A new system called DataWeave has been developed to assist journalists in analyzing large structured datasets for data journalism. It combines conversational interaction with LLMs, schema grounding, and analytical planning to generate executable queries. The system is designed to function as an interactive partner, allowing users to inspect, correct, and steer the LLM's outputs as their hypotheses evolve. A case study involving professional journalists analyzing the U.S. Department of Education's IPEDS dataset demonstrated the system's utility and informed its architectural design and analytical workflow. AI
IMPACT Enhances the capabilities of journalists in uncovering trends and narratives from complex datasets using LLM assistance.
RANK_REASON The item is a research paper detailing a new system for structured data analysis. [lever_c_demoted from research: ic=1 ai=1.0]
- arXiv
- DataWeave
- Hugging Face
- Integrated Postsecondary Education Data System
- journalists
- LLM
- Raquib Bin Yousuf
- United States Department of Education
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →