›› 2013,Vol. 28 ›› Issue (4): 616-624.doi: 10.1007/s11390-013-1362-0

所属专题: Artificial Intelligence and Pattern Recognition

• Special Section on Selected Paper from NPC 2011 • 上一篇    下一篇

防护推荐系统免遭托攻击

Xiang-Liang Zhang1 (张响亮), Tak Man Desmond Lee1, and Georgios Pitsilis2   

  • 收稿日期:2012-09-10 修回日期:2013-05-02 出版日期:2013-07-05 发布日期:2013-07-05

Securing Recommender Systems Against Shilling Attacks Using Social-Based Clustering

Xiang-Liang Zhang1 (张响亮), Tak Man Desmond Lee1, and Georgios Pitsilis2   

  1. 1. King Abdullah University of Science and Technology, Thuwal 23955-6900, Saudi Arabia;
    2. Faculty of Science, Technology and Communication, University of Luxembourg, Luxembourg
  • Received:2012-09-10 Revised:2013-05-02 Online:2013-07-05 Published:2013-07-05

推荐系统在电子商务中扮演着非常关键的角色,不仅为商家带来了利益,而且为用户提供了合理选择商品的便利.随着推荐系统的广泛应用,不良商家也趁机利用推荐系统的推荐机制来推销自己的产品.托攻击就是其中一种简单有效但又难以检测的攻击.具体地,不良商家通过注册一些假用户作为“托”,假意与正常用户兴趣相投,使得推荐系统误认为“托”与正常用户的喜好一致.在此条件下,“托”极力表现出对某特殊产品的喜爱,推荐系统也会错误地将这些产品推荐给正常用户.正常用户在不知情的状况下购买了这些产品,托攻击的目的就成功达到了.本文研究了如何利用社交网络信息来防护推荐系统,使得托攻击不能得逞.本文提出的两种方法,CluTr 和 WCluTr,其主要思想是利用社交网络中的信任信息来加强用户群内部相同喜好的真实性.实验采用Epinions.com数据来验证所提方法的有效性.实验结果表明本文提出的方法可以大大降低托攻击蛊惑正常用户误选择的概率.

Abstract: Recommender systems (RS) have been found supportive and practical in e-commerce and been established as useful aiding services. Despite their great adoption in the user communities, RS are still vulnerable to unscrupulous producers who try to promote their products by shilling the systems. With the advent of social networks new sources of information have been made available which can potentially render RS more resistant to attacks. In this paper we explore the information provided in the form of social links with clustering for diminishing the impact of attacks. We propose two algorithms, CluTr and WCluTr, to combine clustering with "trust" among users. We demonstrate that CluTr and WCluTr enhance the robustness of RS by experimentally evaluating them on data from a public consumer recommender system Epinions.com.

[1] Lam S K, Riedl J. Shilling recommender systems for fun andprofit. In Proc. the 13th International Conference on WorldWide Web (WWW), May 2004, pp.393-402.

[2] Mehta B. Unsupervised shilling detection for collaborative fil-tering. In Proc. the 22nd National Conference on ArtificialIntelligence, Jul. 2007, pp.1402-1407.

[3] Mobasher B, Burke R D, Sandvig J J. Model-based collabora-tive filtering as a defense against profile injection attacks. InProc. the 21st National Conference on Artificial Intelligence,Jul. 2006, pp.1388-1393.

[4] Su X F, Zeng H J, Chen Z. Finding group shilling in recom-mendation system. In Special Interest Tracks and Posters ofthe 14th WWW, May 2005, pp.960-961.

[5] Zhang S, Chakrabarti A, Ford J, Makedon F. Attack detec-tion in time series for recommender systems. In Proc. the12th ACM SIGKDD International Conference on KnowledgeDiscovery and Data Mining, Aug. 2006, pp.809-814.

[6] DuBois T, Golbeck J, Kleint J, Srinivasan A. Improving rec-ommendation accuracy by clustering social networks withtrust. In Proc. ACM Workshop on Recommender Systems &the Social Web, Oct. 2009.

[7] Sarwar B M, Karypis G, Konstan J, Riedl J. Recommendersystems for large-scale e-commerce: Scalable neighborhoodformation using clustering. In Proc. the 5th Int. Conf. Com-puter and Information Technology, Dec. 2002.

[8] Pitsilis G, Zhang X L, Wang W. Clustering recommendersin collaborative filtering using explicit trust information. InProc. the 5th IFIP WG 11.11 International Conference onTrust Management (IFIPTM), Jun. 2011, pp.82-97.

[9] Truong K Q, Ishikawa F, Honiden S. Improving accuracy ofrecommender system by item clustering. Transactions on In-formation and Systems, 2007, E90-D(9): 1363-1373.

[10] O'Connor M, Herlocker J. Clustering items for collaborativefiltering. In Proc. SIGIR 2001 Workshop on RecommenderSystems, Sept. 2001.

[11] O'Donovan J, Smyth B. Trust in recommender systems. InProc. the 10th International Conference on Intelligent UserInterfaces, Jan. 2005, pp.167-174.

[12] Marsh S P. Formalising trust as a computational concept

[Ph.D. Thesis]. University of Stirling, Apr. 1994.

[13] Mobasher B, Burke R, Bhaumik R, Williams C. Effective at-tack models for shilling item-based collaborative filtering sys-tems. In Proc. WebKDD Workshop, Aug. 2005.

[14] Mobasher B, Burke R, Bhaumik R, Williams C. Toward trust-worthy recommender systems: An analysis of attack modelsand algorithm robustness. ACM Trans. Internet Technol.,Oct. 2007, 7(4): Article No.23.

[15] Cheng Z P, Hurley N. Effective diverse and obfuscated attackson model-based recommender systems. In Proc. the 3rd ACMConf. Recommender Systems, Oct. 2009, pp.141-148.

[16] Cheng Z P, Hurley N. Robustness analysis of model-basedcollaborative filtering systems. In Proc. the 20th Irish Conf.Artificial Intelligence and Cognitive Science, Aug. 2009, pp.3-15.

[17] Ziegler C, Lausen G. Analyzing correlation between trust anduser similarity in online communities. In Proc. the 2nd Int.Conf. Trust Management, May 2004, pp.251-265.

[18] Jamali M, Ester M. TrustWalker: A random walk model forcombining trust-based and item-based recommendation. InProc. the 15th ACM SIGKDD Int. Conf. Knowledge Dis-covery and Data Mining, Jun. 2009, pp.397-406.

[19] Massa P, Avesani P. Trust-aware recommender systems. InProc. the 1st ACM Conference on Recommender Systems,Oct. 2007, pp.17-24.

[20] Ma H, King I, Lyu M R. Learning to recommend with so-cial trust ensemble. In Proc. the 32nd Int. ACM SIGIRConf. Research and Develop. in Inform. Retrieval, Jul.2009, pp.203-210.

[21] Ma H, Lyu M R, King I. Learning to recommend with trustand distrust relationships. In Proc. the 3rd ACM Conferenceon Recommender Systems, Oct. 2009, pp.189-196.

[22] Arthur D, Vassilvitskii S. k-means++: The advantages ofcareful seeding. In Proc. the 18th Annual ACM-SIAM Sym-posium on Discrete Algorithms, Jan. 2007, pp.1027-1035.

[23] Rodgers J, Nicewander A. Thirteen ways to look at the corre-lation coefficient. The American Statistician, 1988, 42: 59-66.
No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!
版权所有 © 《计算机科学技术学报》编辑部
本系统由北京玛格泰克科技发展有限公司设计开发 技术支持:support@magtech.com.cn
总访问量: