Incorporating Differential Privacy Protection to a Basic Recommendation Engine

Year 2020, Volume: 9 Issue: 1, 1 - 12, 01.03.2020

Ali Inan

Abstract

Recommendation engines analyze ratings data to suggest individuals new products or services based on their past experiences. However, the set of items that an individual has rated and the ratings on these items are critical for protecting individual privacy. Existing work on the problem focus on overly complicated recommendation engines. In this study, we concentrate on the case of a very simple engine protected with a very strong mechanism. Towards this goal, we incorporate differential privacy to an item-based neighborhood predictor. Empirical analyses over large-scale, real-world rating data indicate the efficiency of our proposed solution. Even at very high levels of protection, the rate of loss in prediction accuracy is below 5%, a reasonable trade-off for privacy protection.

Keywords

Recommendation Engine, Collaborative Filtering, Differential Privacy, Laplace Mechanism

References

• [1] “Booking.com,” http://www.booking.com, accessed: 2019-12-03.
• [2] “The internet movie database,” http://www.imdb.com, accessed: 2019-12-03.
• [3] F. Ricci, L. Rokach, and B. Shapira, “Introduction to recommender systems handbook,” in Recommender systems handbook. Springer, 2011, pp. 1–35.
• [4] Z. Batmaz and H. Polat, “Randomization-based privacypreserving frameworks for collaborative filtering,” Procedia Computer Science, vol. 96, pp. 33–42, 2016.
• [5] A. J. Jeckmans, M. Beye, Z. Erkin, P. Hartel, R. L. Lagendijk, and Q. Tang, “Privacy in recommender systems,” in Social media retrieval. Springer, 2013, pp. 263–281.
• [6] C. Dwork, “Differential privacy: A survey of results,” Theory and Applications of Models of Computation, vol. 4978, 2008.
• .[7] P. Hustinx, “Privacy by design: delivering the promises,” Identity in the Information Society, vol. 3, no. 2, pp. 253–255, 2010.
• [8] F. McSherry and I. Mironov, “Differentially private recommender systems: Building privacy into the netflix prize contenders,” in Proceedings of the 15th ACM SIGKDD international conference on Knowledge discovery and data mining. ACM, 2009, pp. 627–636.
• [9] J. A. Calandrino, A. Kilzer, A. Narayanan, E. W. Felten, and V. Shmatikov, “” you might also like:” privacy risks of collaborative filtering,” in 2011 IEEE Symposium on Security and Privacy. IEEE, 2011, pp. 231–246.
• [10] I. Gunes, and H. Polat, “Robustness analysis of privacypreserving hybrid recommendation algorithm,” International Journal of Information Security Science, vol. 4, no. 1, pp. 13–25, 2015.
• [11] I. Gunes and H. Polat, “Detecting shilling attacks in private environments,” Information Retrieval Journal, vol. 19, no. 6, pp. 547–572, 2016.
• [12] B. D. Okkalioglu, M. Koc, and H. Polat, “Reconstructing rated items from perturbed data,” Neurocomputing, vol. 207, pp. 374– 386, 2016.
• [13] H. Polat and W. Du, “Achieving private recommendations using randomized response techniques,” in Pacific-Asia Conference on Knowledge Discovery and Data Mining. Springer, 2006, pp. 637–646.
• [14] H. Polat and W. Du, “Privacy-preserving collaborative filtering on vertically partitioned data,” in European Conference on Principles of Data Mining and Knowledge Discovery. Springer, 2005, pp. 651–658.
• [15] H. Polat and W. Du, “Privacy-preserving top-n recommendation on horizontally partitioned data,” in The 2005 IEEE/WIC/ACM International Conference on Web Intelligence (WI’05). IEEE, 2005, pp. 725–731.
