential for mitigating risks associated with crises, such as earthquakes.

Agent-based Modeling (ABM) and System Dynamics (SD) are widely used in simulation research. ABM aims to explore how individual behavior affects macro-level patterns. And present their behavioral characteristics and future development trends through interactions between individuals. However, this study categorizes individuals involved in online public opinion into the same group for analysis, such as netizens, media, etc. Individual variability makes the behavioral manifestations of participation in online public opinion vary. This makes it difficult to grasp the magnitude of the extent of a group’s role in the overall system at the macro level. Unlike ABM, SD modeling emphasizes system-wide behaviors and overall dynamics at the macro level. This aligns with this paper’s research goal to analyze the evolutionary trend of online public opinion overall (Yang et al. 2024). Therefore, this paper chooses system dynamics to construct the online public opinion evolution model (Fig. 3) because it can capture the macro dynamics and feedback loops of online public opinion in complex systems. SD modeling has more robust advantages for understanding the evolution of the whole online public opinion system.

Fig. 3
figure 3

Evolution of earthquake online public opinion.

Full size image

Before applying system dynamics to analyze online public opinion on earthquakes, this paper presents several assumptions before applying system dynamics. (1) It is assumed that the dissemination of earthquake online public opinion will not conflict with other types of online public opinion. (2) It is presumed that the variables considered are comprehensive for evaluating the influence of each component, with the possibility that other variables may be disregarded or not accounted for. (3) This paper uses Weibo data as the primary source, overlooking the online public opinion generated by earthquakes on other platforms. (4) It is posited that sub-events associated with secondary effects arising from earthquake disasters do not conflict with the main event and collectively influence earthquake online public opinion. (5) It is maintained that the retrospective nature of data collection does not significantly impact the findings. This implies that the analysis does not account for potential biases or gaps introduced by the time lag between the event and data collection.

Entropy weight

This paper employs the entropy weight method to scientifically and rigorously determine the coefficients preceding the variables in the public opinion evolution model. The amount of information reflects the content of a signal from an information source, while information entropy quantifies the information within that source (Shen et al. 2024). Specifically, it represents the expected amount of information (Grandell et al. 1980). Therefore, it is assumed that a smaller information entropy for an indicator corresponds to more significant variability. This implies that such an indicator provides more information and plays a more substantial role in comprehensive evaluation, resulting in a higher weight.

Firstly, the data for the four indicators, which include earthquake events, media, netizens, and the government, are standardized, and the weight of each indicator is calculated (Eq. 1). The P indicator represents information entropy (Eq. 2), and the weight of the P indicator is then determined (Eq. 3). Among them, \({X^{\prime} }_{{ip}}\) represents the standardized value of each indicator, and \({Y}_{{ip}}\) denotes the proportion of the P-th indicator in the i-th scheme, reflecting the varying magnitude of the indicator. Subsequently, Eq. (2) calculates the information entropy \({E}_{p}\) for each index, with values ranging between 0 and 1. Finally, the weight \({W}_{p}\) of the P-th index is determined based on the information entropy of each index.

$${Y}_{{ip}}=\frac{{X^{\prime} }_{{ip}}}{{\sum }_{i=1}^{m}{X^{\prime} }_{{ip}}}$$
(1)
$${E}_{p}={-\mathrm{ln}}\left(m\right)^{-1}\mathop{\sum }\limits_{i=1}^{m}\left({Y}_{{ip}}\times {\mathrm{ln}}{Y}_{ip}\right)\left(0\le {E}_{p}\le 1\right)$$
(2)
$${W}_{p}=\frac{1-{E}_{p}}{{\sum }_{p=1}^{n}(1-{E}_{p})}$$
(3)

Construction of public opinion evolution model based on system dynamics

Step 1: Construction of the earthquake impact model

For the convenience of discussing the impact of earthquake events on earthquake online public opinion, letters are used here to represent relevant indicators. The degree of effects of earthquake events on online public opinion is represented by \({W}_{1}\), which is a primary indicator. \({W}_{1}\) consists of five secondary indicators, including the seismic magnitude influence coefficient, denoted as \({A}_{1}\). The coefficient of the distance between the epicenter center and the nearest county or city is represented by \({A}_{2}\). \({A}_{3}\) represents the impact coefficient during the earthquake outbreak period. \({A}_{4}\) represents the historical earthquake impact coefficient. The number of secondary disasters and casualties is represented by \({A}_{5}\) (Table 1).

Table 1 Selection of the index for assessing the severity of earthquake events.
Full size table

The intensity of an earthquake is directly related to the degree of damage it causes. According to international standards, earthquakes with a magnitude of fewer than 4.5 Ms are considered “felt earthquakes,” which are usually not destructive. However, when the magnitude exceeded 4.5 Ms, it attracted attention and impacted online public opinion differently (Table 2).

Table 2 Seismic magnitude impact variables.
Full size table

This categorization was used to distinguish the differences in online public opinion \({A}_{1}\) triggered by earthquakes of different magnitudes. According to international standards, earthquake magnitudes are categorized into four categories: weak, felt, medium-intensity, and strong. For every 1.5 MS increase in the magnitude of an earthquake, the damage caused increases. To distinguish the change in the magnitude impact factor, the expression \({2}^{a-3}\) is used, where “\(a\)” denotes the earthquake’s magnitude.

The location of an earthquake also directly influences the intensity of public opinion. The attention given to an earthquake in a remote, sparsely populated mountain village differs significantly from that in an urban or suburban area. This paper determines the impact coefficient based on the distance between the epicenter and the nearest three county and city centers. The distance between the three closest counties is \(x\), \(y\), and \(z\). The formula is set to \({A}_{2}={\mathrm{ln}}\left(x\right)+1/{\mathrm{ln}}\left(y\right)+1/{\mathrm{ln}}\left(z\right)\). The specific time of the earthquake is mainly during the day, early morning, and night. When an earthquake occurs at night, it is difficult for the sleeping public to react quickly. People are more likely to be harmed in an earthquake, which can lead to greater panic. Therefore, it is set to three different influence parameters \({A}_{3}\) (Table 3).

Table 3 Variables influencing the timing of earthquake occurrences.
Full size table

Moreover, since human beings are social creatures driven by emotions, it is necessary to consider the psychological impact of earthquakes on the public. Generally, when the topic of earthquakes is discussed, the public vaguely feels a sense of fear and distress. Survivors of earthquakes that they have personally experienced may exhibit this more prominently. This is because the earthquake has already caused more incredible psychological trauma. Therefore, it is essential to establish links between current earthquakes and historical seismic events. Throughout human history, there have been major seismic events that have claimed tens of thousands of lives. Examples include the Chilean earthquake (Vigny et al. 2011), the Kanto earthquake, the Pakistan earthquake, the Tangshan earthquake, the Kansai earthquake, and the Wenchuan Earthquake (Fan et al. 2018).

The closer the association with previous earthquake events, the greater the potential to evoke media resonance. To differentiate the impact of these variables on earthquake online public opinion, this study categorizes it into three primary types, denoted as \({A}_{4}\) (Table 4). Earthquakes in remote, sparsely populated areas do not cause much concern. However, Sichuan Province is more densely populated. There have been many destructive earthquakes in history, which have left a deep impression on the public. If an earthquake with the same name breaks out, such as the Wenchuan earthquake in Sichuan Province and the Kangding earthquake, it will trigger a new round of network public opinion.

Table 4 Historical earthquake impact variables.
Full size table

In addition to the above four secondary indicators, it is also necessary to consider the variables of the number of secondary disasters and casualties caused by earthquakes, represented by A5. Due to the uncertainty and randomness of the number of secondary disasters (Han et al. 2021) and casualties (Fan et al. 2019) at different time points, A5 is represented by a table function.

Considering these uncertainties, the use of table functions is considered scientific. At present, there is no scientifically accurate earthquake warning system. Even if predictions are feasible, they are often short-lived and imprecise. The worse the warning effect, the greater the damage caused by earthquakes. In addition, over time, online public opinion gradually weakened. Analyzing the life cycle theory of public opinion, the more the online public opinion declines later, the more obvious it is. Therefore, to better highlight the life cycle characteristics of public opinion, this model uses the EXP function because the EXP function exactly represents the decay law.

Degree of earthquake action:

$${{\boldsymbol{W}}}_{{\bf{1}}}={A}_{1}* {A}_{2}* {A}_{3}* {A}_{4}* {A}_{5}* {EXP}\left(-{Time}\right)$$
(4)

Step 2: Construction of the Netizens and Media Impact Model

Similarly, the impact of netizens on online public opinion is represented by \({W}_{2}\), while the effect of media on online public opinion is represented by \({W}_{3}\). Both of these are primary indicators. \({W}_{2}\) includes the netizens’ attention, the geographical distribution of netizens, the age group of posting bloggers, and the netizens’ emotional tendencies. These four variables function as secondary indicators. \({W}_{3}\) comprises the media participation, the media influence, the media diffusion, the media dissemination of information, and the media propagation of rumors. These five variables serve as secondary indicators (Table 5).

Table 5 Indicators selected for netizens’ and media’s degree of influence.
Full size table

The role of netizens in shaping earthquake online public opinion is evident. To some extent, netizens are the main driving force behind the development of online public opinion. Many netizens use the Internet to obtain, understand, and disseminate information, thus forming multiple opinion streams. These aligned opinion streams can coalesce, continually articulating their demands or emotions (Ferree, 1985). Naturally, opposing camps will emerge. At this juncture, the two primary opinion streams will utilize the Internet as a battleground to accuse each other. Subsequently, this confrontation may escalate into abuse, heightening the emotions of Internet users (Chen et al. 2021).

