The study aims to address Chinese universities’ image repair strategies after network public opinion events in the field of crisis management; therefore, it takes 43 network public opinion events in Chinese universities as the research object, encodes the official texts issued by universities according to the image restoration strategy, and sums up the image repair strategies commonly used by Chinese universities. Then, natural language processing is used to conduct the sentiment analysis of the online comments obtained. Accordingly, the sentiment index is constructed to evaluate the effect of Chinese universities’ image repair strategies. We find that Chinese universities commonly use the image repair strategy combination of bolstering, provocation, and corrective action; they have not used the apology strategy commonly used in western discourse systems. We also find that the complete information release process has a better image repair effect, particularly in teachers’ lapse and personal safety events. The sentiment index in teachers’ lapse events is the highest and is related to the universities’ corrective actions. The sentiment index in different public opinion hot events is quite different, which may be related to the nature of specific events. In personal safety events, netizens are more satisfied with image repair strategies.

1. Introduction

In the new development period, Chinese universities shoulder the responsibility of personnel training, scientific research, social service, cultural inheritance, and innovation. University's image is the overall impression and evaluation of a university’s internal quality and external performance in the public’s eyes. Like enterprises, universities pay more and more attention to their image design and brand construction [1]. They shape their image by continuously improving academic, educational, and scientific research qualities [2] and teachers’ quality [3]. Moreover, a good university image enhances students’ loyalty and satisfaction [4] and affects alumni’s sense of identity [5]. Therefore, under the background of deepening the reform and development of China’s higher education, a good image has become an essential intangible resource for the survival and development of Chinese universities, which helps to enhance their core competitiveness.

In recent years, academic misconduct, teachers’ lapse, students’ unnatural death, and other emergencies frequently occurred in universities. Under the network background, the negative news of universities reported by the media and the network public opinion generated from social platforms further amplify the harm of emergencies in universities. The negative comments would seriously damage the image of universities and thus would affect the application for admission of the universities involved and their training effect and their long-term development [6]. Furthermore, with the popularity of smartphones and the development of the network, social media has become an essential channel for netizens to obtain news and a part of many people’s daily life [7]. In China, social media such as Weibo and WeChat have become the source of network public opinion, influencing network public opinion and resolving public opinion crises. Therefore, when facing crises, scandals or emergencies, universities will choose effective social media communication strategies and take the initiative to speak out via social media, thus maintaining their reputation and image [8].

At present, the research on university image repair mainly focuses on English-speaking countries. These studies are based on crisis communication strategies (CCSs) and the image repair theory proposed by Coombs [912], Benoit [1316], and other scholars. They examine how a crisis is appropriately resolved through effective language and behaviour. In terms of the research object, the previous studies focus on how universities use crisis communication and image repair strategies to eliminate the adverse effects of scandals such as sexual assault, corruption, and academic affairs in some famous western universities (e.g., [1719]). However, the research object is relatively not diversified and the research conclusion based on the cultural background of western crisis communication lacks a case test from the Chinese cultural background. In terms of research methods, most of the studies are case studies of a specific university (e.g., [17, 18, 20]). They collect the texts published by the universities on social platforms and netizens’ comments to measure the public’s response to the university image repair strategy. Nevertheless, there is a lack of discussion on the general tendency of the image repair strategy selection in various event types, and the research conclusions are not universally applicable.

Therefore, this study selects 43 network public opinion events that occurred in Chinese universities from 2019 to 2021, that had a great impact on the network platforms, involving academic misconduct, teachers’ lapse, public opinion hot topics, campus safety, and other events. By analysing the official texts issued by Chinese universities, this study attempts to sum up the image repair strategies commonly used by Chinese universities. At the same time, through the sentiment analysis of the online comments obtained by the web crawler and the sentiment index, this study measures netizens’ responses to evaluate the effect of the universities’ image repair strategies.

2. Literature Review

2.1. Crisis Management and Network Public Opinions on Universities

Generally speaking, China and the West differ in their research studies on public opinion events in universities. Western scholars often use terms such as “scandal events” and “crisis events” to refer to network public opinion events. Therefore, when studying the governance of public opinion in universities, they mainly focus on “crisis management” after the “crisis event.” There are studies on universities’ coping with external risks, such as earthquake risk [21] and university risk management mechanisms [22], as well as internal risks, such as sexual assault and corruption scandals [6, 17, 19], and the crisis of racism in higher education institutions [23, 24]. These studies adopt the case study method and analyse a specific event in an university and its countermeasures. Previous studies have confirmed that it is very difficult to rebuild broken trust in educational institutions [25].

Chinese scholars put forward the term “network public opinions,” which refers to media’s or netizens’ influential and tendentious opinions or comments published online about a particular focus issue, public social affairs, and so on [26]. With the popularization of smartphones and mobile Internet technology, some offline events will likely trigger large-scale network public opinions and a public opinion crisis. The outbreak of a public opinion crisis results from the comprehensive action of endogenous factors such as the destructive power, publicity, and social sensitivity of issues and events, and exogenous factors such as media driving force and government regulation [27]. Therefore, generally speaking, the network public opinion crises almost all go through the process of “offline events-online public opinions-public opinion crisis.” In this process, the offline focus event, the online dissemination of the event, and the public opinion coping strategies will all affect the development and solution of the public opinion crisis.

University network public opinions are the network public opinions on universities. This term was initially proposed by Chinese scholars who conducted relevant studies on the unique cultural background of China (e.g., [28, 29]). Due to the highly developed Internet features, the network of public opinions on Chinese universities is characterized by social media disclosure, accelerated dissemination speed, diversified disseminators, and diversified dissemination methods [30]. At present, Chinese scholars mainly focus on the causes (e.g., [31]), dissemination mechanism (e.g., [32]), problems, and counter-measures (e.g., [33]) of the network's public opinions on universities. Therefore, be it “crisis management” or “public opinion governance” in universities, its fundamental purpose is the same, which is to establish a good image and restore a good reputation.

2.2. Social Media as a Tool for Image Repair and Crisis Management

In the Internet age, social media has become an essential tool for crisis management and image repair. Scholars collect the press releases, posts, and related comments of a particular case published on social media, use the content analysis method to analyse the topics involved in the case, the image repair strategies, and the public’s cognition and sentiment, to measure the public’s response to the case and the evaluation of the image repair strategies [18, 20, 34, 35]. Compared with traditional media such as online newspapers, the most significant advantage of using social media for image repair and crisis management is obtaining a higher reputation and less secondary crisis response [36]. It is found that at present, it is mainly organizations or public figures who post information on YouTube, Facebook, or Twitter to clarify the truth or to publicly apologize after a crisis [37, 38]. The effective use of image repair strategies can reduce the severity of the incident [8], which is highly beneficial to the image repair of organizations or individuals. However, when European and American universities use social media for image repair and crisis management, they also need to pay attention to the crisis response from fan groups on social media such as Twitter or Facebook. This is often mentioned in the research on European and American universities’ sports crisis management. Previous studies found that fans not only publicly support and defend their players and teams after a crisis, but also take the initiative to repair their players’ and teams’ image thus relieving the enormous pressure caused by players’ violations by externalizing responsibility to other institutions, adopting certain image repair strategies on behalf of the team, and other ways [3941].

In China, social media is an essential channel for netizens to obtain news and information. Weibo is one of the most well-known social media in China. Its hot search function serves as a weather vane for public opinions to a certain extent, reflecting the hot topics and issues of public concern in Chinese society. When public emergencies occur, Weibo changes from a simple social platform to a source of network public opinions. Thus, it may influence the network's public opinions on public emergencies and can channel negative sentiments. For example, Weibo is widely used in enterprises’ for resolving brand crises [42] and government crisis communication and strategy choice [43, 44] and has achieved good results. Therefore, this study attempts to use data mining methods to obtain Weibo users’ comments after a university public opinion event, and construct a sentiment index through sentiment analysis, to evaluate the effect of Chinese universities’ image repair strategies.