• [16] M. Okkalioglu, M. Koc, and H. Polat, “A Privacy Review of Vertically Partitioned Data-based PPCF Schemes,” International Journal of Information Security Science, vol. 5, no. 3, pp. 51–68, 2016.
• [17] S. Zhang, J. Ford, and F. Makedon, “Deriving private information from randomly perturbed ratings,” in Proceedings of the 2006 SIAM international conference on data mining. SIAM, 2006, pp. 59–69.
• [18] A. Yargic, and A. Bilge, “Privacy-preserving multi-criteria collaborative filtering,” Information Processing and Management, vol. 56, no. 3, pp. 994–1009, 2019.
• [19] A. Boutet, D. Frey, R. Guerraoui, A. Jegou, and A.-M. Ker- ´ marrec, “Privacy-preserving distributed collaborative filtering,” Computing, vol. 98, no. 8, pp. 827–846, 2016.
• [20] J. Li, J.-J. Yang, Y. Zhao, B. Liu, M. Zhou, J. Bi, and Q. Wang, “Enforcing differential privacy for shared collaborative filtering,” IEEE Access, vol. 5, pp. 35–49, 2016.
• [21] Z. Erkin, M. Beye, T. Veugen, and R. L. Lagendijk, Privacypreserving content-based recommender system. ACM, 2012.
• [22] J. Canny, “Collaborative filtering with privacy,” in Proceedings 2002 IEEE Symposium on Security and Privacy. IEEE, 2002, pp. 45–57.
• [23] V. Nikolaenko, S. Ioannidis, U. Weinsberg, M. Joye, N. Taft, and D. Boneh, “Privacy-preserving matrix factorization,” in Proceedings of the 2013 ACM SIGSAC conference on Computer & communications security. ACM, 2013, pp. 801–812.
• [24] Y. Shen and H. Jin, “Privacy-preserving personalized recommendation: An instance-based approach via differential privacy,” in 2014 IEEE International Conference on Data Mining. IEEE, 2014, pp. 540–549.
• [25] Y. Shen and H. Jin, “Epicrec: Towards practical differentially private framework for personalized recommendation,” in Proceedings of the 2016 ACM SIGSAC conference on computer and communications security. ACM, 2016, pp. 180–191.
• [26] H. Shin, S. Kim, J. Shin, and X. Xiao, “Privacy enhanced matrix factorization for recommendation with local differential privacy,” IEEE Transactions on Knowledge and Data Engineering, vol. 30, no. 9, pp. 1770–1782, 2018.
• [27] H. Zhou, G. Yang, Y. Xu, and W. Wang, “Effective matrix factorization for recommendation with local differential privacy,” in International Conference on Science of Cyber Security. Springer, 2019, pp. 235–249.
• [28] S. Zhang, L. Liu, Z. Chen, and H. Zhong, “Probabilistic matrix factorization with personalized differential privacy,” KnowledgeBased Systems, vol. 183, no. 104864, pp. 1–11, 2019.
• [29] X. Meng, S. Wang, K. Shu, J. Li, B. Chen, H. Liu, and Y Zhang, “Towards privacy preserving social recommendation under personalized privacy settings,” World Wide Web, vol. 22, no. 6, pp. 2853–2881, 2019.
• [30] R. Guerraoui, A.-M. Kermarrec, R. Patra, and M. Taziki, “D 2 p: distance-based differential privacy in recommenders,” Proceedings of the VLDB Endowment, vol. 8, no. 8, pp. 862– 873, 2015.
• [31] E. O. Turgay, T. B. Pedersen, Y. Saygın, E. Savas¸, and A. Levi, “Disclosure risks of distance preserving data transformations,” in International Conference on Scientific and Statistical Database Management. Springer, 2008, pp. 79–94.
• [32] J. Ren, X. Xu, Z. Yao, and H. Yu, “Recommender systems based on autoencoder and differential privacy,” in 2019 IEEE 43rd Annual Computer Software and Applications Conference (COMPSAC). IEEE, 2019, pp. 358–363.
• [33] X. Xiao and Y. Tao, “Output perturbation with query relaxation,” Proceedings of the VLDB Endowment, vol. 1, no. 1, pp. 857–869, 2008.
• [34] “Netflix prize data by kaggle,” https://www.kaggle.com/ netflix-inc/netflix-prize-data/, accessed: 2019-12-03.
• [35] “Neo4j graph platform,” http://www.neo4j.com/, accessed: 2019-12-03.