From a CAS theoretical perspective, the online public opinion system in higher education institutions is a complex adaptive system formed through the interaction and organic coupling of four major subsystems: diverse actors, information dissemination, technological support, and socio-cultural factors ([6]). In the era of omni-media, the assessment and guidance of the online public opinion environment have become key issues in the governance of higher education institutions ([10]). The multi-actor system encompasses core participants such as students, teaching staff, university administrators, the media and the general online public, and serves as the primary vehicle for the generation and evolution of public opinion; the information dissemination system is responsible for the transmission and diffusion of viewpoints, emotions and value judgements, acting as the key link connecting all actors; its immediacy and viral diffusion characteristics within the all-media communication landscape significantly alter the pace at which public opinion evolves ([18]); The technological sup-port system provides the platform and technical foundation for the dissemination of online public opinion, determining the immediacy and viral nature of information dissemination; the socio-cultural system influences the cognitive frame-works and interactive logic of the actors, thereby constraining the direction of public opinion evolution. From the perspective of text analysis, research on online public opinion in Chinese higher education institutions has gradually shifted its focus to the interrelationships between actors, issues and dissemination mechanisms ([12]).

The core characteristics of CAS theory align closely with the operational logic of online public opinion systems in higher education institutions. The system’s intrinsic mechanisms can be analysed through key dimensions such as aggregation, non-linearity and flow; these characteristics interpenetrate to form the foundation for the evolution of public opinion, as shown in Table 1 below.

Table 1. An interpretation of the complex system of online public opinion in higher education institutions based on CAS theoretical characteristics.

Within the CAS theoretical framework, the online public opinion system regarding academic misconduct in higher education institutions constitutes a dynamic exchange and multidimensional interplay among diverse actors centred on specific issues; it transcends linear logic, exhibiting non-linearity and multi-layered characteristics ([13]). Non-linearity manifests in the way that minor disturbances by individuals can be amplified through positive feed-back within the network, triggering sudden shifts in public opinion; in the age of algorithms, platform-driven algorithmic recommendations and dissemination pathways further reinforce this non-linear diffusion, thereby driving the optimisation of paradigms and approaches for governing online public opinion among university students ([5]). The multi-layered nature is manifested in a cascading chain of transmission comprising micro-level strategic interactions, meso-level consensus reinforcement, and macro-level public opinion formation. This system possesses both self-organisation and adaptability. Self-organisation enables the system to spontaneously form an evolutionary order, whilst also increasing uncertainty; adaptability allows various actors to dynamically adjust their strategies in response to the progression of public opinion, constituting the endogenous driving force behind the system’s evolution, whilst also placing new demands on enhancing the effectiveness of ideological and political education for university students within the context of online public opinion governance ([11]).

From a CAS theoretical perspective, the core of the adaptive mechanism of university online public opinion systems lies in the coordinated dynamic adjustment between macro and micro levels, which relies on closed-loop feedback to respond to changes in the internal and external environments, thereby maintain-ing system stability and driving its evolution ([16]). At the macro level, the public opinion system is dynamically coupled with the social environment; when subjected to external stimuli, it reconfigures the interaction network among actors through self-organisation, exhibiting characteristics such as shifts in the focus of public opinion. This aligns with the principles of “flow” and non-linearity, whereby the flow of information and emotions, as well as perturbations at key nodes, can easily trigger systemic abrupt changes. The optimisation of micro-level agent behaviour and the restructuring of macro-level structures are interconnected in a bidirectional manner. Individual agents rely on “identifiers” to select interaction partners and update their “internal models” to adjust strategies; this essentially constitutes the reorganisation and adaptation of “building blocks”. Macro-level restructuring provides direction for micro-level behaviour, whilst micro-level optimisation drives the iteration of the macro-level system. The interaction between the two endows the system with continuity in adaptation across events, providing both theoretical and practical foundations for public opinion governance.

The framework of this study is underpinned by CAS theory. Aligned closely with the objective of resilient governance of public opinion regarding academic misconduct in higher education institutions, and incorporating the interactive logic of four subsystems, it constructs a four-tier progressive theoretical framework comprising “objectives - support - operations - implementation”, thereby providing a systematic platform for quantitative analysis and governance practices.

