MM-ArgFallacy2025 Shared Task: Multimodal Argumentative Fallacy Detection

Join the MM-ArgFallacy2025 Shared Task to detect and classify fallacies in political debates using text and audio.

Join the MM-ArgFallacy2025 Shared Task, co-located with ACL 2025, to tackle the challenges of detecting and classifying fallacies in political debates using both text and audio.

Participants will address two challenges: Fallacy Detection and Fallacy Classification. The task features three input settings: Text-only, Audio-only, and Multimodal (Text + Audio).

The challenge is based on the MM-USED-fallacy dataset, with additional resources available via MAMKit.

• Training Data Release: February 25
• Test Set Release: March 24
• Evaluation Period: April 14–25
• Paper Submission Deadline: May 15
• Workshop: July 31

Visit the official webpage here for full details, and join the MM-ArgFallacy2025 Slack Channel here to stay updated.

Tags: MM-ArgFallacy2025, Multimodal Argumentative Fallacy Detection, ACL 2025, Argument Mining, Political Debates, Fallacy Classification