Enhancing Cyberbullying Detection Using Ensemble Learning and Embeddings

Cyberbullying represents a significant challenge in online environments, requiring advanced techniques for its accurate detection and mitigation. This paper introduces a novel approach that leverages ensemble learning and embedding methods to enhance cyberbullying detection. The proposed framework integrates various classifiers, including deep learning models, decision trees, random forests, and logistic regression, in combination with Universal Sentence Embeddings for semantic text representation. The study employs a labeled dataset sourced from offensive language databases, which is preprocessed and divided into training and testing sets. Hyperparameter optimization for traditional classifiers is performed using grid search, while a deep learning model is trained to identify complex patterns in cyberbullying content. Ensemble learning is utilized to combine predictions from multiple models, improving overall detection performance and generalization. The effectiveness of the proposed approach is evaluated using metrics such as accuracy and confusion matrices, demonstrating superior performance compared to individual models. The results indicate that the ensemble learning framework significantly enhances the accuracy of cyberbullying detection, contributing to the growing body of research on online safety and machine learning applications in digital platforms

Ensemble learning · Machine learning · Advanced Malware Detection Techniques · Bullying, Victimization, and Aggression · Computer Science · Hate Speech and Cyberbullying Detection · Artificial Intelligence