Ziegler, J., & Loepp, B. (2023).
In Palanque, P., & Winckler, M. (Eds.), Handbook of Human Computer Interaction. Cham, Germany: Springer.
Abstract:
Recommendation systems have become an indispensable part of many online platforms such as online shops or social media, designed to facilitate users’ search and decision-making when confronted with a very large or otherwise difficult-to-manage set of choices. A large range of algorithmic methods has been developed in recent years for predicting users’ assessment of so far unseen items. While the accuracy of the predictions has long been the main focus of research, it has increasingly been found that the user interface of a recommender system, and particularly the possibility of interactively influencing the recommendation process have a strong impact on users’ assessment of the recommendations and their overall experience of using the system. In the present chapter, we take a user-centric perspective and discuss relevant design aspects of the human-recommender interaction. We present methods and empirical findings concerning the presentation of recommendations as well visualization techniques for large item sets. A particular focus lies on interactive recommending methods which allow users to control different steps of the recommendation process. Among other methods, critiquing techniques and conversational recommender systems are presented as prominent examples of systems with a high degree of interactivity. Furthermore, we address methods for explaining recommendations which play an increasingly important role in empowering users to understand and assess the system. The chapter concludes with an account of evaluation metrics and user-centric evaluation methods.
Resources: