AN IMPROVED PRIVACY-PRESERVING COLLABORATIVE FILTERING RECOMMENDATION ALGORITHM

Privacy-preserving collaborative filtering is an emerging web-adaptation tool to cope with information overload problem without jeopardizing individuals’ privacy. However, Collaborative filtering with privacy schemes commonly suffers from scalability and sparseness. Moreover, applying privacy measures causes a distortion in collected data, which in turn defects accuracy of such systems. In this work, the concept of privacy-preserving intensity weight, its measurement and an improved method of similarity calculation are introduced to solve the accuracy decreasing problem of the Randomized Perturbation Techniques (RPT) based recommendation algorithm. A new formula of similarity is proposed which considers both users’ rating similarity and the level of perturbation. Experimental results show that the improved algorithm outperforms the initial one in accuracy without affecting the effectiveness of privacy protection