A Robust Collaborative Filtering Approach Based on User Relationships for Recommendation Systems

Table 2

Detection results based on user models.

	Precision recall
User models	Attacks
	Attack size = 5%; Filler size = 5%	Attack size = 10%; Filler size = 10%
(Slotted In_Sim,; Slotted L_depd)	0.85; 0.62	0.99; 0.79
(Slotted R_Sim,; Slotted L_depd)	0.96; 0.72	1; 0.82
(Slotted R_Sim,; Slotted In_Sim)	0.95; 0.98	1; 0.95