On the discovery of fake binary ratings
Murat Okkalıoğlu, Mehmet Koç, Hüseyin Polat · 2015
Privacy-preserving collaborative filtering methods promise to preserve privacy of individuals. In general, privacy has two aspects, preserving the rating values of users and masking who rated which items. In this study, we analyze a privacy-preserving collaborative filtering method for binary data referred to as randomized response technique. We develop a method targeting the second aspect of privacy to discover fake binary ratings using auxiliary and public information.