Ecient, Secure, and Privacy-preserving Distributed Hot Item Identification
Lea Kissner, Hyang-Ah Kim, Dawn Xiaodong Song, Oren Dobzinski, Anat Talmy · 2006
Millions of people every day identify many types of popular or widespread items, including interesting entertainment content and malicious network attacks. Unfortunately, much of the information required for these popularity contests has privacy concerns attached to its use. Unless their privacy is preserved, participants may refuse to participate honestly, reducing the quality of the results. Provably ensuring the privacy of participants, while retaining the level of eciency and robustness necessary for a practical protocol, is a serious challenge to privacy-preserving popular item identification. Previous eorts have not obtained the necessary eciency [23] or provide only unproven, heuristic levels of privacy [24, 16, 39, 19]. In this paper, we propose and experimentally evaluate protocols that provably protect the privacy of participants, while adding only a small amount of overhead.