University of Zurich

The Pleasure(s) of Griefing: Tracing Multimodal Toxicity in League of Legends

The paper examines multimodal toxicity in League of Legends, arguing that hostile communication in competitive online play is not confined to text chat but emerges from the combined use of pings, emotes, UI interactions, champion positioning and chat messages.

• Toxicity in League of Legends is multimodal: hostile signals emerge from combined pings, emotes, UI interactions, champion positioning and chat rather than from text chat alone, so text-only classification misses much of it.
• Toxic episodes tend to appear as clusters across channels, and understanding them requires temporally synchronized, context-aware analysis instead of surface-level classification.
• The paper's main contribution is MuMoDeTox, a prototype pipeline combining YOLOv8-based detection of pings and champion positions from video frames, minimap localization, OCR of chat and on-screen text, and external match metadata such as role and performance statistics.