TAFM: A Recommendation Algorithm Based on Text-Attention Factorization Mechanism

Table 5

Results of ablation experiment.

Algorithm model	Model accuracy	AUC	Log_loss
TAFM_ae	0.82497	0.69691	0.4266
TAFM_a	0.82490	0.69709	0.4265
TAFM_ae_a	0.82485	0.69743	0.4264
TAFM_l	0.82478	0.69659	0.4272
TAFM_b	0.82573	0.69546	0.4262
TAFM_c	0.82515	0.69596	0.4265
TAFM_l_b	0.82543	0.69676	0.4269
TAFM_l_c	0.82516	0.69707	0.4266
TAFM_b_c	0.82607	0.69815	0.4251
TAFM_b_l_c	0.82051	0.69554	0.4268
TAFM	0.82694	0.73018	0.4144