Research on Privacy-Preserving Collaborative Filtering Recommendation Based on Distributed Data
Chang Hui · Chinese Journal of Computers · 2006
Privacy-preserving data mining is a cutting-edge research direction in recent years. As one of its sub-directions, privacy-preserving collaborative filtering aims at protecting users' privacy while providing high-quality recommendations efficiently. To reserve privacy in collaborative filtering recommender systems under distributed data scenario, the core challenge——how to securely rate a specific item——is addressed. A protocol employing commutative encryption as its major privacy-preserving technique is introduced. This protocol produces the same results as the traditional memory-based collaborative filtering recommender systems while preventing any user’ ratings from being known by other sites rather than by itself. Based on secure multi-party computation and random oracle model, the protocol's security is proved. The protocol's computation complexity and communication costs are analyzed as well.