Table 6 Result of argument analysis using deep learning model with attention mechanism (Word Embedding from Scratch)
From: Argument annotation and analysis using deep learning with attention mechanism in Bahasa Indonesia
No | Model name | Accuracy (%) | Recall (%) | Precision (%) | F1 Macro (%) | ROC-AUC |
|---|---|---|---|---|---|---|
1 | CNN | 74.35 ± 6.12 | 66.49 ± 11.45 | 74.21 ± 9.36 | 63.84 ± 14.81 | 80.86 ± 0.05 |
2 | CNN + Att | 70.84 ± 3.33 | 63.02 ± 6.58 | 71.80 ± 5.80 | 61.17 ± 9.00 | 74.06 ± 0.07 |
3 | LSTM | 65.79 ± 0.58 | 51.10 ± 2.04 | 43.80 ± 13.29 | 43.54 ± 5.49 | 56.43 ± 0.06 |
4 | LSTM + Att | 69.98 ± 5.46 | 62.22 ± 8.53 | 69.14 ± 10.36 | 61.14 ± 9.51 | 72.40 ± 0.10 |
5 | GRU | 67.15 ± 3.69 | 53.46 ± 6.54 | 51.25 ± 14.22 | 47.48 ± 10.26 | 57.54 ± 0.09 |
6 | GRU + Att | 68.71 ± 4.73 | 58.89 ± 10.85 | 53.55 ± 17.62 | 52.39 ± 15.12 | 67.69 ± 0.10 |
7 | BiLSTM | 67.64 ± 1.56 | 56.67 ± 3.85 | 64.01 ± 8.70 | 54.27 ± 6.46 | 65.72 ± 0.07 |
8 | BiLSTM + Att | 67.54 ± 4.21 | 57.50 ± 7.11 | 62.60 ± 16.99 | 52.86 ± 10.94 | 73.35 ± 0.09 |
9 | BiGRU | 66.96 ± 1.01 | 54.92 ± 4.01 | 61.59 ± 4.09 | 51.28 ± 6.29 | 61.83 ± 0.05 |
10 | BiGRU + Att | 69.10 ± 2.01 | 61.06 ± 9.05 | 74.14 ± 6.76 | 56.28 ± 11.17 | 71.47 ± 0.10 |