In the modeling components of the netizen’s role, the first thing to consider is the degree of the netizen’s participation. The degree of netizen participation is proportional to the dissemination volume of earthquake online opinion information. This dissemination volume mainly comprises three variables: the number of microblog originals \(O\), the number of retweets \(P\), and the number of comments \(Q\). This paper uses logistic curves to model the dissemination of earthquake online public opinion (Chang et al. 2012). This approach was chosen because its trend closely reflects the evolution of online opinion about earthquakes and provides a more accurate fit for modeling the spread of such information (Zhou and Li, 2021). The relevant equations are as follows:

$$N=O{P}^{j}\left(1+{\mathrm{ln}}Q\right)$$
(5)
$${Y}_{t}=\frac{N}{1+M{e}^{-{it}}}\left({\rm{N}}\, >\, 0{;\,i}\, >\, 0\right)$$
(6)

In Eq. (5), t represents the time in days. Yt denotes the magnitude of the earthquake online public opinion at time \(t\), which is modeled using logistic curves validated by several experimental datasets. N indicates the amount of microblog information on a particular day. M, i, and j are the fitting parameters. The method calculates the daily volume of microblogs and models the evolution of earthquake online public opinion by a three-stage summation method. Parameters M, i, and j are iteratively adjusted to achieve the best fit for the data. After extensive empirical verification, the model effectively captures the dissemination of online public opinion on a particular day of an earthquake. Similarly, the magnitude of the quake will affect Internet users’ attention. Therefore, the shadow variable of earthquake magnitude is defined as \(P{1}^{{\prime} }\), and \({B}_{1}=\varepsilon * P{1}^{{\prime} }* N\).

The data of multiple microblog topics were brought into the model through many experiments. This paper used regression statistical analysis, and the value of ε for the fitting result was 0.002. After testing, the coefficient of determination fell within a reasonable range, reflecting netizen participation variables’ influence on earthquake online public opinion. In addition, the geographic location of bloggers’ postings should be considered in the model. When the topic discussion comes from different geographic regions, this indicates that the public opinion triggered by the topic is widespread. To illustrate this variable, the number of provinces participating in the discussion at different points in time, denoted as \({B}_{2}\) was collected.

From the perspective of the age distribution of Internet users participating in topic discussions, a larger span of age stages indirectly indicates a higher level of user engagement. Therefore, the age stratum of daily blog postings is counted and denoted as \({B}_{3}\). Last but not least, the affective tendencies of the content of Internet users’ postings should also be considered within the model. This is because the more intense the emotional inclination will lead to greater group polarization. Abnormal fluctuation of emotional tendency indicates that the topic discussion carries obvious positive or negative emotions. Against this background of extreme emotions, the public will tend to engage in more in-depth topic discussions, thus driving online public opinion (Birjali et al. 2021). This study uses Python software for semantic extensive text data mining to investigate the emotional tendencies in microblog postings. Specifically, the text was sentiment analyzed by constructing a sentiment lexicon. The sentiment dictionary mainly includes two categories: positive emotions and negative emotions. After pre-processing, the crawled microblog topic data can be inputted into the trained dictionary. The number of words for positive and negative emotions is obtained through data mining, quantifying positive or negative emotions (Li et al. 2024).

The initial value of affective tendency is set to 1. The value of effective fluctuation is recorded as \(b\). \(b\) = (number of positive sentiment (pos) sentences + number of negative sentiment (neg) sentences) / total number of sentences. The sentiment tendency of netizens on a specific day is represented by the equation \({B}_{4}=b+1\). The level of influence netizens can be described using these four variables. Given the inherent relationships among all four variables, which collectively affect the magnitude of earthquake online public opinion, their product form is employed. None of the four variables is continuous, and thus, a table function is utilized, which can be input into the system dynamics model.

Degree of netizens’ role:

$${{\boldsymbol{W}}}_{{\bf{2}}}={B}_{1}* {B}_{2}* {B}_{3}* {B}_{4}* (1+{EXP}(-{Time}))$$
(7)

The role of media in the composition of online public opinion cannot be ignored. Compared to ordinary netizens, the media has a huge fan base. Moreover, the media has multiple channels through which to obtain information. For example, the media generally knows the latest information about the disaster after an earthquake. From a particular perspective, the media guides the development of online public opinion. On the one hand, it bridges the information gap between official media and ordinary netizens, playing a buffering role. On the other hand, if the media has a biased understanding of the event, they will release inaccurate reports. Due to various media outlets interpreting events from different perspectives on the internet, a large amount of false information and even rumors may be generated. The main reason is the asymmetry of information. A vast number of netizens lack an understanding of actual events and only believe in media reports, leading to the further spread of rumors and pushing online public opinion to a climax (Wang et al. 2021). The primary variable to consider is the degree of media participation, represented as variable \({C}_{1}\), which can be quantified by the daily number of media reports derived from collected microblogging data. The influence of the media is denoted by variable \({C}_{2}\).

Generally, the more followers a media organization has, the greater the influence of its postings. To better differentiate the impact of media with varying follower counts, the influence coefficient is divided into four categories (Table 6), with w denoting units of 10,000 followers. The media organization with the most significant number of followers is selected, and its influence coefficient is determined based on its follower volume.

Table 6 Media impact variables.
Full size table

After the media releases information, many followers further disseminate it through retweets. The degree of media diffusion, denoted as \({C}_{3}\), can be measured by the number of retweets, \(P\). It is defined as \({C}_{3}={\root{6}\of{P}}\). The effectiveness of information dissemination is also related to the content of media posts. The impact of different posting methods on information dissemination varies. According to the principles of communication studies, video has the best information dissemination effect, followed by images and, finally, pure text. The content of media posts determines the degree of information alienation \({C}_{4}\) (Table 7).

Table 7 Dissimilation variables for information.
Full size table

In addition to the previously mentioned variables, the influence of rumors on online public opinion must also be considered. Certain media outlets may distort the truth or selectively emphasize particular aspects of an issue to attract public attention. Hence, the daily count of impactful rumors is tallied and designated as \({C}_{5}\). These five variables interact with each other and collectively influence the evolution of online public opinion.

Degree of netizens’ role:

$${{\boldsymbol{W}}}_{{\bf{3}}}={C}_{1}* {C}_{2}* {C}_{3}* {C}_{4}* {C}_{5}* \left(1+{EXP}\left(-{Time}\right)\right)$$
(8)

Step 3: Construction of the government impact model

The overall degree of the government’s impact on online public opinion is denoted as \({W}_{4}\) which serves as a primary indicator. This includes five secondary indicators: the government’s attention, the government’s information disclosure, the government’s actual disaster relief progress deficiencies in government response, and donations from entrepreneurs and celebrities. These secondary indicators are detailed in Table 8.

Table 8 The choice of government role degree indicators.
Full size table

The government plays an essential role in forming and developing public opinion about the earthquake online. On the one hand, it releases official information and implements rescue operations. On the other hand, it handles erroneous information on the platform and takes punitive measures against the perpetrators. For instance, following the Yibin earthquake, official media actively reported on the latest developments in affected areas and initiated rescue operations.

The higher the magnitude of the earthquake and the greater the spread of rumors, the more attention the government devotes to the event. In this study, governmental attention is represented as variable \({D}_{1}\), which is measured by the frequency of government updates on the Yibin earthquake. The number of daily government broadcasts related to the event is directly recorded and incorporated into the software using a table function. Throughout the relief effort, the government will consistently update the public on the latest developments in the disaster-stricken areas, reflecting the extent of government information disclosure, denoted as variable \({D}_{2}\).

The degree of information disclosure includes the number of casualties, the number of affected people, and the progress of rescue efforts. The actual progress of government disaster relief is marked as \({D}_{3}\), directly related to the earthquake warning system. The more complete the earthquake warning system (Xu et al. 2020), the smaller the losses the public suffers in earthquake disasters (Table 9).

Table 9 Variables influencing seismic warning effects.
Full size table

Moreover, if the government makes significant mistakes in disaster relief work, it will cause public dissatisfaction. The result is an increase in the risk of online public opinion, thereby reducing the government’s credibility. The lack of government work is defined as \({D}_{4}\). After screening the data, no apparent deficiencies in government work were found. Thus, the indicator is set to unit 1. Research revealed that society mobilized a wave of earthquake relief efforts following the earthquake. This included donations posted on Weibo by entrepreneurs, celebrities, and others, which created a trending topic on the platform. Relevant institutions and celebrities have a large fan base. Therefore, they have a strong appeal in organizing donations. This indicator is defined as \({D}_{5}\) mainly represented by the number of publicly available donation activities by organizations or individuals. The above five variables interact with each other and affect public opinion.

Degree of government Role:

$${{\boldsymbol{W}}}_{{\bf{4}}}={D}_{1}* {D}_{2}* {D}_{3}* {D}_{4}* {D}_{5}* (1+{EXP}(-{Time}))$$
(9)

Step 4: Construction of the earthquake online public opinion evolution model based on system dynamics

The previous section constructed models to gauge the degree of online public opinion on earthquakes from four perspectives: earthquakes, netizens, media, and government. This study has developed a comprehensive model for disseminating online public opinion about earthquakes to integrate all these variables. This model provides a framework for understanding how online public opinion about earthquakes is disseminated. The general equation is as follows:

Earthquake online public opinion:

$$E=\alpha * {W}_{1}+\beta * {W}_{2}+\gamma * {W}_{3}+\delta * {W}_{4}$$
(10)

