Sentiment Analysis of Hate Speech Using SVM Method

Abstract

This research analyses the sentiment of hate speech on social media using the Support Vector Machine (SVM) method. The Indonesian dataset from Kaggle is processed through text normalisation, filtering, and stemming to ensure the data is suitable for use in machine learning models. The SVM model was compared with Naive Bayes and Random Forest. Results showed SVM excelled with 75.40% accuracy, compared to Naive Bayes (67.34%) and Random Forest (46.64%). Performance evaluation is done with a confusion matrix that measures accuracy, precision, recall, and F1-score. The advantage of SVM lies in its ability to find optimal decision boundaries in a multidimensional feature space, making it more effective in handling complex interactions between features compared to Naive Bayes and Random Forest. The findings show that SVM is more effective for the classification of hate speech on social media. This research contributes to the development of automated monitoring systems that are more accurate and efficient in detecting and classifying hate speech content, thus improving countermeasures on social media platforms.