Table 3.

Comparison of performance between our proposed non-local GNN and other classical methods.

Method	Accuracy	Precision	Recall	F-measure	AUC
DC	0.7335	0.4050	0.3470	0.3737	0.5977
CC	0.7150	0.3580	0.3067	0.3304	0.5716
BC	0.7139	0.3550	0.3041	0.3276	0.5699
EC	0.7194	0.3690	0.3161	0.3405	0.5777
LAC	0.7563	0.4630	0.3967	0.4273	0.6299
NC	0.7469	0.4390	0.3761	0.4051	0.6166
PeC	0.7555	0.4610	0.3950	0.4254	0.6288
WDC	0.7630	0.4800	0.4113	0.4430	0.6394
PSGN	0.7614	0.4771	0.4301	0.4524	0.6450
LSTM-AM	0.7340	0.4688	0.5674	0.5134	0.6781
BiLSTM	0.7369	0.4803	0.5742	0.5231	0.6829

Bold values mean best scores.