Table 4 Comparison between the proposed model and the counterpart models
From: Sentiment analysis of Indonesian datasets based on a hybrid deep-learning strategy
Model | Accuracy* | Precision* | Recall* | F-score* | AUC* |
|---|---|---|---|---|---|
BERT + LSTM-CNN-SVM | 75.50 | 73.73 | 77.10 | 74.63 | 76.33 |
BERT and feature union | 82.60 | 82.23 | 82.83 | 82.50 | 82.67 |
BERT + LSTM-CNN fusion model | 83.50 | 82.23 | 84.83 | 83.17 | 83.77 |
BERT-large + BiGRU | 84.60 | 86.23 | 83.60 | 84.83 | 84.77 |
RoBERTa-LSTM | 83.30 | 82.17 | 84.10 | 83.07 | 83.33 |
RoBERTa-CNN | 83.00 | 82.77 | 83.37 | 82.93 | 83.23 |
Roberta + BiGRU-ATT | 82.87 | 81.40 | 84.03 | 82.57 | 83.03 |
SBERT-MLP | 80.83 | 82.53 | 78.60 | 80.30 | 81.17 |
Proposed model | 85.13 | 85.41 | 84.94 | 85.17 | 85.14 |