Artificial intelligence is a persistent topic in the news, driven by enormous research efforts that require vast amounts of electrical energy. One subfield of artificial intelligence - recommender systems - addresses generating personalized recommendations.
Large technology corporations, e.g., Microsoft, Amazon, Google, Netflix, and Spotify, actively research recommender systems. However, analyses of electrical energy consumption in recommender systems research are scarce, have limited methodologies, and do not consider a broader scope. Our research project with Alan Said at the Department of Applied IT aims to quantify and contextualize the carbon footprint of research on recommender systems. In particular, we analyze how energy consumption in recommender systems research increased compared to ten years ago, show how the computer's location impacts carbon footprint, and estimate the carbon footprint generated by a high-ranked recommender systems conference.
Lukas Wegmeth, Tobias Vente
Lukas Wegmeth and Tobias Vente are both third-year PhD students under the supervision of Joeran Beel at the Intelligent Systems Group of the University of Siegen in Germany.
Their research broadly focuses on the automation of recommender systems (AutoRecSys) and the evaluation of recommender systems. In particular, Lukas focuses on the algorithm selection problem, while Tobias focuses on model selection. Lukas and Tobias are visiting Alan Said at the Department of Applied IT for one month to work on a collaborative recommender systems research project.