Sentiment Prediction of Textual Data Using Hybrid ConvBidirectional-LSTM Model

Table 4

Comparison of the experimental outcomes obtained from the US airline dataset.

Method	Sentiment class	Precision	Recall		Accuracy (%)
GloVe-CNN-LSTM	Negative	0.99	1.00	1.00	92.79
	Neutral	0.84	0.84	0.84
	Positive	0.79	0.78	0.78
	Weighted average	0.93	0.93	0.93
GloVe-HeBiLSTM	Negative	0.99	1.00	1.00	92.47
	Neutral	0.81	0.84	0.83
	Positive	0.78	0.73	0.75
	Weighted average	0.93	0.93	0.93
Proposed ConvBiLSTM	Negative	1.00	1.00	1.00	93.25
	Neutral	0.86	0.85	0.85
	Positive	0.80	0.80	0.80
	Weighted average	0.93	0.93	0.93