Table 9 Ablation study on different components of CAFIIE.
From: Cyberattack event and arguments extraction based on feature interaction and few-shot learning
Method | CASIE | DNRTI | ||||||
|---|---|---|---|---|---|---|---|---|
P | R | F1 | Accu | P | R | F1 | Accu | |
CAFIIE | 0.8547 | 0.7034 | 0.7717 | 0.8992 | 0.8628 | 0.8433 | 0.8529 | 0.9545 |
-w/o FI | 0.7781 | 0.7273 | 0.7518 | 0.8162 | 0.7863 | 0.7415 | 0.7632 | 0.8246 |
-w/o BiLSTM | 0.7976 | 0.7847 | 0.7911 | 0.8363 | 0.8216 | 0.7633 | 0.7914 | 0.8435 |
-w/o ANN | 0.8155 | 0.7761 | 0.7953 | 0.8429 | 0.8521 | 0.7772 | 0.8129 | 0.8768 |
Method | MalwareTextDB | Private Dataset | ||||||
|---|---|---|---|---|---|---|---|---|
P | R | F1 | Accu | P | R | F1 | Accu | |
CAFIIE | 0.5973 | 0.6424 | 0.6190 | 0.8863 | 0.6681 | 0.6765 | 0.6723 | 0.8122 |
-w/o FI | 0.5436 | 0.5673 | 0.5552 | 0.8157 | 0.5612 | 0.5376 | 0.5491 | 0.7248 |
-w/o BiLSTM | 0.5751 | 0.5824 | 0.5787 | 0.8367 | 0.5833 | 0.5674 | 0.5752 | 0.7785 |
-w/o ANN | 0.5862 | 0.5961 | 0.5911 | 0.8449 | 0.6273 | 0.6128 | 0.6199 | 0.7852 |