Data sources

This article uses Python software to crawl keywords and topics. It is worth noting that the data collection in this study is retrospective analysis and may be influenced by platform censorship and content moderation. Although the specific degree of quantitative review and moderation is challenging, this factor should be considered a potential research limitation. The research focuses on a particular time frame, analyzing posts within 25 days following the initial Weibo release. The scraping targeted the keywords “Yibin M5.7 Earthquake” and the hashtag “#Yibin M5.7 Earthquake#.” Data were collected from December 16, 2018, at 13:08 to January 11, 2019, at 18:38, covering 25 days. According to the backend data, the release data for keywords was 812 posts (excluding hashtag and repost data). In comparison, the hashtag post data totaled 88,650 posts (including hashtags and reposts) (Fig. 4). The horizontal axis in the figure represents the 25 days after the earthquake occurred. The left vertical axis represents the number of posts on the original Weibo, and the right axis represents the total number of reposts on Weibo. Bar charts and line charts represent the number of posts and reposts.

Fig. 4
figure 4

Statistical chart of Weibo posts and forwards.

Full size image

Regarding original Weibo postings, the earthquake that occurred on December 16, 2018, quickly peaked, characterized by a sharp increase in posts following the event. This surge was especially notable on December 16, 17, and 18 before the activity subsided. The number of provinces from which original Weibo posts originated is used as a parameter in later analysis. ArcGIS software was employed to visually represent the origins of bloggers, using different colors to denote the two indicators: the number of original Weibo posts and the number of forwards (Fig. 5).

Fig. 5
figure 5

Distribution statistics of Weibo posts by province.

Full size image

Different numerical ranges represent the total amount of Weibo posts and shares. The darker the color, the higher the value of the post. The darker the color of provinces and regions with greater attention, the greater the earthquake online public opinion. Figure 5 visually illustrates the spatial distribution of public participation in earthquake topic discussions.

From the spatial distribution of public opinion, Weibo posts are mainly concentrated in the central and western regions due to the relatively dense population. Sichuan Province has the highest online public opinion as the epicenter of earthquakes. The local people actively shared the earthquake situation through photos and videos, which resonated with other netizens. The eastern coastal region has a developed industrial foundation and is a destination for labor migration in Sichuan Province. Therefore, the people in the area are also very concerned about the latest earthquake. Of course, as a public Weibo communication platform, there are differences in the level of attention to earthquakes in different regions, which does not affect the overall trend of the evolution of online public opinion on earthquakes.

Subsequently, the daily volume of Weibo information was calculated using Eq. (5) with the indicators “original Weibo number (\(O\)),” “forwarding number (\(P\)),” and “number of reviews (\(Q\)).” The computed daily information volume is presented in Table 10. After determining the daily Weibo information volume, the simulation of changes in earthquake online public opinion was conducted using the three-stage summation method. Through multiple iterations, the parameters \(M\) \(i\) and j were established, resulting in the values \(M\) = 123.34, \(i\) = 1.27, and \(j\) = 0.5.

Table 10 Daily information volume of microblogs on earthquake topics.
Full size table

This paper adopts regression statistical analysis to verify the model’s effectiveness and the rationality of the parameters. The determination coefficient \({R}^{2}\) describes the difference between actual and predicted values. The closer it is to 1, the better the fitting effect of the model. The result of \({R}^{2}\) is 0.82, indicating that the model in this study has a high degree of fitting.

Substitute the fitted \(M\) and \(i\) into formula (6) to obtain the trend of public opinion information dissemination (Fig. 6). According to the inflection point formula \(\frac{{\boldsymbol{ln}}{\boldsymbol{M}}}{{\boldsymbol{i}}}\) of the logistic function curve, the time interval is calculated to be 3.79 days. The end time of the spreading period is obtained as follows: \({T}_{2}\) = 18:57 on December 20th, and the corresponding information amount at this time is 7233.42. Using this information as a dividing line, we will find the first intersection point on the curve, which occurs at \({T}_{1}\) = 20:23 on December 16th. The period \({(0,T}_{1}\)) is defined as the burst period, the interval [\({T}_{1}\), \({T}_{2}\)] is defined as the spreading period, and the period (\({T}_{2}\), +∞) is defined as the fading period. The horizontal axis represents the 25 days following the outbreak of the Yibin earthquake, while the vertical axis indicates the volume of earthquake online public opinion dissemination.

Fig. 6
figure 6

Stages of online opinion on the Yibin earthquake.

Full size image

Model simulation analysis

The earthquake online public opinion evolution model was constructed and input into simulation software to better understand the evolution patterns of earthquake online public opinion. The Sichuan Yibin M5.7 earthquake microblog topic data was then used to simulate the online public opinion (Fig. 3). Given the complexity of the variables influencing online public opinion, this section first incorporates four components of the public opinion evolution model into the software, to study the respective contributions of earthquake events, netizens, media, and government to earthquake online public opinion. Finally, the entire model will be simulated in software to analyze the overall trend of online public opinion on earthquakes.

Simulation analysis of the action degree of earthquake disaster events

This paper crawled the Weibo data of the “Yibin M5.7 earthquake” for one month. Analyze the changes in earthquake online public opinion daily. System dynamics were used to examine the variations in public opinion. According to the data, the earthquake’s epicenter is approximately 20 kilometers from Xingwen, 30 kilometers from Gongxian, and 40 kilometers from Changning. The specifics are as follows: The effect level of the earthquake event is calculated as \({W}_{1}\) = \({2}^{5.7-3}* \left[1/\mathrm{ln}\left(20\right)+1/\mathrm{ln}\left(30\right)+1/\mathrm{ln}\left(40\right)\right]* 1* 3/2* (1+{EXP}(-{Time}))\). All data were entered into the simulation software to determine the impact level of the earthquake event (Fig. 7). In the process of the earthquake’s effect on online public opinion, the greater the earthquake’s magnitude, the more likely it is to threaten people’s lives. The impact of earthquakes in the initial stage is more prominent, as they directly cause massive casualties. In the later stages, it is mainly the aftershocks or secondary disasters that increase the number of casualties.

Fig. 7
figure 7

Simulation of the individual effects of the four main entities.

Full size image

Simulation analysis on the role degree of netizens and media

The inputs of the variables related to netizens were put into the model (Fig. 7). Netizens play a fundamental role in the evolution of online public opinion. Netizens are both presenters and promoters of public opinion. In other words, netizens directly influence the spread, development, and diffusion of online public opinion. This analysis is because netizens are the largest and central to online public opinion. The horizontal coordinate is 25 days after the earthquake outbreak, and the vertical coordinate indicates the value of the earthquake online public opinion. The large number of Internet users makes them an integral part of the netizen community. While the role of individual netizens alone is small, the collective voice of tens of thousands of netizens becomes essential when they form an interest group. For example, a topic such as “Support Sichuan” can resonate widely, forcing the Government to pay attention to the group voice.

Input five media-related variables into the model and run it to obtain a trend chart of online public opinion (Fig. 7). The media significantly impact public opinion in the early stages. During this stage, the press mainly reports on the actual situation of earthquake disasters promptly. In the later stage, the impact on public opinion was relatively small. With the completion of earthquake relief, the media’s attention to the Yibin earthquake decreased.

In this paper, media information dissemination is analyzed using Gephi. It is mainly based on the number of media postings and their retweets. This results in a network graph of posting subjects (Fig. 8), where the same color indicates a cluster of protocols. The graph visualizes the influence and reach of various posts, demonstrating how the topic gained momentum and how media coverage contributed to expanding public opinion. After the Yibin earthquake, residents continued to post earthquake-related articles, gradually forming the topic “Yibin 5.7 magnitude earthquake”. The topic gradually became a popular search term. Then, Cover News reported a 5.7 magnitude earthquake in Xingwen County, Yibin City, Sichuan Province, accompanied by a local video. The report gained more attention and was retweeted by the public, thus further promoting the development of online public opinion. The media plays an essential role in developing online public opinion about earthquakes. The media not only presents facts through reporting but also promotes the formation of public discourse. Due to the characteristics of information asymmetry, false information and rumors are easy to breed. Some media may exaggerate or distort facts to guide public opinion. In addition, to attract public attention, things unrelated to the report may be linked. The purpose of manipulating online public opinion can be achieved through the abovementioned propaganda tactics. Due to the lack of facts, most of the public tends to support the media’s claims, thus triggering heated online debates. For example, after the Yibin earthquake, under the guidance of the media, netizens engaged in in-depth discussions on issues such as “inefficiency of the government in disaster relief,” “celebrity donations,” and “escalation of the disaster.” If the Government and the State media had failed to dispel the rumors, it would have significantly impacted the relief efforts and further undermined the Government’s credibility.

Fig. 8
figure 8

Main network diagram of online public opinion releases.

Full size image

Simulation analysis of the degree of government’s role

After the earthquake, the government’s main task is to organize earthquake relief efforts. Firstly, actively organize rescue teams to go to the disaster-stricken areas, adhering to the principle of putting the people at the center, with officers, soldiers, and firefighters racing against time to enter the earthquake-stricken areas and carry out rescue activities. Secondly, the whole society should be called to join the earthquake relief camp and organize fundraising activities for disaster-stricken areas. Thirdly, public opinion must be fully collected, and reasonable disaster relief strategies should be formulated. Finally, the latest rescue progress and information should be released promptly, relevant guarantees for the people in the disaster area should be provided, and plans for rebuilding their homes should be formulated. From this perspective, the government has consistently promoted the development of the earthquake online public opinion development. Data related to the M5.7 earthquake in Yibin were input into the model to generate a graph illustrating the extent of government involvement (Fig. 7). The vertical axis represents the magnitude of the earthquake online public opinion. Notably, government involvement exerts a more lasting impact compared to other variables. In addition to its significant influence during the early stages, it occasionally stimulates online public opinion about earthquakes in later phases. This phenomenon can be attributed to officials frequently highlighting recovery efforts in the affected area during the waning phase of public discourse, thereby triggering new rounds of discussion.

