1The presentation is based on a tutorial on the topic presented at ACM IUI '17 [1]
Even though one of the roots of recommender systems lies in the ifeld of human computer interaction, most of today’s academic research focuses on algorithmic aspects of such systems, for example, on predicting the relevance of items for individual users. At the same time, also in many practical applications only limited means are provided for users to interact with the recommendation system, e.g., to provide feedback on the system-generated suggestions.
The chosen user interface of a recommender can however have a significant impact on its efectiveness and success in practice. Research on how to build more interactive, intelligent user interfaces for recommenders, while not sparse, is somewhat scattered across diferent research subfields. Based on a recent survey work [2], this talk1 aims to provide an overview of existing works on user interaction aspects of recommender systems. The talk covers both aspects of how user preferences can be acquired and how recommendation results can be presented and explained to users, with the goal of increasing the acceptance and efectiveness of such systems. Furthermore, various examples of real-world systems that implement advanced interaction mechanisms are discussed in the talk. Dietmar Jannach is a full professor of Computer Science at TU Dortmund, Germany, where he heads the e-service research group of the department. His research focus is on applying artificial intelligence technology to practical application with a special focus on recommender systems. In 2003, he co-founded a technology startup company that focused on adaptive interactive selling solutions. Dietmar Jannach is the author of numerous scientific publications in diferent fields of AI and one of the authors of the textbook “Recommender Systems - An Introduction”.