Table 6 Sentiment model performance on the testing dataset.
Index | Embedding Method | Classification method | Accuracy | F1 score |
|---|---|---|---|---|
1 | LIWC | Voting | 0.288 | 0.394 |
2 | VADER | Voting | 0.285 | 0.378 |
3 | Bag of words (5000 features) | MultinomialNB | 0.769 | 0.770 |
4 | Bag of words (5000 features) | LinearSVC | 0.791 | 0.788 |
5 | BERT | Logistic Regression | 0.808 | 0.810 |
6 | BERT (our paper) | Neural Network | 0.829 | 0.829 |
