Özet:

This study was carried out to contribute to the literature and to define and compare the algorithms used in the detection of cyberbullying. Research articles published in academic journals in the last 10 years, in Turkish and English, whose full text can be accessed, were included in the study. The literature search was conducted with the keywords “Cyber Bullying”, “Guess” and “Artificial Intelligence” in Turkish and “Predicting Cyberbullying” and “Artificial Intelligence” in English from Google Scholar, ProQuest, ScienceDirect, Scopus, Wiley Online Library and Pubmed online databases. Since no results could be reached with Turkish keywords, only English keywords were made in October 2022 with the joint decision of the researchers. 19 research articles were included in the study. Eighteen of the included studies are in English and one is in Turkish. The aim of the studies is to detect and/or prevent cyberbullying with the help of algorithms based on artificial intelligence. As a result of the studies examined, the proposed artificial intelligence models have achieved successful results in achieving their goals for cyberbullying. In one of the studies, an artificial intelligence model was developed that can detect cyberbullying from the message content and prevent the message from being sent. In addition, artificial intelligence applications have been able to successfully detect cyberbullying not only in online social networks, but also in networks used in-house at workplaces. A contribution has been made to the literature on the basis of the theory that cyberbullies abuse cyber victims by using an accusatory language in social media. Studies on cyberbullying usually include text-based analysis, but it is recommended to develop techniques that can analyze images, videos and sounds in order to better detect cyberbullying. Additional updates to apps can be scaled up to perform real-time text prediction before programs suggest any abusive content to the user for typing.

Anahtar Kelimeler: