Project Abstract

All online multiplayer games face the issue of toxicity between players in one form or another. Competitive and esports titles allow players to compete against one another, often grouped in teams. For these teams to be able to collaborate with each other, a greater level of communication is needed compared to non-competitive games. This often takes the form of a text chat, or a voice chat system to allow the teams to communicate with one another, and even sometimes to the opposing team(s). The unfortunate truth is, when a form of communication is granted to players, there will inevitably be bad actors that aim to abuse it. Players may get angry or upset at themselves or others, and this can materialise into toxicity or abuse towards other players.This project aims to look at the usage of machine learning and natural language processing to detect toxicity through the analysis of real game data from Valve Software?��s ?��Counter Strike?��, and to evaluate its effectiveness and viability with feedback from real people. In the process, we look at the methods that are currently employed in existing systems across other eSports titles, to further assess viability. The overall objective of this paper is to contribute to the effort of tackling toxicity in gaming, or perhaps even on the wider internet, which will hopefully create a safer online environment in the future for all.

Keywords: Natural Language Processing, Artificial Intelligence, Esports

 Conference Details

Session: Poster Session A at Poster Stand 9

Location: Sir Stanley Clarke Auditorium at Tuesday 7th 13:30 – 17:00

Markers: Betsy Dayana Marcela Chaparro Rico, Fabio Caraffini

Course: BSc Computer Science, 3rd Year

Future Plans: I have a job lined-up