Analysis of model simulation results

The previous article examined the impact of earthquakes, netizens, media, and the government on earthquake online public opinion. By inputting crawled microblog data into an evolution model, the study analyzed the changes in public opinion associated with different subjects over time. Consequently, the author compared the impact of earthquakes, netizens, media, and the government on earthquake online public opinion. Utilizing crawled Weibo topic data from the M5.7 earthquake in Yibin, the degree of effect was normalized to a scale from 0 to 5, where a higher number indicates a more significant impact on earthquake online public opinion (Fig. 9).

Fig. 9
figure 9

Trend chart of the degree of impact on public opinion by different subjects.

Full size image

Earthquakes and netizens have a direct impact on online public opinion. Earthquakes play a decisive role in the formation of online public opinion. The media plays an essential role in spreading and developing online public opinion. The government has always played a regulatory and guiding role throughout the development of public opinion. Tens of thousands of netizens are the foundation of opinion groups and are in a guided position.

In addition to examining the influencing variables for each subject, it is crucial to construct a comprehensive evolution model for earthquake online public opinion. Therefore, \(\alpha\), \(\beta\), \(\gamma\) and \(\delta\) in Eq. (10) must be determined. This section employs the entropy weighting method to calculate the significance of the indicator. Standardize the four primary indicators by selecting other earthquake Weibo topic data for fitting. Calculate the proportion of the four primary indicators using Eq. (1). Calculate the information entropy of the four primary indicators using Eq. (2). Finally, the weights of the four primary indicators are calculated using Eq. (3). After the abovfour steps of calculation, \(\alpha =0.34,\beta =0.39,\gamma =0.45,\delta = 0.39\). This study uses the determination coefficient \({R}^{2}\) to describe the magnitude of the difference between the actual value and the predicted value. The closer \({R}^{2}\) is to 1, the better the model fitting effect. The final result is R2 = 0.61, indicating that the model performs well.

The four coefficients were substituted into Eq. (10) and entered into software to simulate changes in earthquake online public opinion. The gray line labeled “current” illustrates the trajectory of public opinion, showing a rapid peak immediately following the earthquake’s onset, followed by a gradual decline into a dissipation phase (Fig. 10). Analyzing this through a life-cycle approach reveals a prolonged process from the emergence to the eventual disappearance of earthquake online public opinion. Public opinion evolves alongside the progression of an earthquake event. From the perspective of public opinion crisis management theory, the evolution of online public opinion is consistently accompanied by a public opinion crisis. It may escalate into a more significant crisis when it exceeds the average critical threshold. Therefore, it is crucial to identify the key variables affecting the development of online public opinion on earthquakes. Thus, the author will adjust the model parameters to analyze the trend of online public opinion on earthquakes under different conditions.

Fig. 10
figure 10

Schematic diagram of the earthquake online opinion simulation.

Full size image

Assuming other variables remain constant, this study explores the impact of increasing the role of netizens on overall earthquake online public opinion. The blue line represents the 30% increase in netizen influence on earthquake online public opinion, which is a function of both the quantity and emotional content of posts (Fig. 10). Compared to the “Current,” the simulated public opinion on the seismic online is slightly higher. In the early stages, the blue line closely follows the original curve but peaks later. Additionally, the decline in public opinion extends over a longer period. Netizens are the most active group in the expression of online public opinion. Netizens are not restricted by time and space; they can post, forward, and comment anytime and anywhere. When more and more netizens discuss a specific topic, it is straightforward to form a clustering effect, thus driving online public opinion to a climax.

Additionally, while controlling for other variables, the media’s engagement and dissemination have increased. More media outlets focus on online public opinion about earthquakes and provide real-time coverage of the affected areas. This increases the extent of the media’s role in earthquake online public opinion by 30%, as indicated by the red lines (Fig. 10). The simulation reveals a significantly higher peak in public opinion compared to previous scenarios, suggesting a more significant influence of the media on online public opinion about netizens. With more resources and larger audience bases, the press exerts substantial control over public opinion. Capable of setting the online narrative, the media can direct public opinion in specific directions. However, the diverse media landscape has created multiple opinion hubs, resulting in conflicting viewpoints that intensify and propel online public opinion toward heightened risks and climactic outcomes.

