A Shilling Group Detection Framework Based on Deep Learning Techniques

Table 3

Comparison of detection performance for seven methods

	Netflix-1			Netflix-2			Amazon			Movielens 1M
	Precision	Recall	F1-measure	Precision	Recall	F1-measure	Precision	Recall	F1-measure	Precision	Recall	F1-measure
CBS	0.8973	0.8468	0.8713	0.6024	0.5718	0.5867	0.8793	0.8353	0.8567	0.8862	0.8396	0.8623
DCEDM	0.9838	0.5627	0.7159	0.7053	0.4599	0.5568	0.8637	0.7221	0.7866	0.9785	0.5869	0.7337
TP-GBF	0.9968	0.7926	0.8830	0.0124	0.0070	0.0089	0.9283	0.6467	0.7623	0.9856	0.8327	0.9027
GD-BKM	0.9044	0.9712	0.9366	0.7963	0.8052	0.8007	0.8234	0.6732	0.7408	0.2469	0.7683	0.3737
CoDetector	0.4328	0.3473	0.3854	0.4624	0.3094	0.3707	0.8056	0.8345	0.8198	0.1972	0.2043	0.2007
SA-GraphSAGE	0.9823	0.9446	0.9631	0.9023	0.8527	0.8768	0.8327	0.8652	0.8486	0.9756	0.9587	0.9671
SA-GPSA	0.9991	0.9992	0.9991	0.9861	0.9726	0.9793	0.9299	0.8907	0.9099	0.9983	0.9865	0.9924