Table 11 Comparison with state-of-art techniques
Study | Techniques | Features | P (%) | R (%) | F (%) | A (%) |
|---|---|---|---|---|---|---|
Wei et al. [8] | Machine learning (KNN) | Classical features (tf, idf, tfxidf) BOW | 73 | 71 | 71 | 74 |
Azizan and Aziz [7] | Machine learning (NB) | Classical features (tf, idf, tfxidf) | 70 | 68 | 69 | 72 |
Proposed | Deep learning (LSTM + CNN) | Word embedding | 90 | 86 | 84 | 92 |
Chalothorn and Ellman [17] | Lexicon-based (Senti WordNet) | Sentiment scores and polarity classes | 69 | 68 | 69 | 73 |
Other Models | Deep learning (LSTM) | Word embedding | 85 | 84 | 83 | 85 |
Deep learning (CNN) | Word embedding | 88 | 84 | 83 | 88 | |
Machine learning (SVM) | Tf xidf | 79 | 78 | 79 | 79 | |
Machine learning (RF) | Tf xidf | 83 | 81 | 82 | 84 |