A Comparative Study on Shilling Detection Methods for Trustworthy Recommendations

Abstract

Uncovering shilling attackers hidden in recommender systems is very crucial to enhance the robustness and trustworthiness of product recommendation. Many shilling attack detection algorithms have been proposed so far, and they exhibit complementary advantage and disadvantage towards various types of attackers. In this paper, we provide a thorough experimental comparison of several well-known detectors, including supervised C4.5 and NB, unsupervised PCA and MDS, semi-supervised HySAD methods, as well as statistical analysis methods. MovieLens 100K is the most widely-used dataset in the realm of shilling attack detection, and thus it is selected as the benchmark dataset. Meanwhile, seven types of shilling attacks generated by average-filling and random-filling model are compared in our experiments. As a result of our analysis, we show clearly causes and essential characteristics insider attackers that might determine the success or failure of different kinds of detectors.

Keywords

Recommender system / shilling attack detection / supervised classification / unsupervised clustering / statistical analysis methods

References

[1] Bhaumik R., Williams C., Mobasher B., Burke R.. Securing collaborative filtering against malicious attacks through anomaly detection, 2006

[2] Bhaumik R., Mobasher B., Burke R.. A clustering approach to unsupervised attack detection in collaborative recommender systems, 2011

[3] Bryan K., O’Mahony M., Cunningham P.. Unsupervised retrieval of attack profiles in collaborative recommender systems, 2008.

[4] Buja A., Swayne D., Littman M., Dean N., Hofmann H., Chen L.. Data visualization with multidimensional scaling. Journal of Computational and Graphical Statistics, 2008, 17(2): 444-472.

[5] Burke R., Mobasher B., Williams C., Bhaumik R.. Classification features for attack detection in collaborative recommendation systems, 2006

[6] Cao J., Wu Z., Mao B., Zhang Y.. Shilling attack detection utilizing semi–supervised learning method for collaborative recommender system. World Wide Web Jornal, 2013, 16(5–6): 729-748.

[7] Chirita P., Nejdl W., Zamfir C.. Preventing shilling attacks in online recommender systems. In Binufati A, Lee D (eds.), Proceedings of the 7th annual ACM international workshop on Web Information and Data Management: 67–74, Bremen, Germany, November, 2005, 4: 2005

[8] Gao M., Yuan Q., Ling B., Xiong Q.. Detection of abnormal item based on time intervals for recommender systems. The Scientific World Journal, 2014

[9] Gao M., Tian R., Wen J., et al. Item anomaly detection based on dynamic partition for time series in recommender systems. PloS One, 2015, 10(8): e0135155.

[10] Gunes I., Kaleli C., Bilge A., Polat H.. Shilling attacks against recommender systems: a comprehensive survey. Artificial Intelligence Review, 2014, 42(4): 767-799.

[11] Günnemann N., Gnnemann S., Faloutsos C.. Robust Multivariate Autoregression for Anomaly Detection in Dynamic Product Ratings, 2014.

[12] Hurley N., Cheng Z., Zhang M.. Statistical attack detection. In Bergman L, Tuzhilin A, Burke R, Felfernig A, Schmidt–Thieme L (eds.), Proc. of ACM RecSys, 2009, 2009: 149-156.

[13] Lam S., Riedl J.. Shilling recommender systems for fun and profit. In Feldman S, Uretsky M, Naiork M, Wills C (eds.), Proc. of ACM WWW, 2004, 2004: 393-402.

[14] Lee J., Zhu D.. Shilling attack detection: A new approach for a trustworthy recommender system. INFORMS Journal on Computing, 2011, 21(1): 117-131.

[15] Li C., Luo Z.. A hybrid item–based recommendation algorithm against segment attack in collaborative filtering systems. In Isaev D (ed.) Proceeding of Information Management, Innovation Management and Industrial Engineering, Vol., 2011, 1: 403-406.

