Propasafe-Hybrid: a text-based hybrid propaganda detection tool (FLAIRS-39 Presentation)

Presentation slides for the paper “Propasafe-Hybrid: A Text-Based Hybrid Propaganda Detection Tool,” presented at the 39th Florida Artificial Intelligence Research Society Conference (FLAIRS-39), held in Marco Island, Florida, USA, on May 18, 2026 (Session 6B). The presentation introduces Propasafe-Hybrid, a hybrid propaganda detection framework that combines local transformer-based sentence classification with selective large language model (LLM) explainability. The system uses a fine-tuned BERT classifier as a local first-pass sieve and forwards only high-probability propagandistic sentences to an LLM-based explanation layer, reducing token usage while preserving interpretability and deployment practicality. The slides discuss the system architecture, threshold-gated explainability pipeline, cost-aware filtering strategy, 17-technique propaganda taxonomy, implementation details, example outputs, and deployment considerations for real-world propaganda analysis workflows. Slides presented by Avijit Roy at FLAIRS-39. Conference:FLAIRS-39 — The 39th Florida Artificial Intelligence Research Society ConferenceMarco Island, Florida, USAMay 17–20, 2026 Paper DOI:https://doi.org/10.32473/flairs.39.1.141595