Keynote Speaker

Dr. Marcel Hauck is a technical project manager at the German public service media organisation ARD, where he works on the development, deployment and evaluation of recommender systems. He is also a professor of media computer science at Mainz University of Applied Sciences and holds a PhD from Johannes Gutenberg University Mainz. We are delighted to have him for a talk on Public Service Media Recommender Systems – Engagement, Public Value, and Efficiency in the Wild at INRA 2025.

Abstract: Public service media (PSM) face the dual challenge of meeting societal mandates while ensuring platform effectiveness. This talk presents empirical findings from an A/B test on ARD Mediathek’s Video-on-Demand recommender system. The results demonstrate that less complex models can match or exceed advanced approaches in user engagement, public value metrics, and operational efficiency, including cost and carbon footprint. The analysis addresses observed discrepancies between offline and online evaluations and discusses practical challenges for PSM recommender systems. To support further research, a real-world dataset from the study is available to the scientific community.