Systematic LLM evaluation for news framing detection reveals prompt sensitivity and emotional-language bias, introduces an out-of-domain headline dataset, and shows cross-model consensus aids annotation auditing.
Modeling
5 Pith papers cite this work. Polarity classification is still indexing.
5
Pith papers citing it
verdicts
UNVERDICTED 5representative citing papers
Cross-lingual analysis of 1.76M Singapore comments finds culturally specific hate targets but shared binding moral grammar and threat frames across languages.
LLMs show low endorsement of persuasion-infused messages unless given partisan personas, which then increase polarized endorsements varying by technique and topic.
Reddit analysis finds MAHA users show strong cross-theme belief bundling and network coherence unlike anti-MAHA users, with pandemic-era shifts from anti-fluoride/mask to anti-vaccine to broader anti-science engagement.
The study introduces a framework and reports differences in consistency, cohesiveness, and correctness of themes produced under synchronous versus asynchronous collaboration across three interactive NLP tools.
citing papers explorer
-
Decoding News Narratives: A Critical Analysis of Large Language Models in Framing Detection
Systematic LLM evaluation for news framing detection reveals prompt sensitivity and emotional-language bias, introduces an out-of-domain headline dataset, and shows cross-model consensus aids annotation auditing.
-
Cultural Targets, Structural Frames, Binding Morals: A Cross-Lingual Audit of Online Hate in Multicultural Singapore
Cross-lingual analysis of 1.76M Singapore comments finds culturally specific hate targets but shared binding moral grammar and threat frames across languages.
-
Political Persuasion and Endorsement in Large Language Models
LLMs show low endorsement of persuasion-infused messages unless given partisan personas, which then increase polarized endorsements varying by technique and topic.
-
The Structure and Dynamics of the Online MAHA-sphere
Reddit analysis finds MAHA users show strong cross-theme belief bundling and network coherence unlike anti-MAHA users, with pandemic-era shifts from anti-fluoride/mask to anti-vaccine to broader anti-science engagement.
-
Effects of Collaboration on the Performance of Interactive Theme Discovery Systems
The study introduces a framework and reports differences in consistency, cohesiveness, and correctness of themes produced under synchronous versus asynchronous collaboration across three interactive NLP tools.