Classifying cyberbullying data on Indonesian social media feeds utilizing sentiment analysis technique with decision tree model

Abstract

In the technological sophistication era, social media has rapidly become a site that is favored by many people. However, it can harm each other when accessed that does not have a certain limitation. Cyberbullying is an act to attack an individual through social media. This study aims to classify the cyberbullying data that occurs through conversation on social media. We used the sentiment analysis technique to extract and preprocess the text from tweets. Furthermore, we implemented two classifier models using C4.5 and C5.0 methods to classify the data into two classes. The experiment was conducted by comparing the performance of both methods C4.5 and C5.0 to classify cyberbullying data. From the evaluation using accuracy, we obtained the performance of C5.0 increased with several variations of attributes.