Investigating User-Generated Content with a BERT-SVM Framework for Digital Age

This study presents a novel BERT-SVM framework for the legal and linguistic analysis of political parody in India. By collecting over 1,000 instances of politically themed user-generated content from platforms like Twitter, Facebook, and YouTube, we created a rich dataset of satire, parody, and criticism. The data underwent rigorous preprocessing and was annotated by legal and linguistic experts. BERT was employed for semantic feature extraction, capturing contextual nuances, while a Support Vector Machine (SVM) was used for classification tasks. The system achieved a high classification accuracy of 92.4% in distinguishing parody from non-parody content and 87.9% in identifying legally high-risk expressions. The findings reveal how language, sentiment, and legal sensitivity intersect in digital discourse. This research contributes to understanding the role of AI in monitoring politically charged content while respecting freedom of expression. The BERT-SVM model demonstrates promise in enabling nuanced, context-aware moderation systems that align with Indian legal standards and democratic values.