Relationship Discovery and Hierarchical Embedding for Web Service Quality Prediction

Table 3

Overall performance comparison of TP on average.

Method	MAE/matrix density						RMSE/matrix density
	0.10	0.15	0.20	0.25	0.30	Average	0.10	0.15	0.20	0.25	0.30	Average
LDA	16.29	15.10	14.02	13.97	13.73	14.6220	58.25	57.2	56.56	55.1	54.55	56.3320
Node2vec	14.85	13.76	13.35	13.19	12.82	13.5940	56.85	55.98	54.15	53.29	51.18	54.2900
GCN	15.25	14.30	14.01	13.83	13.55	14.1880	57.93	57.1	55.97	54.35	53.83	55.8360
Wide & deep	14.98	14.21	14.02	13.87	13.85	14.1860	56.09	55.12	54.4	52.97	51.95	54.1060
GraphSAGE	14.76	14.10	13.66	13.11	12.97	13.7200	55.57	54.04	53.52	52.81	53.19	53.8260
NGCF	14.85	14.20	13.86	13.31	12.85	13.8140	55.06	54.02	53.21	51.56	50.94	52.9580
PinSage	14.69	14.15	13.78	13.23	12.95	13.7600	52.87	51.83	50.3	49.78	48.96	50.7480
KGAT	14.76	14.10	13.66	13.11	12.97	13.7200	51.87	51.33	50.25	49.98	48.85	50.4560
GraphRec	14.18	13.53	12.95	12.86	12.75	13.2540	54.23	53.6	52.15	51.42	50.07	52.2940
NIA-GCN	13.58	13.22	13.05	12.99	12.87	13.1420	52.23	51.73	50.05	49.68	48.77	50.4920
RDHE	12.58	12.49	12.45	12.40	12.17	12.4180	48.26	47.41	46.16	45.32	45.01	46.4300