Call for Participation: MM-ArgFallacy2025 - Shared Task on Multimodal Argumentative Fallacy Detection and Classification on Political Debates at ACL 2025

11 views

Skip to first unread message

Eleonora Mancini

unread,

Mar 11, 2025, 8:14:12 AMMar 11

to Natural Language Processing DC

MM-ArgFallacy2025 Shared Task

Launch: February 25 | Submission Deadline: April 25

Co-located with ACL 2025 – 12th Workshop on Argument Mining, Vienna

The MM-ArgFallacy2025 Shared Task focuses on detecting and classifying fallacies in political debates using both text and audio. Participants will address two challenges:

Fallacy Detection: Determine whether a given statement (text, audio, or both) contains an argumentative fallacy.
Fallacy Classification: Identify the type of fallacy present.

Why Multimodal?
Fallacies are not only in what is said but also in how they are conveyed. Paralinguistic cues such as tone, emphasis, and rhythm play a crucial role in persuasion and deception.

Input Settings

Text-only: Analyze written arguments.
Audio-only: Examine paralinguistic features.
Multimodal (Text + Audio): Combine both perspectives.

Data
The challenge is based on MM-USED-fallacy, with a version specifically designed for argumentative fallacy detection, along with additional datasets for multi-task learning. All resources will be available via MAMKit.

Key Dates (AoE)

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

For full details, visit the official webpage.
To stay updated, join the MM-ArgFallacy2025 Slack Channel.

Best,

Eleonora Mancini

Ph.D. Candidate in Computer Science and Engineering at Language Technologies Lab, University of Bologna

Visiting Researcher at Sony AI, Barcelona, Spain

External Collaborator at MILA, Québec AI Institute, Montréal, Canada

Email: e.ma...@unibo.it
Webpage: https://helemanc.github.io/

Reply all

Reply to author

Forward

0 new messages