2.3. Image Repair Theory

One aspect of crisis management is crisis communication which gradually plays a role in coordinating the relationship between stakeholders and enterprises in promoting effective communication. Crisis communication means the communication between an organization and the public before, during, and after a crisis. Its primary function is to influence public opinion and reduce the damage to the organization’s reputation [45]. With the introduction of rhetoric theories to crisis communication research in the 1990s, Benoit’s image repair theory and Coombs’ crisis response strategy have emerged and they have become essential measures for organizations to carry out crisis management.

Coombs [10] has put forward seven crisis response strategies, including attacking the accuser, denial, excuse, justification, ingratiation, corrective action, and full apology. Subsequently, the situational crisis communication theory (SCCT) was formed. The foundation of the crisis response strategy is the image repair theory put forward by Benoit [14]. Based on the discourse theory of image repair, this theory includes more specific image restoration strategies, which are organizations’ responses after being questioned by the public or the image damaged. Image restoration strategies include denial, evasion of responsibility, reducing offension of event, corrective action, and mortification (Table 1).

At present, the image repair theory has been widely used in response to corporate reputation restoration [46], handling university sexual assault scandals [17], response to medical malpractice [47], government image and international image building in crisis [48, 49], and other fields. The research shows that appropriate image restoration strategies can obtain positive responses and support from the public, and help companies, enterprises, universities, governments, and other organizations to restore trust during the crisis, thus minimizing the impact of the crisis. Therefore, this study attempts to use Benoit’s image repair theory to analyse the image repair strategies commonly used by Chinese universities in dealing with public opinion events.

3. Materials and Methods

3.1. Research Design

In China, most universities have opened an official account on Weibo, mainly used for campus news releases and daily publicity. Besides Weibo, the official account on WeChat and the campus website are also used by some universities to publish authoritative information. However, compared with the WeChat accounts and campus websites of universities, official Weibo posts are relatively more interactive and can obtain netizens’ opinion feedback, attitudes, and information needs. Therefore, by analysing the online comments on universities’ Weibo posts, we can identify netizens’ sentiments and can measure the effectiveness of the image repair strategies.

Therefore, this study attempts to answer the following research questions:(1)What are the image repair strategies commonly used by Chinese universities?(2)How is the sentiment analysis of online comments used to measure the effect of university image repair strategies?

Based on the network public opinion events that happened in Chinese universities, this study obtained the texts published by universities on their official platforms and the online comments of Weibo users through manual retrieval and web crawler, respectively. Then, based on the text encoding data, the strategies used by Chinese universities in image repair are summarized and analysed accordingly. Meanwhile, according to the sentiment analysis of online comments, the sentiment index is constructed to evaluate the effect of universities’ image repair strategies (See Figure 1).

3.2. Data Collection

This study collected 43 network public opinion events in 39 Chinese universities from February 1, 2019, to June 30, 2021. First, to analyse the image repair strategies, this study collected the universities’ 58 texts published on Weibo, WeChat, and campus websites through manual retrieval. These text forms include briefings, notifications, and disposition results. Then, to investigate netizens’ responses to the universities’ image repair strategies, this study employed natural language processing to conduct sentiment analysis of the online comments. However, 13 universities in the case database issued texts only on WeChat and campus websites, where the comment data were inaccessible, and 26 universities used the Weibo platform where the comment data were accessible, although limited due to the default setup of Weibo. Therefore, this study employed the web crawler software “GooSeeker” and finally obtained 14,210 Weibo comments on 26 universities’ and 39 Weibo posts.

