This research demonstrates how few-shot learning techniques with Large Language Models significantly improve extraction of Metal-Organic Frameworks (MOFs) synthesis conditions from scientific literature.

• Develops a human-AI interactive approach for data curation and synthesis extraction
• Achieves superior accuracy compared to zero-shot LLM approaches lacking specialized materials knowledge
• Establishes a more efficient pathway for materials scientists to extract critical synthesis parameters
• Creates a practical framework for automating complex materials science literature analysis

This advancement matters for engineering by streamlining the discovery and optimization of MOFs with tailored properties for applications in catalysis, gas storage, and molecular separation.

LLM-based MOFs Synthesis Condition Extraction using Few-Shot Demonstrations