The core objective of the framework is to enhance the resilience of the public opinion system, encompassing three dimensions: disturbance resistance, recovery capacity and adaptability. Disturbance resistance optimises system structure to reduce the risk of explosive diffusion of public opinion; recovery capacity shortens the cycle of public opinion subsidence and suppresses rebounds through precise intervention; adaptability focuses on establishing long-term mechanisms to achieve early warning of public opinion and the upgrading of cross-incident governance. The system support layer is based on four subsystems, which define functional boundaries and interaction logic, working in concert to form the foundation of the framework. The Operational Variables Layer translates governance objectives into quantifiable indicators such as intervention intensity and public opinion heat, forming a closed-loop with data preparation and the construction of variable relationships to ensure operability. The Dynamic Operation Layer follows CAS theory to form a closed-loop of “agent interaction - information circulation - intervention feedback - system optimisation”. Through agent interaction, multi-stream circulation and variable regulation, it achieves system restructuring and enhanced resilience.

The study took as its subjects a joint complaint lodged by students at a university in Central China and the case of an undergraduate student, Liu A, at a university in South-West China, and systematically collected data covering the entire lifecycle of public opinion.

2.3.1. Data Preparation

Data collection covered the entire lifecycle of both cases: the incident at a university in Central China took place from 16 January to 9 February 2024, whilst the incident at a university in South-West China occurred from 7 May to 20 May 2025. Both cases were comprehensively covered across all stages—from the emergence, outbreak and spread of public sentiment to its eventual subsidence. Web crawlers were utilised to extract interaction data from multiple platforms, official texts and media analyses, ensuring the continuity and comprehensiveness of the data. Data pre-processing was carried out in accordance with standardised guidelines. Following deduplication, filtering of invalid data and removal of outliers, the valid datasets for the two cases comprised 126,000 interaction records, 892 text documents and 576 time-series records respectively.

2.3.2. Selection of Variables

The selection of variables drew upon the public opinion resilience assessment indicator system proposed by [17] and the research findings on public opinion in higher education institutions by [7]. Combining these with the evolutionary characteristics of public opinion systems under CAS theory, core and associated variables were identified, and the scoring criteria were optimised.

Public opinion intensity was assessed using three secondary indicators: Weibo views, platform discussions and news coverage. These were normalised to a scale of 0 - 100 using the range method, with the normalised range defined by the minimum and maximum values of each indicator during the study period for the two case studies. Intervention intensity is calculated as a composite of response timeliness (weighted at 40 per cent) and the degree of information disclosure (weighted at 60 per cent); these two weights reflect their actual impact on the management of public opinion in higher education institutions. Event sensitivity is composed of the degree of ethical controversy (70 per cent) and the degree of professional relevance (30 per cent); these two indicators are first coded on a scale of 0 - 4 and then converted to a percentage scale. Based on these calculations, the observed mean values for each variable in the two case studies are as follows: public opinion intensity 68.3, intervention intensity 52.7, incident sensitivity 75.2, and public participation 62.5. The weightings assigned to the factors influencing the evolution of public opinion are detailed in Table 2 below.

Table 2. Explanatory table of weightings for public sentiment evolution factors.

To accurately characterise the patterns of public opinion evolution under different intervention intensities, a system dynamics model was constructed around the core endogenous variable “public opinion intensity”. All equations were determined through regression analysis of historical data from two case studies; the coefficients passed significance tests (P < 0.01) and are consistent with the logic of public opinion evolution.

Core driving equation: Rate of change in public opinion intensity = Event sensitivity × 0.3 + Public engagement × 0.4 – Intervention intensity × 0.5. Drawing on the relevant research framework proposed by [3], this confirms that the suppressing effect of intervention intensity on public opinion is particularly pronounced. Concurrently, equations for three major subsystems were established:

Public engagement = Event sensitivity × 0.6 + Previous public opinion intensity × 0.4, reflecting the dual drivers of the event’s appeal and existing intensity;

Intervention intensity = University response speed × 0.4 + Transparency of information × 0.6, highlighting the central role of information transparency;

Event sensitivity = Ethical controversy × 0.7 + Professional relevance × 0.3, indicating the dominant role of ethical controversy.

