Table 6 Sentiment classification results on SMP2020-EWECT.
From: A novel approach for multiclass sentiment analysis on Chinese social media with ERNIE-MCBMA
Models | Acc (%) | Pre(%) | Recall(%) | F1 (%) |
|---|---|---|---|---|
TextCNN | 70.00 | 70.59 | 70.00 | 70.15 |
TextRNN | 66.00 | 65.75 | 66.00 | 65.60 |
TextRCNN | 69.26 | 69.09 | 69.26 | 68.77 |
TextRNN_Att | 68.58 | 69.88 | 68.58 | 68.76 |
FastText | 67.52 | 67.60 | 67.52 | 67.45 |
Transformer | 66.78 | 67.80 | 66.78 | 67.16 |
BERT | 74.02 | 75.4 | 74.02 | 74.46 |
BERT-TextCNN53 | 74.44 | 74.78 | 74.43 | 74.47 |
BERT-BiLSTM54 | 74.10 | 74.26 | 74.10 | 74.20 |
BERT-TextRCNN55 | 74.00 | 74.82 | 74.00 | 74.04 |
BERT-MCBMA (comparison) | 74.66+ 0.22 | 75.42 | 74.66 | 74.77+ 0.30 |
ERNIE | 77.64 | 78.49 | 77.64 | 77.91 |
ERNIE-TextCNN56 | 77.02 | 76.79 | 77.01 | 76.85 |
ERNIE-BiLSTM57 | 76.98 | 77.46 | 76.98 | 77.19 |
ERNIE-TextRCNN58 | 76.04 | 76.23 | 76.04 | 75.82 |
BBL(2024)47 | 76.51 | - | - | 72.54 |
BCAM(2024)48 | 77.40 | 74.90 | 74.0 | 74.3 |
ERNIE-MCBMA (ours) | 78.26+ 0.62 | 78.95 | 78.26 | 78.45+ 0.54 |
