Publicly-accessible adaptive systems such as recommender systems present a security problem. Attackers who cannot be readily distinguished from ordinary users may introduce biased data in an attempt to force the system to "adapt" in a manner advantageous to them. Secure personalization is the study of how such public systems can be made more robust in the face of such attacks. We started out looking at collaborative recommender systems and have expanded the research to also include collaborative tagging sytems. This research was supported in part by the National Science Foundation Cyber Trust program under Grant IIS-0430303.
Ramezani, M.; Sandvig, J.J.; Bhaumik, R.; Schimoler, T.; Burke, R.; & Mobasher, B. Exploring the Impact of Profile Injection Attacks in Social Tagging Systems. In Proceedings of the 2008 WebKDD Workshop, Held at ACM KDD'2008 Conference on Knowledge Discovery and Data Mining, Las Vegas, Nevada, August, 2008. (PDF)
Williams, C., Mobasher, B, & Burke, R. Defending Recommender Systems: Detection of Profile Injection Attacks. Journal of Service Oriented Computing and Applications. August 2007. (PDF)
Mobasher, B., Burke, R., Bhaumik, R. & Williams, C. Towards Trustworthy Recommender Systems: An Analysis of Attack Models and Algorithms. ACM Transactions on Internet Technology. 7(2) 2007. (PDF)
The project wiki is for members only.