The importance of consent in user comfort with personalization

Abstract

Numerous research projects have documented concerns that users have with data commonly used by recommender systems. In this paper, we extend that work by specifically investigating the link between consent, explicitly given, and privacy concern. In a study with 662 subjects, we found that the majority of users would not consent to data from outside systems being used to personalize their experience, and sizable minorities object to even internal system data being used. Among those who said they could consent, found they are often uncomfortable with the data being used if they are not asked to consent, but become comfortable after they can explicitly give their consent. We discuss implications for recommender systems going forward, specifically with respect to incorporating data into algorithms when users are unlikely to consent to its use.