Klasifikasi Tweet Pelecehan Online pada Media Sosial X Menggunakan Gated Recurrent Unit

Abstract

The development of communication and information technology is growing rapidly in various fields of human life, especially in the use of social media. One of the most widely used social media is X (twitter) which is often used for real-time information, including the latest news and provides a space for discussion on certain topics. However, the use of X not only has a positive impact but also a negative impact where many parties or users abuse the function of X itself resulting in an increase in crimes that arise, one of which is online harassment. Identifying online harassment is very important to protect users and prevent the widespread practice of online harassment and violence in Indonesia. However, this process will certainly take a long time if done manually. Therefore, an approach is needed that can process harassment tweet data into something that can effectively classify harassment. This research aims to classify online harassment tweets on social media X using gated recurrent units and word embedding fastText. The data used in this research amounted to 7006 data by crawling using the tweet harvest tool in the X application. Based on the evaluation using confusion matrix, the accuracy is 0.91. So it can be concluded that the algorithm used in this study is good enough in classifying online harassment on social media X.