Intelligent Flagged Content Detection with Transformer-Based Models for Secure Online Environments

The increasing volume of user-generated content on global online platforms has made it very crucial to detect and moderate harmful content like criminal activities, illegal substances, extremist codewords, and other prohibited materials across multiple languages. In this study, we present an intelligent system for detecting flagged content using advanced transformer-based models such as BERT, RoBERTa, XLNet, and T5. These models have been further trained to very accurately and effectively detect from illegal activities to extreme language and drug related content. Our method surpasses previous methods of content moderation achieving great improvements in classification accuracy, precision, recall and F1-score. Furthermore, the implemented system enables offline or online real-time process and therefore large-scale throughput. It also gives a detailed overview of the architectural designs of the model, training processes and other problems faced during the model development stage. The results illustrate the power of transformer-based models in offering a robust strategy for automated detection of flagged content across languages to help in cultivating better online safety. This research also contributes to SDG 16 by enhancing online security and SDG 10 by promoting fairer digital spaces. The study concludes with a discussion on potential applications in social media moderation, law enforcement, and global security agencies and future research aimed at improving detection capabilities.