Academic Journal of Computing & Information Science, 2023, 6(6); doi: 10.25236/AJCIS.2023.060603.
Xinhao Wang
School of Information Engineering, Nanjing University of Finance and Economics, Nanjing, China
With the rapid development of the Internet, recommendation systems have become increasingly important components of various e-commerce and social media platforms. However, recommendation systems also face some security issues, including the problem of shilling attacks. Shilling attacks refer to malicious users who use multiple fake accounts or provide fake reviews to influence the recommendation results and evaluations of the recommendation system, in order to obtain their own interests. The existence of shilling attacks not only affects users' shopping experiences, but also disrupts the normal operation of the recommendation system, affecting the platform's economic benefits.At present, some researchers have attempted to use prototype network methods to solve the problem of underattack detection with low attack filling rate and small attack scale, and have achieved good detection results. However, when faced with more complex underattack data, the effectiveness will deteriorate. We propose an attack detection method based on an infinite hybrid prototype network, and validate our method on real datasets to achieve good results in the face of more complex attacks.
Recommendation system; Shilling attack detection; Variational autoencoder; Supervised Prototype network; Infinite mixture model
Xinhao Wang. Infinite Mixture Prototypical Variational Autoencoder for Shilling Attack Detection in Recommender Systems. Academic Journal of Computing & Information Science (2023), Vol. 6, Issue 6: 16-25. https://doi.org/10.25236/AJCIS.2023.060603.
[1] Snell J, Swersky K, Zemel R. Prototypical networks for few-shot learning[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems. 2017: 4080-4090.
[2] Allen K, Shelhamer E, Shin H, et al. Infinite mixture prototypes for few-shot learning [C]// International Conference on Machine Learning. PMLR, 2019: 232-241.
[3] Vivekanandan K. A Study on Shilling Attack Identification in SAN using Collaborative Filtering Method based Recommender Systems[C]//2021 International Conference on Computer Communication and Informatics (ICCCI). IEEE, 2021: 1-5
[4] Lopez R, Regier J, Jordan M I, et al. Information Constraints on Auto-Encoding Variational Bayes [J]. arXiv e-prints, 2018: arXiv: 1805.08672.
[5] Kingma D P, Welling M. Auto-encoding variational bayes [J]. arXiv preprint arXiv:1312.6114, 2013.
[6] Hjort N L, Holmes C, P Müller, et al. Bayesian Nonparametrics [M]. Cambridge University Press, 2010.
[7] Kulis B, Jordan M I. Revisiting k-means: new algorithms via Bayesian nonparametrics [C]// Proceedings of the 29th International Coference on International Conference on Machine Learning. 2012: 1131-1138.
[8] Chua L O, Roska T. The CNN paradigm [J]. Circuits & Systems I Fundamental Theory & Applications IEEE Transactions on, 1993, 40(3):147-156.
[9] Wang X, Zhao H, Wang Y, et al. Supervised Prototypical Variational Autoencoder for Shilling Attack Detection in Recommender Systems[C]//Data Mining and Big Data: 7th International Conference, DMBD 2022, Beijing, China, November 21–24, 2022, Proceedings, Part II. Singapore: Springer Nature Singapore, 2023: 231-245.
[10] Mobasher B, Burke R. A Survey of Collaborative Recommendation and the Robustness of Model-Based Algorithms [J]. Bulletin of the Technical Committee on Data Engineering, 2009, 31(2): 3-13.