Table 15. Comparison of proposed SBi-LSTM with deep learning approaches.

Model	Accuracy	Class	Precision	Recall	F1
CNN	0.62	−1	0.62	0.75	0.68
		0	0.64	0.46	0.54
		1	0.58	0.55	0.57
LSTM	0.81	−1	0.85	0.88	0.86
		0	0.82	0.64	0.72
		1	0.75	0.88	0.81
CNN+LSTM	0.62	−1	0.63	0.73	0.68
		0	0.65	0.46	0.54
		1	0.56	0.60	0.58
GRU	0.81	−1	0.84	0.88	0.86
		0	0.83	0.66	0.73
		1	0.74	0.80	0.77
Proposed	0.92	−1	0.94	0.96	0.95
		0	0.89	0.89	0.89
		1	0.91	0.88	0.90