Analisis Perbandingan Metode Decision Tree Dan K-Nearest Neighbor Untuk Klasifikasi Cyberbullying Pada Sosial Media Twitter

Abstract

This research focuses on analyzing the impact of social media on society, particularly addressing the issue of cyberbullying on the Twitter platform. Based on statistics, the majority of internet users in Indonesia actively utilize social networks, with Twitter being the most dominant platform used for communication and interaction. Therefore, cyberbullying cases often occur on this social media platform. In this study, two classification methods, namely Decision Tree and K-Nearest Neighbor (KNN), were employed to classify cyberbullying-related messages on Twitter. The aim of this research is to compare the performance of these two methods and to identify early signs of cyberbullying as relevant digital evidence for legal proceedings. The dataset used in this study consists of 650 comment records from the period 2019 to 2021, with predefined labels. The analysis results indicate that K-Nearest Neighbor achieved the highest accuracy, reaching 75.99%, compared to Decision Tree with 65.00%. Hence, K-Nearest Neighbor is considered a more effective method for cyberbullying analysis on the Twitter platform. Additionally, the identification of early signs of cyberbullying in comment id 2 can serve as relevant digital evidence for legal purposes. This research provides better insights into the effectiveness of classification in addressing cyberbullying issues on the Twitter platform.