This research demonstrates how large language models can significantly improve metadata standardization for scientific datasets when augmented with structured knowledge.

• LLMs with structured knowledge bases achieved higher adherence to metadata standards than LLMs alone
• Experiments on 200 lung cancer samples from NCBI BioSample showed measurable improvements in data quality
• The approach provides a scalable solution to metadata curation challenges
• Results indicate potential for automating quality control in scientific repositories

Why it matters: Standardized metadata is critical for medical research discoverability, reproducibility, and interoperability across datasets—this approach could dramatically improve access to valuable clinical data.

Use of a Structured Knowledge Base Enhances Metadata Curation by Large Language Models