Online recommender systems have triggered widespread privacy concerns due to their reliance on personal user data for providing personalized services. To address these concerns, some systems have started allowing users to express their preferences before receiving personalized content (i.e., reactive personalization) rather than automatically pushing it to them (i.e., proactive personalization). However, this would mean constant calls for user action, which can adversely affect user experience. One potential solution is to offer users the ability to customize their privacy settings at the outset, thus obviating the need for constant consultation. We conducted a 2 (Personalization: Reactive vs. Proactive) X 3 (Customization of Settings: Absence vs. Action vs. Cue) factorial experiment (N = 299) with a movie recommendation system. Findings show that interface cues suggesting customization enhance user experience, even in the presence of proactive personalization. They also highlight the important role played by negative privacy experiences in the past.
All Science Journal Classification (ASJC) codes
- Human Factors and Ergonomics
- Human-Computer Interaction
- Hardware and Architecture