Table 1. Parameter adjustment table.

Parameter type	Optimal	Test range
Word embedding dimension	100	50–300
Character embedding dimension	100	50–300
CNN convolution kernel size	2, 3	2–7
CNN output size	200	100–300
LSTM output size	300	100–300
Learning rate	0.001	0.1–0.001
Minibatch size	20	10–50
Dropout value	0.5	0.5–1

CNN, convolutional neural network; LSTM, long- and short-term memory.