Enhancing Cyberbullying Detection: A Multi-Algorithmic Approach

Given the widespread use of online communication platforms, cyberbullying has emerged as a significant concern, demanding effective detection strategies to protect individuals from its harmful consequences. Cyberbullying predominantly occurs on digital devices like smartphones and computers, often through social media channels. This study address on the issue by proposing a novel approach for cyberbullying detection, offering a valuable tool for online platforms to identify and remove harmful content. Our research emphasizes the critical role of selecting appropriate machine learning (ML) and deep learning (DL) algorithms, as well as preprocessing techniques, in enhancing detection performance. By conducting thorough experiments on a real-world dataset demonstrating the efficacy of the multi-algorithmic framework. Specifically, the findings indicate that the integration of Support Vector Classifier (SVC) and TF-IDF (Term Frequency-Inverse Document Frequency) Vectorizer through a pipeline method outperforms individual algorithms and deep learning models. This research contributes to advancing cyberbullying detection methodologies, providing valuable insights for the development of more robust systems to combat online harassment and foster a safer digital environment.