Authenticating Users of Recommender Systems Using Naive Bayes

摘要

Knowledge Based Authentication (KBA) verifies the credibility of claimed identities by matching various user-related data. Popular recommender systems hold abundant personalized data that are valuable for KBA. This paper studies how to authenticate users with abundant rating data in recommender systems. For this, we propose a measurable user authentication scheme for recommender systems with secure personalized data under the Naive Bayes model. Next, we analyze its usability and security for knowledge sources under possible guessing strategies and experimentally evaluate its performance in real datasets. And the proposed scheme is practical in recommender systems.