Table 1. Hyperparameter tuning.

BiLSTM layer		TextCNN layer
Parameters	Values	Parameters	Values
Hidden node problem	300	Convolution kernels number	300
Learning rate	0.001	Convolution kernels size	3, 4, 5
Epochs	15	Activation function	ReLU
Batch_size	300	Pooling strategy	1-max pooling
Optimization function	Adam	Dropout	0.5
Loss function	Cross entropy	L2 regularization	Three
Input word vector	BERT