HATE SPEECH PREDICTION USING LOGISTIC REGRESSION AND ADABOOST ALGORITHM

Abstract

The proliferation of hate speech on social media necessitates robust automated systems to identify and mitigate harmful content effectively. While traditional sentiment analysis models struggle with contextual nuances, this paper proposes a hybrid machine learning approach combining Logistic Regression (LR) and AdaBoost to improve hate speech detection accuracy. Leveraging Natural Language Processing (NLP) and data mining techniques, we extract linguistic and semantic features to distinguish hate speech from benign text. The AdaBoost ensemble method enhances the base Logistic Regression classifier by iteratively refining focus on misclassified samples, thereby improving generalization. Our model is evaluated on a benchmark dataset, with results demonstrating superior performance in precision, recall, and F1-score compared to standalone classifiers. Additionally, we analyze prevalent hate speech trends, identifying frequently targeted topics. The proposed framework offers a scalable solution for real-time content moderation, balancing interpretability (via LR) and predictive power (via AdaBoost). This work contributes to the evolving discourse on ethical AI by addressing algorithmic bias and improving detection robustness in dynamic online environments.

Author

Mr.MATHIYAZHAGAN M, Mr.ANANTH E, Dr.SATHYABALAJI N