Table 5 Evaluation results of classification performance of DeepPatentAI on various datasets.
From: A Global Dataset Mapping the AI Innovation from Academic Research to Industrial Patents
Dataset | Overall Accuracy | Class | Precision | Recall | F1 Score |
|---|---|---|---|---|---|
Training Set | 0.984 | Non-AI | 0.989 | 0.981 | 0.985 |
AI | 0.979 | 0.988 | 0.983 | ||
Validation Set | 0.977 | Non-AI | 0.977 | 0.972 | 0.975 |
AI | 0.977 | 0.981 | 0.979 | ||
Test Set | 0.975 | Non-AI | 0.983 | 0.975 | 0.979 |
AI | 0.963 | 0.974 | 0.968 | ||
Manually Annotated Set | 0.908 | Non-AI | 0.993 | 0.837 | 0.915 |
AI | 0.807 | 0.992 | 0.900 |
