Recommender systems are widely used, in social networks and online stores, to overcome the problems caused by the large amount of information. Most of these systems use a collaborative filtering method to generate recommendations to the users. But, as in this method users’ feedback is considered for recommendations, it can be significantly erroneous by the malicious people. In other words, there may be some users who open fake profiles and vote one-sided or biased in the system that may cause disturbance in providing proper recommendations to other users. This kind of damage is said to be shiling attacks. If the attackers succeed, the user's trust in the recommender systems will reduce. In recent years, efficient attack detection algorithms have been proposed, but each has its own limitations. In this paper, we use profile-based and item-based algorithms to provide a new mechanism to significantly reduce the detection error for shilling attacks.
