Identifikasi Komentar Spam Bahasa Indonesia pada Instagram Menggunakan Gated Recurrent Unit

Abstract

Instagram is a popular social media platform that has become a primary space for users to interact and share information. Although Instagram provides a positive environment for interaction, the surge in users has brought about new challenges, such as an increase in disruptive spam comments. These spam comments not only disturb user experiences but also pose risks of harmful link sharing that can adversely affect users. To address this issue, this study employs an approach utilizing the Gated Recurrent Unit model and fastText word embedding. Research outcomes demonstrate the effectiveness of this model in detecting spam comments with an accuracy rate of 0.961. These findings suggest that the Gated Recurrent Unit method holds potential in identifying and mitigating spam comments on Instagram.