• Kona Tulya Sree Simla
• ch.Sowjanya

The increasing prevalence of cyberbullying on digital platforms poses a serious threat to online safety and mental well-being. Detecting such harmful behavior in real-time remains a critical challenge due to the dynamic and ambiguous nature of user-generated content. This research presents a novel probabilistic framework designed to efficiently detect cyberbullying by leveraging the combined strengths of Support Vector Machine (SVM), Naive Bayes (NB), and Random Forest (RF) classifiers. The proposed model operates on a pipeline that begins with rigorous data preprocessing, including text normalization, tokenization, stop-word elimination, and feature extraction using Term Frequency–Inverse Document Frequency (TF-IDF). Each classifier contributes uniquely to the detection process: Naive Bayes captures probabilistic word associations, SVM identifies complex decision boundaries, and Random Forest enhances robustness through ensemble learning. A weighted probabilistic ensemble strategy is implemented to combine predictions, assigning dynamic weights based on individual model confidence scores to maximize overall accuracy and reliability. Experimental validation is performed on benchmark cyberbullying datasets sourced from social media platforms. The proposed framework demonstrates high performance, achieving an accuracy of 94.3% and an F1-score of 0.91, surpassing the effectiveness of standalone models. The results confirm the framework’s capability to generalize across varied linguistic patterns and handle class imbalance effectively. This work establishes a strong foundation for scalable, real-time cyberbullying detection systems. Future enhancements will focus on incorporating deep contextual embeddings and expanding the framework to support multilingual text classification for broader applicability.

Cyberbullying Detection, Probabilistic Framework, Support Vector Machine, Naive Bayes, Random Forest, Text Classification, Ensemble Learning.

"A Novel Probabilistic Framework for Efficient Cyberbullying Detection ", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.12, Issue 5, page no.k454-k464, May-2025, Available :http://www.jetir.org/papers/JETIR2505B35.pdf

"A Novel Probabilistic Framework for Efficient Cyberbullying Detection ", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.12, Issue 5, page no. ppk454-k464, May-2025, Available at : http://www.jetir.org/papers/JETIR2505B35.pdf