Case studies have demonstrated that, following an increase in the intensity of intervention at a university in Central China, the rate of change in public opinion shifted from positive to negative, indicating a marked suppressive effect; simulation results for an incident at a university in South-West China showed over 85 per cent alignment with actual public opinion trends, fully validating the applicability of the formulae.

With a focus on enhancing the adaptability of public opinion management systems, higher education institutions are establishing long-term governance mechanisms from both institutional and technological perspectives to strengthen the cultivation of academic integrity. Academic integrity education is integrated into undergraduate and postgraduate training programmes, with students’ awareness of standards being enhanced through diversified teaching methods; in research, end-to-end record-keeping and data anomaly alerts are implemented to prevent academic misconduct at a technical level; relying on official new media platforms, science popularisation and interactive engagement are carried out, whilst collaborating with local cyberspace administration departments to swiftly address misinformation, guide rational public discourse, and promote the coordinated development of academic integrity and a healthy online ecosystem.

This study used “public opinion on academic misconduct in higher education institutions” as its search topic, limiting the search period to January 2023 to June 2025. Data was sourced from Weibo’s trending topics, Zhihu’s trending topics, as well as public reports from mainstream media outlets such as the People’s Daily, Xinhua News Agency, The Paper and the Beijing News, and official channels of higher education institutions. Following an initial screening, a total of 31 public opinion incidents involving academic misconduct in higher education institutions were identified, constituting the initial set of candidate cases.

Building upon this initial set, a tiered screening process was conducted based on six criteria. Firstly, issue focus: the core of the incident must relate to academic misconduct in higher education (including data fabrication, unauthorised authorship, plagiarism, and disputes over the authenticity of research findings); after excluding non-academic disputes, 19 cases remained; Second, public attention: the main Weibo hashtag must have accumulated over 100 million views or ranked within the top 10 trending topics, ensuring the incident had sufficient public impact; this left 11 cases; Third, complete lifecycle: the incident must have been continuously traceable via public channels for at least 10 days, from the outbreak of the public outcry to the university’s official announcement resolving the matter; this left 7 cases; Fourth, clear institutional outcome: the university involved must have issued an official investigation report or disciplinary decision; this left 5 cases; Fifth, data availability: interactive data from multiple platforms, official texts and media reports must be systematically accessible without significant gaps; 3 cases met this criterion; Sixth, exclusivity: after removing duplicate reports from the same source and incidents not involving higher education institutions, 2 cases were ultimately retained.

Consequently, this paper selects a joint student complaint at a university in Central China and the “Liu A” incident at a university in South-West China as twin case studies. It analyses the evolution of public opinion in each case and compares the differences between them in terms of triggering mechanisms, the pace of development and response strategies, thereby illustrating the practical application of CAS theory in public opinion governance.

3.1.1. The Incident Involving a Joint Complaint by Students at a University in Central China

A joint complaint by students at a university in Central China was triggered primarily by students’ real-name allegations of academic misconduct against their supervisor. On 16 January 2024, 11 students from the university disclosed in a joint letter of complaint that their supervisor, Mr Huang, had engaged in acts such as data manipulation and unauthorised authorship. On the same day, the university responded by establishing a task force to launch an investigation. After evolving through several stages, public sentiment gradually subsided, sparking in-depth societal discussions regarding the academic oversight system and the power dynamics between teachers and students. Based on CAS theory, the patterns of public opinion evolution in this incident are analysed across four dimensions—stages of evolution, key events, actor behaviour and characteristics of public opinion—as summarised in Table 3 below.

Table 3. Characteristics of the evolution of public opinion surrounding a whistleblowing incident at a university in Central China, based on CAS theory.

3.1.2. The Case of Liu A, an Undergraduate Student at a University in the South-West

The case involving an undergraduate student, Liu A, at a university in south-west China was triggered by doubts regarding the authenticity of her academic achievements. On 7 May 2025, publicity surrounding Liu A’s publication of 14 SCI papers and acquisition of three patents attracted public attention. Subsequent revelations that she was the daughter of Professor Liu B, one of the co-authors, led the university to launch a special investigation on 8 May. On 10 May, the university announced its findings, confirming that there had been inappropriate authorship practices; the incident brought the issue of fairness in the allocation of academic resources into sharp focus. Based on CAS theory, the characteristics of the public opinion evolution are summarised in Table 4 below.

