From News Streams to Narrative Intelligence Briefs: LLM-Assisted Political Discourse Analysis in the Hungarian 2026 Pre-Election Context
Year: 2026
Type: conference-paper
Abstract: Journalists, activists and policymakers need scalable ways to convert high-volume political news into actionable narrative intelligence, yet most NLP pipelines stop at classification outputs that are difficult to operationalise.We present a case study assessing whether LLMs, constrained by explicit analytical schemas and multi-stage validation, can reliably produce structured narrative intelligence briefs from Hungarian pre-election news.Our pipeline processes 574 election-relevant articles from 21 sources through three stages: (1) per-article extraction of Narrative Policy Framework event frames and SemEval-taxonomy manipulation techniques; (2) embedding-based narrative clustering; and (3) constrained brief generation with five sections-narrative summary, character map, manipulation profile, escalation assessment, and counter-strategy-where counter-strategies are grounded in curated external sources.Evaluation through human experts and LLM-as-judge reveals a consistent quality gradient: descriptive sections outperform prescriptive ones, with failures driven by contextual insufficiency rather than hallucination.We document failure modes and assess which components support semi-automation.
Language: English
Subfield: General Social Sciences
Field: Social Sciences
Domain: Social Sciences
Open Access status: hybrid
OpenAlex ID: W7166583241