Table 8 Impact of entity and concept knowledge on model performance for twitter15 datasets
From: KAGN:knowledge-powered attention and graph convolutional networks for social media rumor detection
Variants | Accuracy | Precision | Recall | F1 |
|---|---|---|---|---|
PTE | 0.8036 | 0.8053 | 0.8036 | 0.8037 |
PTE + KGE/E | 0.8156 | 0.833 | 0.8149 | 0.8154 |
PTE + KGE/C | 0.8344 | 0.8344 | 0.8354 | 0.8342 |
PTE + KGE | 0.8406 | 0.8486 | 0.8408 | 0.8417 |
PTE + KAE/E | 0.8281 | 0.8323 | 0.8291 | 0.8279 |
PTE + KAE/C | 0.8438 | 0.8436 | 0.8445 | 0.8428 |
PTE + KAE | 0.8636 | 0.869 | 0.8628 | 0.8642 |