Table 4. Characteristics of the evolution of public opinion surrounding the “Liu A” incident at a university in south-west China, based on CAS theory.

Focusing on the core variables of the system dynamics model and resilience assessment indicators, key standardised data were extracted from two case studies for model validation. The public opinion cycle for an incident at a university in Central China lasted 25 days, with a peak intensity of 85; the cycle for an incident at a university in South-West China lasted 14 days, with a peak intensity of 75. Both universities responded within 24 hours; the Central China institution employed an intervention intensity of 70 with full transparency throughout the process, whilst the South-West institution employed an intervention intensity of 75 with targeted disclosure. Regarding resilience data, the Central China institution took 5 days for sentiment to subside, experienced a 3.53 per cent rebound, and had a 35 per cent probability of a secondary public opinion outbreak; the South-West institution took 3 days for sentiment to subside, experienced no rebound, and had a 15 per cent probability of a secondary outbreak. Among the associated variables, the average event sensitivity was 75.2 (with a high proportion of ethical controversies), whilst the average public engagement was 62.5, exhibiting a phased evolution. Universities are simultaneously establishing long-term mechanisms through institutional and technical measures, integrating academic integrity education, implementing end-to-end evidence retention and early warning systems for research, collaborating with cyber security authorities to address misinformation, and fostering a symbiotic relationship between academic integrity and rational public discourse.

A comprehensive analysis of the aforementioned public opinion incidents at universities reveals that, whilst there are significant differences between the various cases in terms of the triggering events, the pace at which the incidents unfolded and the university’s response, there are also some commonalities. The results of the comparison of these university public opinion incident cases are shown in Table 5 below.

Table 5. Comparison of public opinion incidents at higher education institutions.

Both incidents of academic misconduct at universities were characterised by a rapid escalation of public sentiment. In the case of a university in Central China, the prolonged investigation period led to sustained high levels of public concern, which only subsided after the university announced its findings; in the case of a university in South-West China, as the issue involved questions of fairness, public interest rose gradually but subsided rapidly following prompt action. In the age of social media, public sentiment regarding universities spreads rapidly and attracts high levels of public engagement, with the focus of public opinion tending to shift towards deeper-rooted issues such as the academic ecosystem. A swift response and transparent handling by the university administration can effectively quell public sentiment and safeguard the institution’s reputation. Universities should establish long-term governance mechanisms through institutional and technological measures, strengthen education on academic integrity, implement full-process traceability and anomaly alerts for research, and collaborate with cyber security departments to address misinformation and guide rational public discourse, thereby reducing the incidence of academic misconduct-related public sentiment at its source.

Based on the evolution of the aforementioned case, this section examines the intrinsic logic underlying public opinion regarding academic misconduct in higher education institutions from the perspective of the core mechanisms of CAS theory, revealing the key driving factors behind the development of such public opinion from its inception to its resolution.

3.4.1. Mechanisms for the Emergence of Entity Clusters

The mechanism by which groups coalesce and emerge centres on shared identifiers as the core bond, with the scale of the group being positively correlated with the social resonance of these identifiers. In an incident at a university in Central China, 11 students formed a community of shared understanding; the topic triggered an “echo chamber” effect, and individual grievances were elevated into a call for reform of the academic system. An incident at a university in South-West China was triggered by academic scepticism; verification by netizens and the intervention of experts reinforced the consensus on academic fairness and shifted the focus towards institutional reflection. Both incidents demonstrate that shared identifiers drive the aggregation of individuals to form collective demands, propelling public discourse from isolated cases to a broader scrutiny of academic systems within higher education institutions.

3.4.2. Mechanisms of Non-Linear Interaction

At the heart of non-linear interaction lies the cognitive restructuring triggered by key information, the extent of which depends on the degree to which it resonates with latent social contradictions. In an incident at a university in Central China, a student’s exposure of a supervisor’s misconduct resonated with the public regarding the power imbalance between teachers and students, causing public opinion to shift towards a discussion of the educational ecosystem; in an incident at a university in South-West China, the controversy surrounding the age of a patent inventor, once interpreted, escalated public opinion into a critique of the academic system.

