This research introduces a novel retrieve-then-prompt approach using LLMs to significantly improve ontology matching for biomedical data integration without requiring large training datasets.

• Combines Large Language Models with a Prioritized Depth-First Search algorithm to match medical terminologies
• Outperforms existing systems in biomedical ontology matching benchmarks
• Addresses vocabulary processing limitations of traditional machine learning approaches
• Enables more effective knowledge sharing and data interoperability in healthcare settings

Why it matters: Improved ontology matching enables seamless integration of medical databases, clinical terminologies, and research knowledge bases, facilitating better healthcare decision support systems and cross-institutional data sharing.

Original Paper: Ontology Matching with Large Language Models and Prioritized Depth-First Search