A Survey on Methods for Detecting Cyberbullying in Multilingual Documents

Digital technologies are now swallowing the world. People irrespective of the age and gender are influenced by the colourful wings that they provide. Teenagers are the main victims of this digital era. They become addicted to games and the virtual world more quickly than any other age group. Their age is so critical that they are very much sensitive. There is a natural tendency among teenagers to do things so as to catch the attention of others. Sometimes this paves way to bully or harass or embarrass others on the internet or other digital spaces such as social media sites and that causes a negative impact on those who are being targeted then there arises the threat of cyberbullying. Social media users are not merely sticking on English language in their posts or comments but the usage of multilingual code mixing or even code switching is very much prevalent. Surveys have been done among people of different ages in many countries and have demonstrated various consequences of cyberbullying victimisation that lead to change in behaviour and increased anxiety. Researchers identified the necessity of computerbased solutions for determining, preventing, mitigating ow even stopping cyberbullying. This paper is a survey of various computer-based techniques specifically concentrating on machine learning, deep learning and natural language processing that targets to detect cyberbullying in online media.