Rosetta system reconstructs metadata from data values in undocumented warehouses

Researchers have developed a system called Rosetta to reconstruct metadata from data values in undocumented warehouses, addressing the challenge of cryptic identifiers and absent documentation. Rosetta integrates a language model with a verification harness that extracts structural evidence, allowing the model to propose semantics. This approach improves accuracy and coverage compared to using the language model directly, particularly in selecting which columns to provide metadata for. The system demonstrated effectiveness on BIRD databases and i2b2 clinical warehouses, successfully decoding ICD-9 codes and abstaining on NDC drug codes when necessary. AI

IMPACT This research could improve data integration and analysis in environments with poor documentation.

RANK_REASON The cluster contains an academic paper detailing a new system for metadata reconstruction. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• arXiv
• BIRD
• i2b2
• ICD-9
• national drug code
• Rosetta

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →