Enhancing Cyberbullying Detection on Twitter with Psychological Features and Machine Learning

Today, a large number of people dabble in the realm of social media. Due to the pandemic situation, people are even more engaged since they frequently use social media to vent their emotions. One of the many detrimental effects of this pervasive usage of social media is cyberbullying, which is a troubling form of online harassment. Though it can take several forms, the most common one is text. Cyberbullying is common on social media, and instead of confronting the perpetrator, victims often have mental breakdowns as a result of it. This study's computerized cyberbullying detection method accesses Twitter users' psychological traits, including their personalities, moods, and emotions. Our study provides an innovative solution for detecting cyberbullying tweets by an attention-based transformer algorithm combined with embeddings. Our model acts as a detector in classifying tweets that are related to cyberbullied actions. These tweets are converted into numerical vectors by Embeddings and divided into fixed segments through the padding technique. The transformer model learns from the encoder part comprising of self-attention and feed-forward neural network and normalization through the tweet’s dataset. Incredibly accurate cyberbullying detection is made possible by this integrated technology. Our approach promises to identify cyberbullying quickly and precisely to give more control to women over the situation.