Table 3.

Hyperparameter setting.

Hyperparameter	Tuned range	Optimal
Word embedding dimension	[100,200,300]	200
POSa embedding dimension	[10,20,30,40]	20
Entity type embedding dimension	[40,50,60,70,80]	60
GCNb hidden units	[100,200,300]	200
LSTMc hidden units	[100,200,300]	200
Learning rate	[0.1,0.2,0.3,0.4]	0.3
Dropout rate	[0.4,0.5,0.6]	0.5

^aPOS: part-of-speech.

^bGCN: graph convolutional network.

^cLSTM: long short-term memory.