Why Traditional NLP Fails Political News — and How LLMs Can Bridge the Gap
Maryam Fooladi, Federico Bottino
Paper, 3rd Workshop on NLP for Political Sciences (PoliticalNLP 2026) at LREC 2026 — Palma de Mallorca, 2026
Abstract
Traditional sentiment analysis collapses politically substantive news into a 'neutral' bin in roughly 70% of cases, missing framing, ideology, and rhetorical strategy. We compare classical NLP pipelines with LLM-based approaches and show how LLMs can recover the analytic dimensions social scientists actually need.