
Politecnico di Milano’s team, coordinated by Maurizio Ferrari Dacrema, Cesare Bernardis and Fernando Benjamin Perez Maurera of the Dipartimento di Elettronica, Informazione e Bioingegneria, won the first place of the academic part of the RecSys Challenge 2021.
The RecSys Challenge 2021 was organized by Politecnico di Bari, ETH Zürich and Jönköping University with the involvement of Twitter, which provided the data set. The challenge focused on predicting the engagement of tweets in a dynamic environment. For this year's edition, four different types of engagement were considered: likes, retweets, quotes and replies.
This year's challenge has come even closer to Twitter's true recommendation systems, introducing latency constraints. The size and density of the data has also been increased in order to encourage the development of faster methods.
The goal was twofold: to predict the likelihood of different types of engagement of a target user for a series of tweets based on heterogeneous input data, while providing fair recommendations.
Politecnico di Milano’s team will illustrate the approach followed in a paper that will be presented at the official workshop to be held in Amsterdam on the occasion of the 15th ACM RecSys Conference, scheduled from 27th September to 1st October 2021.