In China’s Higher Education Public Opinion Report (2019), China’s higher education network of public opinions were divided into nine categories [50], such as policy and management, quality evaluation, teacher development, and college students. Accordingly, this study divided the collected 43 university network public opinion events into four categories: academic misconduct, teachers’ lapse, personal safety, and public opinion hot events. The definitions of each category are as follows:(1)Academic misconduct: this includes university teaching and research personnel, management personnel, and students’ violation of recognized academic standards, and academic integrity in scientific research and related activities, including plagiarism and falsification of papers(2)Teachers’ lapse: it includes university teachers’ violation of professional ethics and fundamental principles, such as harassing students, engaging in malpractices for personal gain, and abusing students, which is to be severely punished in China(3)Personal safety: this includes incidents that occur in universities and affect the personal safety of teachers and students, such as accidental death, food safety, and infectious diseases(4)Public opinion hot events: these are the events that occur in universities which quickly become the focus of public attention and cause heated discussion after media reports or exposure, including the management of overseas students and inappropriate remarks of teachers and students

The proportion of the four types of events is shown in Figure 2:

3.3. Image Repair Strategies: Data Encoding

To analyse the specific strategies used by Chinese universities in image repair, this study collected 58 texts which were officially released by universities nvolved in 43 cases of network public opinion events. Because it takes time to investigate and obtain evidence, it is generally necessary for universities to deliver a briefing and/or notification before issuing the disposition result, indicating that the universities are aware of the public opinion event and have started the investigation (as in cases 8/9/16). Nevertheless, some universities combine the briefing and notification with the disposition result in one text. That is, the final disposition of the case is formed after the complete investigation process (as in cases 18/27/41). This study encoded the 58 Chinese texts according to the image restoration strategies described by Benoit [14], and the examples of codes are shown in Table 2.

The specific process of encoding is as follows: First, the author and the three master students conducted a pilot encoding based on the comprehensive analysis of the image restoration strategy as per the examples given by Benoit. Then, the four people randomly encoded five texts and rechecked them to unify the encoding standard. Next, the author and the first two master students independently encoded all 58 Chinese texts, and the similarity rate exceeded 87%. Then, the third master student rechecked and recoded the controversial encoding parts and improved the similarity rate of text encoding results to 95%. Finally, the encoding results were determined by the author and the three master students after consultation and after they were tested by random sampling.

3.4. Effect Evaluation: Sentiment Analysis and the Sentiment Index

In the Internet age, user-generated content (UGC) on social platforms has become an essential source of information. A typical example of user-generated content is the online comments on e-commerce platforms. Computer natural language processing (“NLP” hereafter) can conduct the sentiment analysis of the obtained online comments to help evaluate the comments’ sentiment polarity (positive or negative). As the sentiments reflected in online comments will affect the public’s attitude, the more positive the public comments are, the more positive the sentiments will be. Sentiment analysis is thus considered an effective method to collect opinions [51]. In order to successfully manage public opinion and crisis in colleges and universities, it is necessary to examine public sentiment and other sentiment-related reactions [52]. Therefore, this study uses NLP to conduct sentiment analysis of the online comments obtained. The results can reflect, to some extent, the genuine attitude of netizens towards the universities’ image repair to evaluate the image repair effect.

In this study, the sentiment analysis of online comments is realized by using Baidu AI open platform’s NLP technology. Baidu's NLP technology contains a sentiment tendency analysis module that provides two interfaces: general and customized models. The former draws on the subjective information text of a single subject. The sentiment tendency is automatically evaluated and the corresponding confidence level is given. Thus, users can realize “plug and play.” The latter trains and optimizes the former by uploading positive and negative sentiment corpus of specific application scenarios, thus improving the analysis accuracy of the former. Currently, this study chooses the latter model of sentiment tendency analysis as it carries out the sentiment analysis of Weibo comments with higher accuracy through supervised machine learning. The specific operation steps are as follows:

First, according to the operation process of Baidu's AI open platform, a new exclusive sentiment analysis task was created in the NLP module under “product-service.”

