Table 1 Average extraction performance (precision, recall, and F1 score) averaged across 12 nanomaterials and nanozyme parameters for nanoMINER (ours) and three baseline models: GPT-4.1, o3-mini, and o4-mini

From: Agent-based multimodal information extraction for nanomaterials

Method

Avg Precision

Avg Recall

F1 Score

GPT-4.1

0.71

0.65

0.68

o3-mini

0.68

0.57

0.62

o4-mini

0.78

0.69

0.74

nanoMINER

0.89

0.72

0.79

  1. The bold values indicate the best-performing scores across all evaluated methods for each metric (Average Precision, Average Recall, and F1 Score).
Back to article page