Authors
Santiago de Leon
Behnam Rahdari
Róbert Móro
Peter Brusilovsky
Mária Bieliková
Linked Materials
Abstract
Despite the wide-spread use of multi-list or carousel (Netflix-like) interfaces in e-commerce and streaming services, there is little academic research (less than 30 papers), especially when compared to works for single-list interfaces. Recent eye tracking results have shown that users browse multi-list and carousels significantly differently than other interfaces. Carousels are much more complex, allowing a wide-range of browsing/interaction sequences with multiple topic defined-lists that can be swiped to see more items. To account for this complexity and improve recommendations, recommender systems should be designed specifically for the interfaces they are used on, in other words \textit{interface-aware recommenders}.
This tutorial introduces researchers and industry practitioners to the growing area of interface-aware recommenders. The tutorial provides an introduction to varying interfaces and their impact on user behavior, an overview of the research and insights for improving interface specific systems, the open problems and challenges that have not been addressed, and tools/datasets to help tackle these problems. The tutorial’s goal is to provide a strong basis and tools that participant can use to build improved user-centric systems that are interface-aware and also encourage future research in this area.
Tutorial Outline (90 Minutes Total)
1. Introduction to Interface-Aware Recommenders (5 minutes)
Tutorial background, Motivation, Objectives
2. Overview of Interface Types and User Interaction Differences (20 minutes)
Overview of interface types from single lists to multi-lists and sliders to carousels
Comparison of user browsing and interaction behavior across interfaces
3. State-of-the-Art in Interface-Aware Recommenders and Carousel Research (20 minutes)
(Including 7 papers authored by the instructors)
Summary of works
Human-centric research
User studies
How users browse, perceive, and make decisions
Key findings
Computational research
Click models
Evaluation methods
Modeling and evaluation approaches
4. Challenges and Open Problems in Interface-Aware Recommendation Research (20 minutes)
Understanding how users browse and interact
Factors that impact behavior (e.g., topic preference)
Building user models
Leveraging interface insights for algorithms
Whole-page (multi-list) optimization and evaluation
Handling duplicate items
Interactive discussion
Additional challenges (general or interface-specific)
5. Getting Started with Interface-Aware Recommenders: Tools and Datasets (10 minutes)
CARE: Infrastructure for Evaluation of Carousel-Based Recommender Interfaces
RecGaze Dataset: First eye-tracking and interaction dataset in carousel recommenders
6. Interactive Q&A and Discussion (15 minutes)
Tools and datasets
Additional challenges
Future research directions
Learning Objectives
This tutorial will enable attendees to:
– Understand how interface design impacts user browsing and interaction behavior with concrete examples across a range of interface types
– Gain a foundational understanding of prior work in interface-aware and carousel recommenders along with promising directions for further research
– Apply methods and approaches for determining how a given interface impacts user behavior and decision-making and how to then leverage this information into more intelligent systems
– Identify and reason about key challenges and open problems in interface-aware recommendation and evaluate example solutions in carousel interfaces
– Become familiar with available datasets, tools, and evaluation infrastructures for developing interface-aware systems and realistic users models
Target Audience: Introductory and above
The tutorial is targeted at both researchers and industry practitioners that work with interactive systems and are interested in understanding how the interface impacts user interaction and how to build better systems with this in mind. We specifically focus on the case of recommenders systems (where most of our recent work lies). As we begin from the fundamentals and cover all the work done so far, this tutorial is suitable for students or experienced researchers. As experts in the field, we discuss our insights from recent works, how to improve real-world systems using these insights, future directions of research, and provide an introduction to the tools/datasets that we have developed to help research in the field.
Prerequisites. The only prerequisite is general ( student level or above) knowledge in recommenders. Although we discuss many topics (eye tracking, user studies, click modeling, unbiased learning to rank, page optimization), they will be presented in an easily digestible manner with practical examples in carousel recommenders.
Organizers
Santiago de Leon-Martinez is a Horizon Europe MSCA doctoral student at the Kempelen Institute of Intelligent Technologies. In 2022, he began his PhD project focusing on gaze-based user modeling for interface-aware recommender systems, focusing on carousel interfaces. He (with supervisors Moro \& Bielikova) has 4 works in interface-aware recommenders, particularly he is the first author and creator of the RecGaze Dataset, the largest available feedback (including clicks, cursor movements, gaze, and selection explanations) dataset in carousel recommenders and also the first analysis of eye tracked browsing behavior in carousels. He has been a teaching assistant for 2 undergraduate math courses, presented at ETRA 2023, WSDM 2024, and RecSys 2025 and has been invited to give a number of external talks, such as at the Medical University of Vienna (on psychiatric user models), the Czech Technical University of Prague (on VAE recommenders), and UX and market research company Eye-Square (on designing better carousel interfaces).
Behnam Rahdari is a postdoctoral scholar at Stanford Center for Biomedical Informatics research. He completed his Ph.D. at the University of Pittsburgh under the supervision of Peter Brusilovsky. His research focuses on carousel-based recommender systems, with an emphasis on human-AI collaboration, user navigation, and the evaluation of carousel interfaces. He has authored several publications in this area, including an award-winning work presented at the ACM IUI conference and other contributions at top-tier venues. His doctoral dissertation, titled “Human-AI Collaboration in Search and Recommendation: Navigation, Personalization and Evaluation of Carousel-Based Interfaces,” explored the interplay between user guidance and algorithmic personalization in carousel recommenders. He has presented his work at multiple major conferences, and this tutorial is his first official tutorial presentation.
Robert Moro is a Senior Researcher at Kempelen Institute of Intelligent Technologies (KInIT) and a co-supervisor of Santiago de Leon-Martinez. He previously worked as an Assistant Professor at Slovak University of Technology in Bratislava for 3 years where he taught several courses and was a part of User Experience and Interaction Research Center. In his research on user modeling, he applied eye tracking and other modalities (e.g., mouse tracking) to model user traits or confusion. He has also co-authored a paper on auditing recommender systems that has been awarded the Best Paper award at RecSys 2021. He has presented his work at multiple conferences (including UMAP, IJCAI, EMNLP).
Peter Brusilovsky has been teaching multiple graduate courses over his 25 years as a Professor at the University of Pittsburgh, including “Interactive Systems Design” and “Adaptive Information Systems”. He was a presenter of many relevant tutorials at such conferences as WWW, ACM Hypertext, Adaptive Hypermedia, and User Modeling. Most recently, he presented a tutorial on User Control in Adaptive Information Access at ACM UMAP 2022. He is the supervisor of Behnam Rahdari and has authored several relevant publications in this area.
https://sites.pitt.edu/~peterb/
Maria Bielikova has taught many courses over more than 20 years as a professor at the Slovak University of Technology and given many invited talks. Additionally, she led the personalized web research group (PeWe) and was the director of the User Experience and Interaction Center. She co-funded and currently leads the Kempelen Institute of Intelligent Technologies. She was recently a general co-chair for RecSys 2025 and general co-chair for UMAP 2017. She was the supervisor of Robert Moro, currently is the supervisor of Santiago de Leon and she is a co-author of several relevant publications in this area.