Second, under the “personalized customization” taskbar, the model was created, and the training and the generation model were carried out in turn in the customization process. As the platform requires that each training corpus should be over 1,000 to ensure the customization effect, so three groups of 1000, 1100, and 1300 positive and negative comments were manually marked out from the collected 14210 comments. In the training process, the platform selected 10% of the training data by default to evaluate the model and gave the model accuracy before and after training. The model with the highest model accuracy was selected for use after the training was completed.

Finally, comparing the results of the three training models (shown in Table 3), we selected Model 2 with the highest accuracy to analyse the sentiment tendency.

Online comments under a single Weibo post were arranged mainly in two ways: by heat and by time. The online comments collected in this study were arranged by heat. The main reasons include the following: (1) To arrange the online comments by heat is the default setting of Weibo. It prioritizes the online comments with the highest number of netizens’ likes and (2) Studies (e.g., [53]) have found that biased comments in the front will affect users’ attitude to comment later. Therefore, to a certain extent, the number of online comments can reflect the personal attitude of netizens, that is, the more comments, the higher the ranking, and the more likely it is to affect the comments of netizens behind. Furthermore, users’ online comment behaviour also obeys power-law distribution [54]. Thus, this study selected the top 50 Weibo comments with the most considerable number of likes to analyse the sentiment tendency and then constructed the sentiment index to evaluate Weibo users’ responses to universities’ official posts.

Thus, the sentiment index is constructed and the results of a single post are as follows:(1)Use the screening and sorting functionality in an Excel document to obtain the top 50 Weibo comments with the most considerable number of likes under a single post.(2)Use the Python code derived from the Model 2 interface to calculate the sentiment tendency probability of a single post one by one. The probability value range from (0 to 1), and a probability value of 0.5 is considered as neutral. When the value goes over 0.5, it indicates that the probability of positive sentiment is higher, regarded as a positive comment; and when the value <0.5, it indicates that the probability of negative sentiment is higher, regarded as a negative comment.(3)Construct and calculated the online comment sentiment index of a single post. Here, we consider both the weight of the number of likes of each comment among the top 50 and the sentiment polarity of a single post, that is, positive (+) and negative (−). (a) We then calculate the proportion of likes for each comment by using the formula (1). (b) Because the number of likes among the top 50 comments differs widely, we use formula (2) to standardize the results in Step (a) to reduce the difference gradient. (c) Multiply the sentiment probability value of each comment calculated in Step (b) with the weight after standardization processing. The polarity of the product is distinguished by positive and negative, that is, when the value of the calculation result is >0.5 it is considered as positive, and when the value of the calculation result is <0.5, it is considered as negative. (d) Sum up the values of the top 50 online comments to obtain the sentiment index of the online comments of a single post. The range of the sentiment index is (−∞, + ∞). The larger the value, the higher the positive attribute of sentiment. The calculation results of the sentiment index are shown in Table 4.

4. Results and Discussion

4.1. Chinese Universities’ Image Repair Strategies

The current analysis shows that, in releasing information on the official platform after public opinion events, the image repair strategies Chinese universities commonly use are reducing offension of the event (44.1%), evasion of responsibility (31.68%), and corrective action (23.6%). On the other hand, they seldom use denial (0.62%) and never use mortification (0%) (Table 5). Specifically, Chinese universities frequently employ the image repair strategy combination of bolstering (29.19%), provocation (24.84%), and corrective action (23.60%).

Bolstering refers to reducing the public’s negative feelings about the institution by describing the positive characteristics or behaviours of the past [14]. In the official texts released by Chinese universities, bolstering is reflected more in a positive and serious attitude. It implies that the case will be investigated and dealt with according to some functional departments’ relevant regulations and procedures, which have a strong administrative colour. The provocation is a public response to real-name reporting and netizens’ questioning. It also embodies the official attitude of Chinese universities. The corrective action means a kind of error correction behaviour of university officials after public opinion events. It is also the result of events about which the public is most concerned. It affects the development of public opinion events through forwarding, commenting, and praising on social platforms [55]and is an essential strategy for image repair.

