Strategy-specific Preference Elicitation for Group Recommender

Abstract

Group recommender systems propose items to a group of users by taking the preferences of individuals into account. Preference elicitation interfaces in existing solutions mostly use 5-point rating scales and are not tailored for group tasks and the underlying aggregation strategies. There is little work that addresses the design of suitable preference elicitation interfaces for group scenarios. In this paper, we propose, prototype, and evaluate novel user interface concepts that are tailored for aggregation strategies. In total, we introduce 8 solutions which seek to make the underlying strategies more transparent to the users. We present two user interfaces for each selected strategy and compare them in a user study. The results demonstrate that the presented prototypes were well received by most of the participants. Except for one draft solution, most participants agreed or strongly agreed that the proposed user interfaces are suitable for the respective strategy. Moreover, our findings suggest that there is a correlation between the complexity of aggregation strategies and the feedback received by the participants. This implies that it makes sense to hide the underlying logic when using complicated strategies. Furthermore, the results indicate that user interface elements should be tailored to the aggregation strategy.