3.4.3. Mechanisms for the Circulatory Flow of Information

The smooth circulation of information determines the self-regulatory capacity of a public opinion monitoring system; any disruption to this circulation can easily lead to systemic imbalance. An incident at a university in Central China created a closed loop of information flow and a cycle of empathetic diffusion in sentiment, which sustained public interest in the matter. In an incident at a university in South-West China, the flow of resources formed a complete response cycle, enabling the university administration to swiftly investigate, make adjustments and resolve the conflict. These two types of incidents demonstrate that the smooth circulation of information, sentiment and resources is key to the self-regulation and effective governance of the public opinion system.

3.4.4. Self-Organising Adaptation Mechanisms

The core of self-organising adaptation lies in transforming ad hoc measures into long-term mechanisms; this is a key manifestation of the system’s resilience. A university in Central China adopted a tiered escalation model, with the university administration coordinating the handling of academic misconduct, establishing a standardised governance system covering the entire process, and utilising technology to enable proactive early warning. A university in the south-west of China has focused on adaptation through the restructuring of standards, revising academic regulations, introducing new recusal procedures, and implementing integrity reviews and accountability measures to form a complete closed-loop governance system. Both institutions have established long-term mechanisms through institutional development and technological support, strengthening the cultivation of academic integrity and guiding rational public discourse. This approach reduces the incidence of academic misconduct-related public opinion incidents at their source and enhances the adaptability and governance capabilities of the public opinion management system.

The core controllable parameter in this study is the intensity of intervention, which is calculated as a weighted composite with a 40 per cent weighting for response time and a 60 per cent weighting for the degree of information disclosure. The intensity of intervention is categorised into three simulation scenarios—low, medium and high—with the parameter ranges set out in Table 6 below.

Table 6. Parameter range settings.

The exogenous scenario parameters were set to the observed mean values from the two case studies: event sensitivity = 75.2 and public engagement = 62.5. These were used as constant exogenous inputs for the simulation; the simulation did not alter these two types of parameters, but only adjusted the intervention intensity variable to observe the output results for public opinion intensity, the magnitude of the backlash, recovery time and the probability of a secondary outbreak.

The simulation starts at time t = 1, with an initial public opinion intensity set at 50, derived from the average standardised intensity during the initial exposure phase of the two case studies; event sensitivity is set at 75.2 and remains constant throughout the simulation; The initial value of public engagement is set at 62.5; during the simulation, it will be dynamically updated according to the formula “Public engagement = Event sensitivity × 0.6 + Previous public opinion intensity × 0.4” in response to the state of public opinion; representative values for intervention intensity are selected for each scenario: 40 for low intervention, 60 for medium intervention and 80 for high intervention. The simulation tool used is Vensim PLE; the time step DT = 1 day, with the system state updated daily; the overall simulation period is 15 days, covering the entire process from the outbreak of public opinion to its gradual subsiding, and outputting time-series data on public opinion intensity for a total of 15 time points from t = 1 to t = 15.

This study utilises Vensim software to simulate the evolution of public sentiment under different intervention intensities, with the primary aim of quantifying the impact of intervention strategies on the resilience of public sentiment. The simulation experiments strictly adhere to core characteristics of the CAS theory, such as “flow cycles” and “non-linear interactions”. Intervention intensity is categorised into three levels—low (30 - 50), medium (51 - 70) and high (71 - 100). The simulation period was set at 15 days, with recovery time, rebound magnitude and the probability of a secondary public opinion outbreak selected as the core resilience indicators. The data on the evolution of public opinion intensity under different intervention intensities fully illustrate the entire trajectory of public opinion from outbreak, through its peak, to its subsidence. The data on the change in public opinion intensity over time under different intervention scenarios is shown in Figure 1 below.

Figure 1. Data showing changes in public opinion intensity over time under different intervention scenarios.

Recovery time shows a significant negative correlation with intervention intensity: at low intervention intensity, the peak value is 85.0, and it takes 5 days to return to the initial level; at medium intervention intensity, the peak value is 75.0, taking 3 days; and at high intervention intensity, the peak value is 70.0, taking just 1 day. A 3.0 rebound in sentiment intensity occurs with low intervention, whilst there is no significant rebound with medium or high intervention. The probability of a secondary public opinion outbreak, in order of increasing intervention intensity, was 35 per cent, 15 per cent and 5 per cent respectively.

