Researchers have developed ProBel, a new resource for propaganda detection that supports both Arabic and English languages. This resource includes aligned annotations for binary classification, multi-label technique classification, span identification, and reference explanations. Experiments show that a single bilingual multi-task model performs best across various tasks and languages, demonstrating that transfer learning effectiveness depends on the supervision level. AI
IMPACT This research contributes to natural language processing by improving methods for detecting propaganda across languages and different classification levels.
RANK_REASON This is a research paper detailing a new dataset and model for propaganda detection. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →