Table 8 Result of argument analysis using hierarchical attention network (FastText Word Embedding)
From: Argument annotation and analysis using deep learning with attention mechanism in Bahasa Indonesia
No | Batch size | Accuracy (%) | Recall (%) | Precision (%) | F1 macro (%) | ROC-AUC |
|---|---|---|---|---|---|---|
1 | 16 | 70.65 ± 3.40 | 62.97 ± 7.92 | 61.91 ± 14.69 | 60.75 ± 11.64 | 68.00 ± 16.00 |
2 | 32 | 71.33 ± 3.99 | 64.27 ± 7.86 | 62.90 ± 15.50 | 62.41 ± 11.73 | 70.72 ± 13.47 |
3 | 64 | 73.47 ± 0.88 | 64.20 ± 3.80 | 76.37 ± 3.73 | 63.89 ± 3.78 | 76.60 ± 2.95 |
4 | 100 | 72.11 ± 1.66 | 65.35 ± 5.81 | 71.52 ± 3.48 | 64.57 ± 6.73 | 75.83 ± 2.71 |
5 | 128 | 72.51 ± 3.96 | 65.58 ± 5.13 | 75.10 ± 5.02 | 64.57 ± 6.05 | 77.73 ± 2.87 |