DETECTION AND MONITORING MULTIPLE SHILLING ATTACK IN ONLINE MANAGEMENT SYSTEMS PREDICATED ON BISECTING K-MEANS CLUSTRING ALGORITHM

While existing methods for detecting shillings attacks in online recommendation system are efficient in detecting individuals’ offenders, they are not as effective at detecting group shilling operations. Using the bisecting K-means clustering technique, we offer a method for detecting coordinated shilling attacks. To begin, we take the ratings for each item and split them into groups based on a predetermined amount of time. Second, we suggest using the proportion of product concentration and usage data to determine the degree of suspicion around potential groupings. Research performed on the Netflix and Amazon data sets validate the superiority of the suggested strategy over the gold standard techniques.