Helping Users Perceive Recommendation Diversity

The recommendation diversity is increasingly being recognized as an important issue in satisfying users’ needs for recommender systems. Various diversity-enhancing methods have been developed to increase diversity while making personalized recommendations to users. However, one crucial issue remains. Could the diversity, as system designers have carefully incorporated, be perceived by users and influence their interaction behaviors? In this paper, we try to investigate whether this issue can be addressed at the interface level. Our goal is to understand design issues that enhance users’ perception of recommendation diversity and more importantly their satisfaction. A within-subject user study was conducted to compare an organization interface, which groups recommendations into categories, with a standard list interface. Our user study results show that the organization interface indeed effectively increased users’ perceived diversity of recommendations, especially perceived categorical diversity. Correlation results reveal that the perceived categorical diversity in recommendation lists has a significant correlation with users’ perceived ease of use of a system, perceived usefulness of the system and attitudes towards the system, thereby resulting in a positive effect on their intention to use the system. We conclude by proposing design guidelines based on our study observations.