Effective Analysis of Recommender Systems

Recent research has revealed an important tendency for affiliate program programs based primarily on the emergence of profile injections, where malicious users add false profiles to the test record in order to bias the program outcome. “To reduce this risk, a variety of clever ways to detect such types of attacks have been considered. While current diagnostic methods may recognize the normal number of such attacks well, they do have negative effects when they feel the recent effects of these types of attacks. The most important example is the average threat over popular items. Based on this problem, this paper sheds light on various emerging strategies that are helpful in detecting these types of such attacks in real-world situations.