An Intelligent System for Detecting Online Gambling Promotion on Social Media Based on BERT and Adaptive Browser Extension

The proliferation of online gambling promotions on social media has created serious social and legal problems in Indonesia. Government efforts through site blocking and manual monitoring are often outpaced by the rapid spread and diversity of languages ​​used to disguise promotional content. Previous research has developed machine learning models to detect gambling content, but most are limited to post-facto analysis and are unable to handle implicit, domain-specific vocabulary. To address this gap, this study proposes an intelligent detection system utilizing IndoBERT, which undergoes Domain-Adaptive Pre-Training (DAPT) on a corpus of Indonesian language online gambling text, followed by fine-tuning for text classification. The resulting model demonstrated superior performance with an F1-Score of 98.58%, outperforming baselines such as TF-IDF+SVM, BiLSTM, mBERT, and IndoBERT without DAPT. Furthermore, the model was integrated into an adaptive browser extension capable of scanning, classifying, and filtering social media content in real time. Test results on YouTube and X demonstrated the system's effectiveness in detecting and masking online gambling promotions, without disrupting neutral or anti-gambling content. This research contributes to the development of BERT based NLP models in domain-specific scenarios while presenting an applicable solution that is directly beneficial for the protection of social media users.