Tracking Abuse on Twitter Against Football Players in the 2021 – 22 Premier League Season

Abstract

Online abuse against prominent sportspeople, particularly football players, has become a growing concern in recent years. This report analyses 2.3 million tweets directed at Premier League Footballers during the 2021/2022 season using machine learning. It also includes a manual review of 3,000 tweets by experts, helping shed light on the dynamics and patterns of online abuse experienced by players. The study found that most tweets directed at Premier League Players were positive, and only a 2.6% of tweets contained abusive content. The analysis shows that the abuse tends to peak after key events, such as particularly heated games, and that only a small proportion of players receive the majority of abuse online. This research was commissioned by Ofcom and produced by The Alan Turing Institute.

Keywords: online abuse, social media, natural language processing, machine learning