The simulation reveals governance patterns: the marginal effect of intervention intensity on public opinion resilience is significant, with intervention costs being lowest and effectiveness optimal during the early stages of an outbreak; the value of information disclosure as a constraint outweighs that of response timeliness; and high-intensity intervention can simultaneously optimise the three key dimensions of resilience—resilience, disturbance resistance and adaptability.

Higher education institutions should focus on enhancing the adaptability of their public opinion management systems, establishing long-term mechanisms through institutional and technological measures, integrating academic integrity education into talent development, implementing end-to-end evidence retention and data early-warning systems for research, utilising official new media channels for science communication, and collaborating with cyber security authorities to address misinformation. By addressing the root causes, they can reduce the incidence of academic misconduct-related public opinion incidents and safeguard academic integrity and a healthy online ecosystem.

Combining the findings of the simulation with the core mechanisms of CAS theory, and with the aim of enhancing resilience, the governance strategy is developed across three dimensions: stakeholder coordination, response mechanisms and environmental development, with the primary focus on strengthening adaptive coordination capabilities through multi-stakeholder collaboration.

4.3.1. Enhancing the Capacity for Coordination amongst Stakeholders

With the aim of enhancing the resilience of the public opinion monitoring system, a highly efficient and standardised public opinion response framework should be established through coordinated action amongst diverse stakeholders, thereby breaking down information barriers and eliminating operational discrepancies. Universities may establish dedicated academic ethics investigation posts within their academic committees, staffed by specialists in academia and law, and establish regular coordination mechanisms with the public opinion task force of the public relations department to facilitate information sharing and the rapid transmission of instructions. Collaborative handling guidelines should be formulated covering the entire process of public opinion incidents relating to academic misconduct, clarifying requirements for stages such as information registration, verification and consultation, and enhancing handling efficiency through standardised procedures.

At the same time, regional research integrity organisations should be engaged to establish a case database, integrating typical cases and data to provide a reference for decision-making. Establish communication mechanisms with mainstream local media and academic self-media outlets to promptly release authoritative information in response to queries, and set up dedicated columns on official platforms to ensure the uniform and synchronised dissemination of information. Conduct regular training on public opinion response; through scenario simulations and case studies, enhance the professional capabilities and rational engagement of relevant personnel, staff and students, strengthen awareness of multi-stakeholder collaboration, and foster a framework for the governance of academic public opinion characterised by multi-party coordination and efficient collaboration.

4.3.2. Improving Mechanisms for Public Opinion Interaction and Resilient Response

Given the non-linear nature of public opinion evolution, a tiered response system centred on the intensity of public discourse must be established to ensure interventions are precisely matched to the prevailing public sentiment, thereby reducing the risk of backlash and enhancing the system’s resilience. Higher education institutions should collaborate with experts in academic ethics, public opinion and administrative staff to formulate detailed grading criteria for academic misconduct-related public opinion incidents, based on factors such as sensitivity, scope of dissemination and intensity of public discourse. Resources should be allocated and responsibilities clearly defined according to these levels; for high-level incidents, independent external experts should be engaged to conduct verification, thereby enhancing professionalism and credibility.

At the same time, a dynamic assessment system for the intensity of intervention should be established. Taking into account the type of incident, platform characteristics and public engagement, this system should provide reasonable recommendations regarding the disclosure of information and the timeliness of responses, thereby forming differentiated handling plans. An effectiveness evaluation module should be incorporated into the monitoring system to track information disclosure, shifts in sentiment and the emergence of related topics in real time, enabling timely strategy adjustments. Once the public opinion situation has subsided, a cross-departmental debriefing should be conducted to analyse issues, identify potential triggers for a resurgence, optimise the tiered response guidelines, and incorporate the effectiveness of the response into performance assessments. This will establish a closed-loop management cycle of “response - evaluation - optimisation”, continuously enhancing the resilience and precision of universities’ responses to public opinion regarding academic misconduct.

