Real-Time Live Insult Analysis on Twitter-X Social Media Platform

Real-time insult analysis could help platforms identify and filter out harmful content promptly, promoting a healthier online environment and reducing the spread of negativity. Identifying and addressing insults in real-time, platforms can enhance user experience by fostering more civil and respectful interactions among users. It requires robust algorithms capable of accurately detecting insults across various languages and contexts while ensuring user privacy and minimizing false positives. In this study, sentiment analysis was conducted using Turkish Twitter data. The data was analyzed by applying data analysis methods and Naive Bayes (NB), K-Nearest Neighbor (KNN), Support Vector Machine (SVM), Logistic Regression (LR), Random Forest, and Decision Tree. They were classified as positive and negative using classification algorithms. The performance of these classifiers was evaluated with the F1 score. According to the results, the Logistic Regression classifier performed with the highest F1 score (85%). Among other classifiers, SVM (84%), Random Forest (83%), Naive Bayes (83%), and Decision Tree (80%) achieved F1 scores. This study aimed to compare classifiers when performing sentiment analysis from Turkish Twitter data.