Analyzing Cyberbullying Word Distribution Across Social Media to Develop Lexicon Dictionary

Social media platforms have become essential tools for communication, collaboration, and exchanging information, ideas, and knowledge among users worldwide. Despite their benefits, the anonymity offered by these platforms has unfortunately led to a rise in hate speech and cyberbullying, causing concern across the globe. This issue has drawn the attention of researchers and scholars now focused on devising methods for automatically detecting cyberaggression and hate speech. The goal is to mitigate these harmful behaviors and create safer online environments [1]. This study specifically aims to identify and analyze the usage of cyberbullying language within the English Tweet corpus, focusing on distinguishing between bullying and non-bullying language. This research contributes to developing more effective cyberbullying detection models by examining the frequency and patterns of cyberbullying words. These findings are vital for enhancing the efficacy of cyberbullying detection on social networks in future research efforts. The conclusions of the experiments indicate that the SVM achieved the highest average accuracy, approximately $84 \%$.