Propasafe-Hybrid: A Text-Based Hybrid Propaganda Detection Tool (ILA 2026 Presentation)

Authors

• Thomas Kimmeth, CUNY John Jay CollegeFollow
• Avijit Roy, CUNY John Jay CollegeFollow
• Vivek Sharma, CUNY John Jay CollegeFollow

Document Type

Presentation

Publication Date

Spring 5-2-2026

Abstract

This presentation introduces Propasafe-Hybrid, a hybrid system for sentence-level propaganda detection that combines offline transformer-based classification with selective large language model (LLM) explainability. The system employs a two-stage pipeline in which a local BERT-based classifier evaluates all input text and filters non-propagandistic content, while only high-confidence candidates are forwarded to an LLM for rhetorical technique labeling and explanation. This design enables cost-aware, privacy-conscious, and scalable analysis by reducing unnecessary reliance on external models.

Propasafe-Hybrid identifies propagandistic techniques such as loaded language, obfuscation, and appeal to fear, and generates concise natural language rationales that make these techniques interpretable to users. By bridging computational detection and linguistic interpretation, the system supports both large-scale rhetorical analysis and media literacy applications.

The presentation outlines the system architecture, methodological design choices, and limitations, and discusses directions for future work, including multilingual extension and user-centered evaluation.

Comments

Presented at the 69th Annual Conference of the International Linguistic Association (ILA 2026), John Jay College of Criminal Justice, CUNY, New York, NY, USA.

Slides archived on Zenodo (DOI: https://doi.org/10.5281/zenodo.19991615)

Avijit Roy — ORCID: https://orcid.org/0009-0007-8036-0952
Vivek Sharma — ORCID: https://orcid.org/0000-0001-9590-6574