Abstract: Cyberbullying has emerged as a serious issue in recent years with the growth of social media, which has created serious psychological and emotional effects on victims. This study aims to identify cyberbullying through machine learning methods for automatic classification and identification of abusive content. The work analyzes various natural language processing (NLP) techniques for feature extraction, including TF- IDF, word embeddings, and sentiment analysis, to enhance detection accuracy. Support Vector Machines (SVM), Random Forest, and deep learning models like LSTMs and transformers are used for classification. Real-world social media data are used in the dataset to enable robust training and cross-validation of models. Performance measures such as precision, recall, and F1-score are used to compare various methodologies. The results indicate that the newest deep learning models, particularly transformer-based ones, are far better at detecting cyberbullying than traditional methods with a great accuracy rate. The research contributes to constructing independent tools for the early identification of cyberbullying, promoting a safer online community.
Keywords: Cyberbullying detection, social media monitoring, Machine learning classification, Natural Language Processing (NLP), Text classification, Toxicity detection.
Copyright Form