Table 7.

The results of discussion about knowledge attention mechanisms. “M” means $\times 10^6$. Batch time means the runtime of each batch in the model testing.

Method	Accuracy	Precision	Sensitivity	Specificity	F1-score	AUC	The model size of knowledge attention mechanisms	Batch time
Ours (DeepWalk [97])	0.8463	0.8549	0.8506	0.8867	0.8528	0.8507	23M	0.40 s
Ours (Esim [98])	0.8488	0.8567	0.8545	0.8869	0.8556	0.8552	24.3M	0.35 s
Ours (metapath2vec [99])	0.8499	0.8628	0.8554	0.8912	0.8591	0.8568	24M	0.34 s
Ours (HERec [100])	0.8523	0.8637	0.8638	0.8938	0.8638	0.8658	23.75M	0.32 s
Ours (GCN [101])	0.8666	0.8701	0.8703	0.8991	0.8702	0.8678	25.4M	0.46 s
Ours (GAT [25])	0.8670	0.8715	0.8724	0.8992	0.8719	0.8766	25.4M	0.97 s
Ours (MAGNN [102])	0.8690	0.8764	0.8736	0.8992	0.8750	0.8772	37.2M	1.48 s
Ours (RGCN [103])	0.8716	0.8716	0.8714	0.8913	0.8814	0.8744	44M	1.24 s
Ours (GATNE [104])	0.8736	0.8726	0.8734	0.8955	0.8839	0.8764	44.6M	1.54 s
Ours (HGAN [105])	0.8765	0.8746	0.8745	0.8960	0.8852	0.8772	47.3M	2.27 s
Ours (HetGNN [106])	0.9336	0.8764	0.9090	0.9344	0.9045	0.9459	42.1M	1.35 s
Ours (HGT [107])	0.9604	0.8764	0.9444	0.9331	0.9039	0.9536	41M	1.50 s
Ours (MMGCN [108])	0.9633	0.8764	0.9466	0.9371	0.9057	0.9539	46.7M	2.13 s
Ours	0.9810	0.9889	0.9861	0.9859	0.9875	0.9908	39.4M	1.14 s