Detection of Shilling Attack in a Recommender System

In recent times commendation programs have become an integral part of our daily life, from grocery compliments to recommendation movies. People would not like to devote their time to find the best thing on the list according to their needs. The level of recommendations is fundamental for users to recommend anything. The Joint Recommendation Program recommends items for customer engagement and is a widely used and proven way to provide recommendations. Based on user ratings, it recommends that particular item. Since this recommendation program is based on ratings, it is straightforward for attackers to create false profiles and inject biased profiles in very large numbers. These types of attacks are called shin attacks divided into push (hacker attack) and nuke attack (opposite object). This attack is detected using aggregation, separation, element extraction, and possible methods.