Defending recommender systems by influence analysis
Information Retrieval
Hi-index | 0.00 |
Recent research has focus on examining the security of e-commerce collaborative filtering(CF) recommender system. Love/hate attack is one of the most effective model as a nuke attack against the classic user-based CF. In this paper, we examine the effectiveness of Love/hate attack against our topic-level trust based recommendation algorithm that incorporate topic-level trust model into traditional collaborative filtering algorithm. The results of our experiments conducted on well-known dataset show that Love/hate attack is more robust against topic-level trust based recommendation algorithm than against classical user-based CF algorithm.