For the first time, a student team has taken first place overall in the RecSys Challenge, outperforming both academic and industry teams. This result was achieved by the Politecnico di Milano team, after three consecutive years as the best academic team.
Coordinated by Maurizio Ferrari Dacrema, researcher at the Department of Electronics, Information and Bioengineering, and supervised by PhD students Andrea Pisani and Michael Benigni, the team is composed of Abdallah Alkhetiar, Luigi Inguaggiato, Nicolò Locatelli, Roberto Manea, Alessio Pizzini, David Ravelli, Gianmarco Schifone, Matteo Vitali, and Andrea Zhang.
The RecSys Challenge 2026 ran from April to June 2026, with its final event held in conjunction with the 20th ACM Conference on Recommender Systems in Minneapolis, Minnesota, from September 28 to October 2, 2026. The challenge was organized by researchers from Sony Group Corporation, Pandora/SiriusXM, Deezer Research, Amazon, Politecnico di Bari, and Maastricht University.
This year’s challenge focused on music recommendation through conversation. At each turn of the dialogue, the system had to interpret the user’s preferences based on the conversation and the available information about the user, select and rank 20 tracks, and generate a natural response consistent with the dialogue.
Participants worked with a large set of music conversations, together with information about the tracks and user interactions. The evaluation considered both the quality of the recommendations and the quality of the generated responses.
A total of 336 teams registered for the competition, submitting more than 9,500 solutions overall. Forty-one teams, including employees of major tech companies such as Samsung, Uber, and LinkedIn, entered the final leaderboard by outperforming the best reference method provided by the organizers. The Politecnico di Milano team described its solution in the paper “Talk Less, Rank Better: A Decoupled Pipeline for Conversational Music Recommendation”.