Rethinking Recommender System's Sustainability and Dataset Selection / Joeran Beel (U. Siegen)

Joeran Beel presents his latest research, exploring two critical yet often overlooked aspects of recommender systems: their environmental impact and the adequacy of dataset selection. His study reveals a striking environmental cost of modern recommender systems, showing that deep learning-based recommender systems generate up to 42 times more CO2 emissions than traditional methods—comparable to the carbon footprint of a long-haul flight. This alarming finding calls for a shift towards more sustainable practices in recommender system research. In addition to addressing the ecological footprint, Beel introduces a novel approach to dataset evaluation with the Algorithm Performance Space (APS), a framework that maps dataset diversity and algorithm performance in a multi-dimensional space. By applying the APS to a wide range of datasets, Beel demonstrates that many commonly used datasets, such as those from Amazon, are too homogeneous to provide meaningful challenges for modern algorithms. His work highlights the need for more diverse datasets and directs researchers toward underexplored, unsolved problems that could drive the next wave of innovation in the field.

Bio:
Joeran Beel is head of the Intelligent Systems Group at the University of Siegen. His research focuses on automated machine learning & meta-learning, information retrieval, and recommender systems. He has published more than 140 peer-reviewed publications, is a member of the ACM Recommender Systems Steering Committee, an associate editor and the information director at ACM TORS, and he acted as a reviewer for venues such as SIGIR, ECIR, RecSys, UMAP, ACM TiiS, and JASIST. Joeran Beel has founded multiple award-winning business start-ups and has initiated and contributed to various projects, including Recommender-Systems.com, Docear, TensorFlow, and JabRef. He acquired over 2.5 million Euros in funding for his research and business start-ups.