Table 6.
Performance evaluation of the models under cross-dataset environment. (a) Train: Waseem & Hovy, Test: SemEval 2019 (b) Train: SemEval2019, Test: Waseem & Hovy. The first row for each dataset represents the Micro average and the second represents the Macro average.
| Method | Dataset | F1 | Prec. | Rec. |
|---|---|---|---|---|
| Proposed model | (a) | 62.73 | 62.65 | 63.13 |
| 61.45 |
61.85 |
61.36 |
||
| (b) | 61.04 | 60.98 | 61.11 | |
| 55.34 | 55.36 | 55.33 | ||
| Badjatiya et al. (2017) | (a) | 61.85 | 62.07 | 62.84 |
| 60.22 |
61.44 |
60.28 |
||
| (b) | 60.66 | 60.01 | 61.60 | |
| 54.03 | 54.41 | 54.00 | ||
| Agrawal and Awekar (2018) | (a) | 56.57 | 63.57 | 62.39 |
| 53.28 |
63.9 |
56.8 |
||
| (b) | 54.82 | 56.55 | 53.71 | |
| 49.81 | 50.15 | 50.16 | ||