SPAM DETECTION USING MACHINE LEARNING TECHNIQUES IN AN ONLINE RECOMMENDATION SYSTEM

The detection of group shilling attacks, in which a group of attackers conspires to skew the results of an online recommender system by injecting fake profiles, is rarely addressed by existing shilling attack detection approaches, which are primarily focused on systems. This article suggests a method for Spam detection using machine learning techniques in an online recommendation system. First, we separate the rating tracks of each item to create candidate groups that correspond to a 1950s time period. Second, we suggest using user activity and item attention levels to determine how suspicious a group of candidate individuals is. Finally, we use the Machine Learning algorithm to sort the candidate groups into attack groups based on how suspicious they are. Experiments on the Netflix and Amazon data sets have shown that the proposed strategy performs better than the standard methods