The data used in the study collected from the Web of Science core collection database. The database is an important database for obtaining global academic information. The target papers were searched in the core collection of Web of Science on Dec. 31, 2019. The search string “TS = ((high school OR middle school OR secondary school OR senior high school OR senior middle school OR junior high school OR junior middle school) and (cyberbullying OR teenager cyberbullying))” was used to obtain the target literatures. TS stands for topic, which means searching the literature in the form of topic. Here 654 records are shown on the screen (See Figure 1). Based on the focus of research on the development of

this field, the document type was limited to “Article”, and the search period was unlimited. The language was English. Then, the irrelevant papers and less representative record types are excluded, such as news, biography, advertisements, notices and press releases. As a result, 535 relevant papers were found from WoS.

To ensure that the 535 articles retrieved were not exist duplicated and irrelevant, all the literatures were screened by reading the abstract and full text, and a total of 527 target literatures were finally obtained. Bicomb (Bibliographic Item Co-Occurrence Matrix Builder) was used to sort out the keywords of 527 papers, and constructed a text matrix and co-word matrix of high-frequency keywords. Then SPSS 20.0 was used for cluster analysis to reveal the hotspots and trends of the research on teenagers’ cyberbullying in middle and high schools. Finally, the representative articles were selected for content analysis.

As shown in Figure 2, the research related to the middle and high school students’ cyberbullying began to attract the attention of scholars in the early of 20^th century. The first relevant paper was published in 2006. From 2006 to 2011, the average annual number of papers published was no more than 10. The research on teenagers’ cyberbullying in middle and high schools developed slowly in this

period. The number of papers published in 2012 increased rapidly compared with that in 2011, indicating that the problem of teenagers’ cyberbullying has gradually attracted attention since 2011, which is a developing period for the research of teenagers cyberbullying. The number of articles published in 2015 showed a significant increase compared with 2014, indicating that research on cyberbullying among middle and high school students has received widespread attention. There were abundant papers on cyberbullying among the middle and high school students published in 2019 than previous year.

Keywords represent the core content of an article. By analyzing and summarizing the key words in the literature of a certain field, the research hotspots in this field can be revealed to some extent. The first 26 keywords were selected as high-frequency keywords (see Table 1), which were with frequency > 11, and the cumulative frequency was 48.4%, which met the analysis criteria. As represented in Table 1, except for “cyberbullying”, the top ten keywords: “bullying”, “adolescents”, “cyber victimization”, “adolescence”, “victimization”, “bystander”, “victims”, “perpetration”, “gender”, and “traditional bullying”. This result indicated that the research on teenagers cyberbullying mostly focus on the forms and gender of bullying, while the deeper relationship between keywords needs further data mining.

Cluster analysis plays a role of dimensionality reduction, and data with similar topics can be formed into a group. The data within the group are highly similar, while the differences between the groups are large. The aim of cluster analysis is to segregate groups with similar traits and assign them into clusters. After analysis, the clustering analysis chart of high frequency keywords of cyberbullying is obtained (see Figure 3). The keywords were classified into four topics: 1) definitions and characteristics of cyberbullying, which mainly involve cyberbullying,

bullying, cyber victimization, and victimization, etc.; 2) mental health problems caused by cyberbullying mainly involve suicide, suicide ideation, and depression, etc.; 3) strategies of cyberbullying mainly include intervention, coping strategies, and prevention, etc.; 4) the relationship between gender difference and cyberbullying, which mainly include gender difference, gender, etc.

Home and school are the main places in the life of middle and high school students, and the social environment their contact with is exposed to the network environment. However, the middle and high school students who are in adolescence are susceptible to the temptation and influence of external factors. Therefore, to create a relatively healthy and harmonious external environment is conducive to students to establish a correct concept of network security.

Build trust between parents and children, and promote empathy in children. Previous studies have found that parents of cyberbulliers were often prone to use punitive or severe parenting methods to supervise children through physical punishment, filtering friends, and monitoring the Internet ( Kalmus, Blinka, & Lafsson, 2015

Schools are the most relevant environment for the social development of teenagers. From the perspective of school prevention, it is important to ensure that schools have sound anti-bullying norms and introduce anti-cyberbullying programs or content into the classroom to raise the anti-bullying awareness of students. For example, the task-oriented WebQuest course immediately and effectively enhanced the knowledge of cyberbullying, reduced the intentions, and retained the effects after the learning ( Ming-Shinn, Wu, Leif, Koustuv, & Scott, 2013). Some scholars believe that the prevention of cyberbullying should combine the traditional school bullying and cyberbullying, and design a video game (https://www.cybereduca.com/) ( Landazabal & Martínez-Valderrey, 2015; Maite & Martínez-Valderrey, 2018 ). This game covering five topics: 1) network phenomenon, 2) computer technology and security, 3) cybersexual behavior, 4) the consequences of cyberbullying, and 5) dealing with cyberbullying. After a year of empirical research, the results showed that students’ empathy, self-esteem and interpersonal conflict resolution ability were significantly improved, and the number of both types of aggressive behaviors was reduced. In addition, the gender difference in cyberbullying is related to age. Schools should develop personalized measures based on the characteristics of students, such as using the Pestkoppenstoppen online project, tailoring intervention measures for cyberbullying based on the personal characteristics of participants, teaching young people how to cope with cyberbullying more effectively (Jacobs, VoLlink, Dehue, & Lechner, 2014).

Relying on media technology, cyberbullying information is widely disseminated, which can strengthen the transmission of monitoring information, curb the circulation of bullying content, effectively reduce the spread of bullying information and reduce harm. Many scholars use recognition technology and prediction models to identify bullying information, such as analyzing text features, image contents, social graph parameters, and temporal contents, but the accuracy is still limited, and the relevant psychological research results are largely ignored. Silva ( Silva, Hall, & Rich, 2018) adopted an interdisciplinary approach to develop an automated model for identifying and measuring the degree of cyberbullying in social network sites, and a Facebook app. The proposed model matches streams or bursts of messages with information from adolescents’ social media profiles, which notifies parents about the likelihood that their adolescent is a cyberbullying victim. Therefore, social media plays an important role in the network media environment, which can help with families and schools to curb the spread and deepening of cyberbullying. For example, when media reports on issues such as traditional bullying or cyberbullying related to students, the relevant information should be more detailed based on evidence. Social learning theory of suicide contagion ( Pirkis , Burgess, Francis, Blood, & Jolley, 2006) believes that media reports may make suicide more culturally acceptable, stimulating imitation by observers who have similar experiences with the content of the report, which has greater increases in suicides.

Under the protection of a good external environment, when students face cyberbullying, their reactions have an important connection with their internal psychological qualities. These findings revealed that adolescents’ level of cyberbullying increases as their hopelessness increases and decreases as their value increases (Bülent, 2017). For the victim of bullying, those who tend to attribute the cause of the accident to themselves are more likely to suffer than those who attribute it to others or accidents (Bauman, 2011). Therefore, teenagers should improve their internal psychological quality, learn to control anger, enhance prosocial behavior, and improve their skills in conflict resolution.

Establish our own awareness of responsibility, and increase the intervention of cyberbullying. Whether it is traditional school bullying or cyberbullying, nearly half of teenagers may be bystanders (Knauf, Eschenbeck, & Hock, 2018). Therefore, mobilizing bystanders and raising their awareness of responsibility are promising goals in anti-cyberbullying projects. However, the role of bystanders in cyberspace is complicated. As for bystanders of cyberbullying, there will be different performances, such as helping the victim, reinforcing bullying behavior or keeping silent (Wright, 2016). For example, bystander’s behavior that express approval by forwarding and thumbing up bullying messages will increase the bulliers’ satisfaction with their own behavior, thus strengthening bullying. Bystanders who ignore bullying and keep silent may also be considered as acquiescence to bullying; Bystanders commenting from the perspective of the victim in online comments will inspire other bystanders to defend the victim. The anonymity of the Internet reduces bystanders’ fear of the negative consequences caused by the intervention, but cyberspace technology has also changed the way observers view the incident. More than half of adolescents consider cyberbullying to be unrelated to themselves and choose to be silent (Desmet, Bastiaensens, Van Cleemput, Poels, Vandebosch, Cardon et al., 2016). In cyberspace, when bystanders witness cyberbullying, it is difficult for them assess the situation correctly because they can’t catch the emotional response from the victim directly and sometimes do not even know the identity of the victim. In addition, they may blame the victim because they may have less contextual information about the victim in case of cyberbullying, which makes the attacks seem justified. At the same time, the invisibility of the bystander, the physical distance from the victim, the delayed attention to the bullying incident, and the potential presence of many other bystanders also promote the transfer of responsibility, weaken their sense of responsibility, and promote the passivity of intervention. Therefore, it is a major task of anti-cyberbullying to improve students’ information interpretation ability and form correct network moral values.