Multi-criteria collaborative filtering recommender by fusing deep neural network and matrix factorization

Journal of Big Data

Table 6 Evaluation results

Dataset	Metric	SVD	SVD++	SlopeOne	Baseline estimates	Single DNN	DMCCF [19]	Our
TripAdvisor	MAE	0.8071 ± 0.0043	0.8222 ± 0.0071	0.9160 ± 0.0053	0.8094 ± 0.0019	0.7855 ± 0.0153	0.7552 ± 0.0050	0.7434 ± 0.0068
	F₁	0.8370 ± 0.0035	0.8390 ± 0.0035	0.8666 ± 0.0018	0.8391 ± 0.0020	0.8572 ± 0.0024	0.8769 ± 0.0043	0.8811 ± 0.0031
	F₂	0.7951 ± 0.0031	0.7991 ± 0.0057	0.9248 ± 0.0011	0.7986 ± 0.0029	0.9176 ± 0.0057	0.9295 ± 0.0014	0.9319 ± 0.0204
	FCP	0.5024 ± 0.0210	0.5092 ± 0.0056	0.5096 ± 0.0661	0.5025 ± 0.0117	0.4957 ± 0.0102	0.5126 ± 0.0160	0.5164 ± 0.0048
	MAP	0.3935 ± 0.0057	0.3483 ± 0.0049	0.3006 ± 0.0026	0.3881 ± 0.0023	0.4481 ± 0.0091	0.4643 ± 0.0072	0.4836 ± 0.0047
	MRR	0.7272 ± 0.0054	0.7513 ± 0.0018	0.7502 ± 0.0037	0.7374 ± 0.0028	0.7244 ± 0.0092	0.7664 ± 0.0039	0.7693 ± 0.0036
Movies dataset	MAE	2.1072 ± 0.0168	2.1594 ± 0.0050	2.1798 ± 0.0112	2.1333 ± 0.0069	2.0613 ± 0.0347	2.0474 ± 0.0119	2.0188 ± 0.0181
	F₁	0.8061 ± 0.0049	0.7985 ± 0.0010	0.7860 ± 0.0026	0.8107 ± 0.0031	0.8344 ± 0.0106	0.8414 ± 0.0041	0.8563 ± 0.0036
	F₂	0.7608 ± 0.0062	0.7553 ± 0.0023	0.7325 ± 0.0045	0.7699 ± 0.0045	0.8256 ± 0.0287	0.8383 ± 0.0049	0.8442 ± 0.0056
	FCP	0.6022 ± 0.0107	0.5985 ± 0.0029	0.6251 ± 0.0098	0.6233 ± 0.0027	0.6226 ± 0.0055	0.6258 ± 0.0083	0.6289 ± 0.0029
	MAP	0.1910 ± 0.0056	0.2144 ± 0.0035	0.2089 ± 0.0067	0.1612 ± 0.0027	0.2136 ± 0.0143	0.2345 ± 0.0109	0.2458 ± 0.0090
	MRR	0.6651 ± 0.0091	0.6502 ± 0.0048	0.6085 ± 0.0048	0.6721 ± 0.0052	0.6922 ± 0.0145	0.7061 ± 0.0078	0.7092 ± 0.0049

Evaluation results for our model against the other algorithms. F₁, F₂, FCP, MAP, and MRR the higher the better while MAE the lower the better