Privacy-Preserving Context-Aware Recommender Systems: Analysis and New Solutions

Abstract

Nowadays, recommender systems have become an indispensable part of our daily life and provide personalized services for almost everything. However, nothing is for free -such systems have also upset the society with severe privacy concerns because they accumulate a lot of personal information in order to provide recommendations. In this work, we construct privacy-preserving recommendation protocols by incorporating cryptographic techniques and the inherent data characteristics in recommender systems. We first revisit the protocols by Jeckmans et al. at ESORICS 2013 and show a number of security and usability issues. Then, we propose two privacy-preserving protocols, which compute predicted ratings for a user based on inputs from both the user's friends and a set of randomly chosen strangers. A user has the flexibility to retrieve either a predicted rating for an unrated item or the Top-N unrated items. The proposed protocols prevent information leakage from both protocol executions and the protocol outputs: a somewhat homomorphic encryption scheme is used to make all computations run in encrypted form, and inputs from the randomly-chosen strangers guarantee that the inputs of a user's friends will not be compromised even if this user's outputs are leaked. Finally, we use the wellknown MovieLens 100k dataset to evaluate the performances for different parameter sizes.

Privacy-Preserving Friendship-Based Recommender Systems

“…Generally speaking, we validate the hypothesis made in [16], [29] and comprehensively analyse existing and newly designed protocols. In more detail, our contribution lies in three aspects.…”

“…In this paper, we revisit the concept of friendshipbased recommender system and the protocols from [16], [29]. Generally speaking, we validate the hypothesis made in [16], [29] and comprehensively analyse existing and newly designed protocols.…”

“…Later, Tang and Wang [29] pointed out some security flaws in the protocol from [16] and gave improved solutions. All solutions from [16], [29] rely on the hypothesis that friends share more similar tastes than with strangers without any validation.…”

“…All solutions from [16], [29] rely on the hypothesis that friends share more similar tastes than with strangers without any validation. Moreover, some proper analysis is missing for security and implementation results are missing to demonstrate the efficiency.…”

“…• We clarify the single prediction and Top-n protocols from [29] and correct some flaws in their specifications. The clarified single prediction protocol is more efficient than the original one.…”