earticle

영어

One of the concerns users have to confront when using an IPTV system is information overload, which makes it difficult for them to find suitable content according to their personal preferences. A recommendation service is one of the most widely adopted technologies for alleviating this problem; these services intend to provide people with referrals of items that they will appreciate based upon their preferences. IPTV users must ensure that their sensitive preferences collected by any recommendation service are properly secured. In this work, we introduce a framework for a private recommender service based on Enhanced Middleware for Collaborative Privacy (EMCP). EMCP executes a two-stage concealment process that gives the user complete control over the privacy level of his profile. We utilise a trust mechanism to augment the recommendation’s accuracy and privacy. These trust heuristic spot users whom are trustworthy with respect to the user requesting the recommendation (target-user). Later, the neighbourhood formation is calculated using proximity metrics based on these trustworthy users. Finally, users submit their profiles in an obfuscated form without revealing any information about their data, and the computation of recommendations proceeds on the obfuscated data using a secure multi-party computation protocol. We expand the obfuscation scope from a single obfuscation level for all users to arbitrary obfuscation levels based on different trust levels between users. In other words, we correlate the obfuscation level with different levels of trust, so the more trusted a target user is, the less obfuscatedcopy of users’ profiles he can access. We also provide an IPTV network scenario and experimentation results. Our results and analysis show that our two-stage concealment process not only protects the users’ privacy, but can also maintain the accuracy of the recommendations.