Deep Learning And Machine Learning For Indonesian Online Gambling Website Promotion Detection
Jovan Amarta Liem, Dionisius Tyogo Djatmiko Utomo, Bernardus Ignasio, Tjeng Wawan Cenggoro · Procedia Computer Science · 2025
In Indonesia, online gambling activities are prohibited by current legal regulations, although online gambling activities are prohibited, their promotion continues on digital platforms and is disguised as harmless content. This study explores deep learning and machine learning methods such as BERT, IndoBERT, RoBERTa, Random Forest, XGBoost, CNN, and LSTM. Furthermore, this study also explores the impact of pre-trained feature extractors on Bahasa Indonesia, such as IndoBERT, FastText, and other methods, such as TF-IDF and Word2Vec. The result of these experiments shows that IndoBERT is the most effective model in detecting online gambling website promotion with 98.53% of accuracy, 97.24% of F1-score, 95.65% of recall, 98.88% of precision, and 98.21% of F-beta score.