The apology strategy has never been used in image repair by Chinese universities. Similarly, most Chinese companies avoid public apologies and save face by keeping silent [56]. In contrast, an apology is often considered the most effective image repair strategy in the eyes of western scholars. Stamato ([57]: 397) maintains that “By finding the right expression and circumstance, and by acknowledging the wrongs done and the harm caused, an apology can serve as an instrument of reconciliation, and this help to create the conditions for seeking a just and sustainable peace.” However, Compton ([58]: 357) points out that “parts of the image repair effort seem to emphasize that the apologists did not really mean the apology in the first place,” and thus it may be “apologizing for an apology.” The difference between China and the West in choosing the apology strategy mainly originates from cultural differences. China’s political culture places more emphasis on the government’s ability to control and deal with the aftermath of an event. However, an apology often reduces the government’s credibility in handling emergent public events, so the apology strategy is often regarded as the last resort [59]. Although China’s political culture does not use the apology strategy in image repair, it pays more effort in taking practical actions to recover losses after the crisis, that is, actions speak for themselves. Therefore, Chinese universities avoid using the apology strategy, either based on the connotation of apology in Chinese culture or to maintain their credibility.

For different types of public opinion events, the emphasis of Chinese universities' image repair strategies is also different (Table 6).(1)In the public opinion events of academic misconduct, Chinese universities choose the image repair strategies of provocation, bolstering, and correct behavior more often. Correspondingly, all universities choose to issue their disposition results to deal with the teachers and students involved in academic misconduct to varying degrees. The provocation to reports from various channels, supported by the punishment provisions of academic misconduct in schools and then adopting the strategy combination of punishment and corrective action, fully reflects Chinese universities’ efforts to handle academic misconduct.(2)In the public opinion hot events, Chinese universities also choose the image repair strategy combination of provocation, bolstering, and corrective action. This combination of strategies is realized more through official notifications. Specifically, the universities involved clarify the truth, announce corrective measures, reasonably respond to hot public discussions, support relevant regulations and procedures, and then issue specific corrective actions in the notifications.(3)In personal safety public opinion events, most Chinese universities have adopted the image repair strategy combination of bolstering and minimization. This strategic combination is supported by the disposition results released by authoritative organizations, which downplays the adverse consequences of such events, emphasizes the irresistibility of accidents, and finally issues official notifications.(4)In the public opinion events of teachers’ lapses, the vast majority of Chinese universities have adopted the image repair strategy combination of provocation, bolstering, and corrective action. They issue separate notifications or disposition results to repair images. Specifically, the universities involved give a reasonable response to various reports and support the severe punishment of teachers’ lapses with various regulations of the state and schools. This strategy combination shows Chinese universities’ zero-tolerance attitude towards teachers’ lapses.

