Cyberbullying Predictive Analysis on Twitter(X) Data with Multi-model Supervised Technique

The study presents a predictive analysis approach for cyberbullying detection on Twitter, utilizing a multi-model supervised technique. The research aims to develop an effective strategy to identify and classify instances of cyberbullying before they escalate further on social media platforms. A dataset of tweets related to cyberbullying was collected, preprocessed, and transformed into numerical features. Three machine learning models, namely Support Vector Machines (SVM), Random Forest (RF), and Deep Neural Networks (DNN), were trained and evaluated to predict the presence of cyberbullying in tweets. The results demonstrated the effectiveness of the proposed technique, achieving high accuracy, precision, recall, and F1-score. Furthermore, additional experiments were conducted to investigate the impact of different feature engineering methods on model performances. The findings highlight the importance of considering various linguistic and context-based features in cyberbullying detection tasks. Overall, this study contributes to the ongoing efforts in developing proactive measures to combat cyberbullying, providing insights and potential approaches for monitoring and addressing this prevalent issue on Twitter.