Researchers have developed a new pipeline for analyzing media bias and framing in online news articles. This system groups articles by topic and event, then annotates them with named entities and sentiment. The pipeline was applied to 8,358 Albanian news articles from the GDELT Project, comparing its annotations with GDELT's existing ones. The study found moderate agreement in sentiment and entity extraction, with the new pipeline identifying additional person-entity pairs that could aid bias analysis. The research also explored different annotation prompts, concluding that a simpler prompt is more efficient despite slightly reduced coverage. AI
IMPACT This LLM-based pipeline offers a novel approach to understanding media framing and bias, potentially improving news analysis tools.
RANK_REASON The cluster contains an academic paper detailing a new methodology for media bias analysis. [lever_c_demoted from research: ic=1 ai=1.0]
- Albanian
- alphaXiv
- CatalyzeX
- DagsHub
- GDELT Project
- Gotit.pub
- Hugging Face
- Influence Flower
- ScienceCast
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
