Deteksi Komentar Promosi Judi Online Pada Platform Youtube Menggunakan Metode LSTM Dalam Sistem Keamanan Konten Digital

The increasing spread of online gambling promotions in YouTube comment sections has become a significant digital security concern, especially as perpetrators often disguise words, symbols, and writing patterns to bypass the platform’s moderation system. This phenomenon highlights the need for an adaptive detection mechanism capable of understanding linguistic variations and contextual cues. This study aims to design and develop a web-based system for detecting online gambling promotional comments using the Long Short-Term Memory (LSTM) method integrated with the YouTube Data API v3 for real-time comment retrieval. The methodology includes text cleaning, tokenization, vector transformation through word embedding, LSTM model training, and performance evaluation. The system implementation utilizes Laravel as the backend platform, while the deep learning model is processed using Python. The results indicate that the LSTM model can accurately classify comments containing online gambling elements, even when disguised through unconventional spelling. The developed system enables users to view detection results and manually delete flagged comments directly through the dashboard interface. This research contributes to enhancing intelligent content moderation technology that responds effectively to the evolving patterns of disguised language across digital platforms.