4.3.3. Fostering a Symbiotic Environment of Academic Integrity and Rational Public Discourse

With a focus on enhancing the adaptability of the public opinion monitoring system, establish a long-term governance mechanism through institutional development and technical support, strengthen the cultivation of academic integrity, and reduce the likelihood of academic misconduct-related public opinion incidents in higher education institutions at source. Higher education institutions will integrate academic integrity education into the entire talent development process, incorporating relevant courses into undergraduate and postgraduate programmes, and enhancing students’ academic integrity and compliance capabilities through case-based teaching and thematic seminars.

At the research level, a mechanism for end-to-end recording and evidence preservation will be implemented. Relying on technology, traceability will be ensured across all stages—including experiments, data and research outputs—and an early warning system for data anomalies will be established to prevent academic misconduct through technical means. Official new media platforms will be utilised to launch dedicated columns on research transparency, promote academic standards, and open interactive channels to address queries and clarify doubts. At the same time, universities will collaborate with local cyberspace administration departments to combat academic rumours, establish rapid response channels for handling false information, guide rational public discourse through professional analysis, and maintain a healthy online ecosystem, thereby fostering a virtuous cycle in which the promotion of academic integrity and the cultivation of rational public discourse mutually reinforce one another.

Taking into account the theoretical evolution of CAS and the practical requirements of governance, regulatory recommendations centre on multi-stakeholder collaboration and end-to-end standardisation, with a focus on refining institutional frameworks and enhancing effectiveness.

4.4.1. Recommendations for Regulatory Authorities

Regulatory authorities should, based on the complex adaptive nature of public opinion systems as outlined by CAS theory, establish an institutionalised, collaborative and technology-driven regulatory framework. Regulations governing the monitoring of public opinion regarding academic misconduct in higher education institutions need to be refined, with the responsibilities of all parties and accountability mechanisms clearly defined; resilience indicators such as the level of information disclosure and response timeliness should be incorporated into the performance assessment of higher education institutions. A cross-departmental coordination mechanism should be established to integrate resources from departments such as education and cyberspace administration, and a national public opinion sharing platform should be set up to facilitate the exchange of leads and data, as well as coordinated handling. Relying on big data and artificial intelligence, an intelligent monitoring and early-warning system should be developed to focus on highly sensitive issues, enabling real-time tracking of relevant public opinion, tiered early warnings, and early intervention at high-risk junctures. At the same time, conduct routine supervision and inspection, spot-checking the implementation of academic integrity education and the documentation of public opinion handling at universities, promote successful governance practices, and encourage universities to shift from a reactive to a proactive approach, thereby strengthening the institutional defences of the academic integrity ecosystem.

4.4.2. Recommendations for Internet Users

As key participants in public discourse, internet users should enhance their awareness of rational participation and their level of academic understanding, working together to maintain a healthy online public opinion environment. It is recommended that, when following public opinion regarding academic misconduct in higher education institutions, netizens maintain a rational judgement, refrain from blindly following trends or spreading unverified information, and consciously resist academic rumours and emotional rhetoric. They should proactively learn about academic norms and research integrity, enhance their ability to identify academic misconduct, view public opinion incidents objectively, and avoid being misled by one-sided perspectives. At the same time, they should cultivate a sense of responsibility, express their views rationally, and focus on deeper issues such as the optimisation of academic systems and the improvement of talent development. They should avoid personal attacks and extreme rhetoric, and channel their concerns through appropriate channels to actively contribute to the rational evolution of public discourse and the resolution of issues, thereby jointly promoting the development of an ecosystem of academic integrity in higher education.

4.4.3. Recommendations for the Media

The media should fulfil their central role in processing and disseminating public opinion information, upholding professional standards and social responsibility to establish an accurate and rational model of communication. When reporting on incidents of academic misconduct in higher education institutions, they must rigorously verify information sources, utilise authoritative first-hand materials, and report objectively using a model of “factual presentation + professional interpretation + institutional reflection”. They should avoid emotional sensationalism and one-sided interpretations, focusing instead on deeper issues such as the optimisation of academic systems and the improvement of oversight mechanisms, thereby guiding the public towards rational thinking. At the same time, the media should establish regular communication channels with universities to obtain authoritative information in a timely manner and correct any inaccuracies in their reporting. They should proactively fulfil their duties of public opinion oversight and guidance, promote the principles of academic integrity, dispel misconceptions among the public, and foster a favourable public opinion environment conducive to the rational development and effective management of public discourse surrounding academic misconduct in higher education.