4.2. Online-Comment-Based Effect Evaluation of Image Repair Strategies

In this study, the sentiment analysis and the sentiment index of online comments on Weibo are used to measure the effect of Chinese universities’ image repair strategies. The current findings are as follows:(1)When the universities involved release only notifications but no disposition results, the sentiment index of the public opinion events is lower. In contrast, the sentiment index significantly improves, and the image repair effect is better when the universities release notifications first and disposition results after. This practice conforms to the general cognition of the public and reflects the serious attitude that universities should have. Therefore, it is conducive to reshaping universities’ images and can gain strong support from alumni, students, and other relevant parties in universities [60]. This phenomenon is most apparent in the events of teachers’ lapses and personal safety (Table 7).(2)The sentiment index of teachers’ lapse events is higher than that of other types of events. This result is associated with the corrective action of universities. To the public, university teachers are the highly educated class in society, so the public has higher requirements for university teachers’ moral standards. Therefore, they think teachers’ lapses should be avoided and thus pay more attention to such events. As a result, all universities use corrective action to repair images of such events, either dismissing the teacher involved or giving severe punishment. This strategy reflects the official attitude of zero tolerance for teachers’ lapse. It also verifies the research conclusion of Claeys et al. [61] that the extremely severe and almost reconstructive punishment measures for such incidents (such as dismissal) have a tremendously positive influence on image repair. Therefore, the more severe the universities’ corrective action in teachers’ lapse events, the higher the sentiment index of online comments (Table 8).(3)There are significant differences in sentiment indexes among different public opinion hot events. This result may be related to the nature of a specific event and should be analysed on a case-by-case basis. For example, in Case 13, the damage to the university’s image was purely caused by the media’s errors which had nothing to do with the universities themselves. Thus, the sentiment index of netizens’ online comments is the highest. In contrast, Case 29 involved Chinese people’s patriotic sentiment. Although the university issued detailed notifications and specific corrective actions, the image repair effect was still unsatisfactory, and the sentiment index was the lowest (Table 9).(4)The sentiment index in personal safety public opinion events is all positive, indicating that netizens are satisfied with the image repair strategies. This is because such incidents are related to the life safety of teachers and students in school. Moreover, the official texts of the universities involved are generally released based on the authoritative investigation results of government agencies with absolute credibility, that is, the public security organs. Therefore, this image repair strategy has high authority and credibility, and the sentiment index of corresponding cases is also more positive, and the image repair effect is better (Table 10).

5. Conclusions

This study examines 43 cases of network public opinion events in Chinese universities. First, it draws on Benoit’s image restoration strategy to analyse the universities’ image repair strategies used in their official texts. Then it constructs the sentiment index based on the sentiment analysis of Weibo users’ online comments to discuss the universities’ image repair effect in public opinion events. This study’s findings are as follows:(1)Chinese universities commonly use the image repair strategy combination of bolstering, provocation, and corrective action. The strategy combination has a positive effect on responding to public queries. On the other hand, Chinese universities have not used the apology strategy commonly used in western discourse systems.(2)Through the sentiment analysis and the sentiment index to evaluate the image repair strategy effect, we find that the public’s response to different types of university network public opinion events varies. Complete information release processes including briefings, notifications, and disposition results have a better image repair effect, which is most apparent in teachers’ lapse and personal safety events. The sentiment index in teachers’ lapse events is higher than that of other types of events and is related to the universities’ corrective actions. The sentiment index in different public opinion hot events is quite different, which may be related to the nature of specific events. In personal safety public opinion events, netizens are more satisfied with the image repair strategies.

Previous studies mainly focus on the image repair strategies applied in enterprises and governments’ crisis management, and most of them adopt case study methods to discuss the image repair strategies selected in a specific case. This study takes Chinese universities as the research subject and thus extends the application scope of the image repair strategy. At the same time, to evaluate the effect of the universities’ image repair strategies, the study employs NLP to carry out the sentiment analysis of Weibo users’ online comments and constructs the sentiment index accordingly. This research method may provide a reference in the strategy effect evaluation for relevant research studies.

The limitation of this study mainly lies in the limited number of online comments in the database. Due to the system setup of the Weibo platform, the study failed to obtain comments that are set as unable to comment or invisible. It is hoped that a more effective method to obtain online comment data may be explored to enlarge the database in the follow-up research of university network public opinion events.

Data Availability

The data used in this paper were obtained from the Sina Weibo platform by using the crawler named GooSeeker. The web crawler GooSeeker (http://www.gooseeker.com) is a free web page data capture tool in China, with strong usability and operability. The study collected 43 network public opinion events in 39 Chinese universities from February 1, 2019, to June 30, 2021. All of these cases have had a significant impact on China’s online platforms, and detailed information can be obtained through Baidu search engine (http://www.baidu.com). However, for privacy reasons, these cases and the universities involved will not be provided as attachments.

Conflicts of Interest

The authors declare that they have no conflicts of interest.


This article was supported by the National Social Science Foundation of China (No. 17CGL074).