Detection of hate: speech tweets based convolutional neural network and machine learning algorithms.

Area of Science:

• Natural Language Processing
• Machine Learning
• Computational Social Science

Background:

• Social media facilitates information sharing but also enables harmful content like hate speech.
• Developing automated systems to identify and mitigate hate speech is crucial for online safety.

Purpose of the Study:

• To develop and evaluate a machine learning-based approach for the automatic identification of hate speech on social media platforms.
• To assess the effectiveness of Term Frequency-Inverse Document Frequency (TF-IDF) for feature engineering in hate speech detection.

Main Methods:

• Utilized TF-IDF for feature engineering on text data from three distinct datasets: "Hate speech offensive tweets by Davidson et al.", "Twitter hate speech", and a merged dataset including "Cyberbullying dataset (toxicity_parsed_dataset)".
• Evaluated eleven machine learning and deep learning classifiers: Logistic Regression (LR), Naive Bayes (NB), Multi-layer Perceptron (MLP), Support Vector Machine (SVM), Random Forest (RF), K-Nearest Neighbor (KNN), K-Means, Decision Tree (DT), Gradient Boosting classifier (GBC), Extra Trees (ET), and Convolutional Neural Network (CNN).

Main Results:

• The TF-IDF based feature engineering approach, when applied to the eleven classifiers, demonstrated high efficacy in identifying hate speech.
• Achieved a maximum accuracy exceeding 99% across the evaluated models, indicating a robust performance in hate speech detection.

Conclusions:

• The proposed TF-IDF based feature engineering method is highly effective for automatic hate speech detection using machine learning and deep learning models.
• The study highlights the potential of computational approaches to address the challenge of online hate speech, offering a scalable solution for content moderation.