A Clustering Approach to Unsupervised Attack Detection in Collaborative Recommender Systems

Securing collaborative filtering systems from malicious attack has become an important issue with increasing popularity of recommender systems. Since recommender systems are entirely based on the input provided by the users or customers, they tend to become highly vulnerable to outside attacks. Prior research has shown that attacks can significantly affect the robustness of the systems. To...