Ultimately, under varying conditions, the impact of increased government intervention on overall earthquake online public opinion is explored, as represented by the green lines. Measures include enhancing control over online public discourse, promptly dispelling internet rumors, utilizing official media to broadcast the latest developments in disaster areas, and actively encouraging societal support for earthquake relief. These interventions led to a 10% reduction in the other three parameters modeled, as shown in the simulation results of the green line in Fig. 10. The study results show that the effect is significant when the government takes a series of measures to control earthquakes online public opinion. Overall, the value of earthquake public opinion decreases, and the rapid decline shortens the cycle of public opinion, thus reducing the risk of escalation of earthquake online public opinion. Official media play a crucial role in this process. Hence, it is essential to establish a comprehensive early warning and forecasting system for earthquake online public opinion. After an earthquake, the government must proactively manage online public opinion about the quake and guide its dissemination to reduce the associated risks.

The simulation reveals that the government intervention led to a 10% reduction in the other three parameters. The total amount of online public opinion decreased rapidly, as indicated by the green line (Fig. 10). This shows that the effect of the series of measures taken by the government is significant. Overall, the rapid decline in public opinion directly shortens the cycle of public opinion evolution, reducing the risk of online public opinion escalation. Among the many measures, the role of the official media is crucial and indispensable. Therefore, it is essential to establish a complete early warning system for earthquakes and public opinion online. After an earthquake, the government and the official media must take the initiative to intervene in the dissemination of online public opinion. They should actively guide the dissemination of public opinion, thereby reducing the risk of online public opinion.

Conclusions and Discussion

Conclusions

This article focuses on the formation process of earthquake online public opinion, which fills the gap in this field. This paper creatively analyzes the importance of the roles of different subjects in online public opinion from four perspectives: earthquakes, netizens, media, and government. To explore the evolution mechanism of earthquake online public opinion, this paper combines life cycle theory, crisis management theory, and system dynamics methods to construct a comprehensive public opinion evolution model. Taking the Yibin earthquake with a magnitude of 5.7 as an example, online public opinion can be divided into three stages: the burst period, the spreading period, and the fading period. Then, simulations are conducted separately from the earthquake, netizens, media, and government to explore the impact of each subject on the evolution of online public opinion. Finally, the evolutionary trajectory of online public opinion on earthquakes is simulated by adjusting various variables.

This study finds that (a) Earthquake online public opinion possesses unique characteristics. Earthquakes affect large areas, resulting in extensive disaster zones and a more complex array of participants in the public discourse. (b) The magnitude and severity of an earthquake directly influence the development of earthquake online public opinion. In particular, incredibly destructive earthquakes that result in significant casualties tend to provoke higher-risk earthquakes online public opinion. (c) The media wields considerable influence over earthquake online public opinion, demonstrating strong dominance and the ability to shape the direction of public discourse. (d) All aspects of government earthquake relief efforts are subject to public scrutiny. Effective disaster response that minimizes casualties can garner significant public support and trust.

Of course, earthquakes bring significant risks to online public opinion mainly due to several factors. Firstly, there is an insufficient earthquake warning system and the destructive nature of the disaster itself. Secondly, social platform anonymity and privacy allow users to express their opinions freely at any time. However, it is difficult to trace the sources of online public opinion, which poses a challenge for the government in managing online public opinion. Again, the government’s lack of transparency in earthquake relief work has led to information asymmetry, which can easily facilitate the spread of rumors. In addition, some media maliciously spread false information, inciting public anger and further complicating earthquake online public opinion.

Suggestion and Discussion

Once the operational mechanisms of earthquake online public opinion are understood, strategies can be formulated based on the varying degrees of influence that different variables exert on this public opinion. The specific suggestions are as follows:

  1. (1)

    The government should establish an early warning system for online public opinion management of earthquakes (Wang et al. 2020). The system is based on artificial intelligence (AI) technology to improve its effectiveness in regulating and managing online discussions. Although the debate over artificial intelligence technology’s benefits and potential risks is ongoing (Allen and Melgar, 2019), AI provides feasible ideas for monitoring and controlling the spread of misinformation (Cheng et al. 2020).

  2. (2)

    The government should intervene in the evolution of online public opinion on earthquakes and actively discover hot search content with topics. After investigating the focus of the dispute, release the investigation results as soon as possible. Simultaneously, paying attention to the cutting-edge information on earthquake relief and disaster relief, the live streaming connection method has a solid positive guiding effect. The government should promptly release authoritative details and deal with false information and rumors, which helps to increase public trust in the government.

  3. (3)

    The government should actively organize professional personnel for disaster relief activities. The public sector should promptly donate necessary and produced living materials to disaster-stricken areas and plan for rebuilding homes in the disaster-stricken areas. Official media should call on the whole society to join the disaster relief camp. Finally, the psychological condition of the affected people should also be taken seriously, and providing targeted psychological counseling is necessary.

Of course, the research also has limitations. The dissemination of online public opinion involves more social platforms. The data source of this paper only selects microblog data. Thus, data from different online platforms can be used for further research. Second, this paper adopts a retrospective data collection method, which may overlook the dynamic nature of public opinion. Future research can adopt real-time data collection methods.