Personalized Recommendation via Suppressing Excessive Diffusion

Table 2

Algorithms’ performances on three datasets with . For fair comparison with parameter-free algorithms, the corresponding parameters which achieve the lowest ranking score are chosen with resolution of 0.01 ( for HNBI, for RENBI, for PD, and and for SED in MovieLens; for HNBI, for RENBI, for PD, and and for SED in Amazon; and for HNBI, for RENBI, for PD, and and for SED in RYM). The sampling number in AUC is fixed as . The best values of each metric are emphasized in boldface.

MovieLens		AUC		Recall
CF	0.1225	0.8990	0.0639	0.3856	0.3758	0.5796	243
NBI	0.1143	0.9094	0.0670	0.4048	0.3554	0.6185	235
HNBI	0.1075	0.9144	0.0693	0.4183	0.3392	0.6886	220
CSI	0.0970	0.9278	0.0759	0.4584	0.3315	0.7530	200
RENBI	0.0875	0.9349	0.0812	0.4901	0.3250	0.7923	188
PD	0.0877	0.9341	0.0798	0.4819	0.2902	0.8392	161
SED	0.0873	0.9352	0.0833	0.5031	0.2895	0.8508	158
Amazon		AUC		Recall
CF	0.1212	0.8811	0.0157	0.2711	0.0928	0.8650	82
NBI	0.1170	0.8844	0.0162	0.2795	0.0900	0.8620	82
HNBI	0.1169	0.8843	0.0162	0.2803	0.0896	0.8653	81
CSI	0.1036	0.8936	0.0190	0.3284	0.0881	0.9667	49
RENBI	0.1103	0.8848	0.0181	0.3124	0.0861	0.9245	68
PD	0.1031	0.8935	0.0190	0.3277	0.0855	0.9745	43
SED	0.1024	0.8962	0.0194	0.3356	0.0753	0.9782	41
RYM		AUC		Recall
CF	0.0756	0.9548	0.0130	0.3932	0.1605	0.8216	1114
NBI	0.0675	0.9612	0.0132	0.3989	0.1580	0.7912	1196
HNBI	0.0588	0.9641	0.0132	0.3997	0.1548	0.8114	1155
CSI	0.0463	0.9715	0.0156	0.4731	0.1467	0.8922	869
RENBI	0.0456	0.9701	0.0157	0.4748	0.1527	0.8820	918
PD	0.0440	0.9719	0.0156	0.4719	0.1359	0.9114	734
SED	0.0401	0.9720	0.0166	0.5042	0.1340	0.9214	708