[16] Mehta B., Hofmann T., Fankhauser P.. Lies and propaganda: detecting spam users in collaborative filtering. Proceedings of the 12th International Conference on Intelligent User Interfaces: 14–21, Honolulu, Hawaii, USA, 2007

[17] Mehta B., Nejdl M.. Unsupervised strategies for shilling detection and robust collaborative filtering. User Modeling and User–Adapted Interaction, 2009, 19(1–2): 65-97.

[18] Mobasher, B., Burke, R. & Sandvig, J. (2006). Model–based collaborative filtering as a defense against profile injection attacks. In Gil Y, Mooney R (eds.), Proceedings of the 21st National Conference on Artificial Intelligence: 1388–1393, Boston, Massachusetts, July 16, 2017, AAAI Press.

[19] Mobasher B., Burke R., Bhaumik R., Williams C.. Towards trustworthy recommender systems: an analysis of attack models and algorithm robustness. ACM Transactions on Internet Technology, 2007, 7(4): 1-41.

[20] Mobasher B., Burke R. B. R., Sandvig J. J.. Attacks and remedies in collaborative recommendation. IEEE Intelligent Systems, 2007, 22(3): 56-63.

[21] Oh H., Kim S., Park S., et al. Can you trust online ratings? A mutual reinforcement model for trustworthy online rating systems. IEEE Transactions on System, Man, and Cybernetics: System., 2015, 45(12): 1564-1576.

[22] O’Mahony M., Hurley N., Silvestre G.. Collaborative filtering–safe and sound?. Foundations of Intelligent Systems, 2003

[23] O’Mahony M., Hurley N., Silvestre G.. Detecting noise in recommender system databases. In Paris C, Sidner C (eds.), Proceedings of the 11th International Conference on Intelligent User Interfaces: 109–115, Sydney, Australia, January 29–February, 2006, 1: 2006

[24] Williams C.. Profile injection attack detection for securing collaborative recommender systems, 2006

[25] Wu Z., Cao J. M. B., Wang Y.. Mobasher B., Burke R., Jannach D.. Semi–SAD: Applying semi–supervised learning to shilling attack detection. Proceedings of the fifth ACM Conference on Recommender Systems: 289–292, Chicago, USA, 2011

[26] Wu, Z., Wu, J., Cao, J. & Tao, D. (2012). HySAD: A semi–supervised hybrid shilling attack detector for trustworthy product recommendation. In Yang Q, Agarwal D, Pei J (eds.), Proceedings of the 18th ACM SIGKDD International Conference On Knowledge Discovery and Data Mining: 985–993, Beijing, China, August 12–16, 2012, ACM.

[27] Wu Z., Wang Y., Wang Y., Wu J., Cao J., Zhang L.. Spammers detection from product reviews: a hybrid model, 2015

[28] Xia H., Fang B., Gao M., Ma H., Tang Y., Wen J.. A novel item anomaly detection approach against shilling attacks in collaborative recommendation systems using the dynamic time interval segmentation technique. Information Sciences, 2015, 306: 150-165.

[29] Zhang F., Zhou Q.. A meta–learning–based approach for detecting profile injection attacks in collaborative recommender systems. Journal of Computers, 2012, 7(1): 226-234.

[30] Zhang, S., Ouyang, Y., Ford, J. & Makedon, F. (2006). Analysis of a low–dimensional linear model under recommendation attacks. In Efthimiadis E, Dumais S, Hawking D, Jarvelin K (eds.), Proceedings of the 29th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval: 517–524, Seattle, USA, August 6–11, 2006, ACM.

[31] Zhang, S., Chakrabarti, A., Ford, J. & Makedon, F. (2006). Attack detection in time series for recommender systems. In Eliassi–Rad T, Ungar L, Craven M, Gunopulos D (eds.), Proceedings of the 12th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining: 809–814, Philadelphia, USA, August 20–23, 2006, ACM.

[32] Zhang Y., Tan Y., Zhang M., Liu Y., Chua T., Ma S.. Catch the black sheep: unified framework for shilling attack detection based on fraudulent action propagation, 2015

[33] Zhou W., Wen J., Koh S., et al. Shilling attacks detection in recommender systems based on target item analysis. PloS One, 2015, 10(7): e0130968.