A Study on the Detection of Cyberbullying using CNN with IbI Logics Algorithm (ILA)

While it's true that OSNs make it easier for people to connect with one another, they also fuel an increase in antisocial behaviors like hate speech, or trolling. Especially for women and children, cyberbullying has caused extreme emotional and physical suffering, and in extreme cases, it has even driven victims to take their own lives. Depending on users to report instances of bullying is one conventional method for detecting cyberbullying, but it isn't foolproof. Cyberbullying may be detected and flagged automatically using deep learning along with machine learning methods. These algorithms can also identify patterns of behavior that could indicate cyberbullying. Consequently, the purpose of this research is to identify and categorize instances of cyberbullying using Twitter data by means of an ensemble deep learning system called IbI Logics system (ILA). The research aims to analyze social media data by means of an ensemble learning procedure and natural language processing (NLP). The last step was to try every conceivable combination using the ensemble models. According to the findings, ensemble models are more likely to provide optimal outcomes. The optimal ensemble CNN, which includes all three CNNs, outperforms the others with respect to precision (97.7%), accuracy (97.7%), recall (97.8%), in addition F1 average score (97.7%).