Touché-Argument-Images

Motivation

Argumentation is a communicative activity in which reasons are exchanged. In addition to words, images are often used in argumentation, either
to illustrate, to exemplify or to arouse emotions. In this task, we want to investigate how to find suitable images for arguments.

News

You can find the dataset here: https://zenodo.org/records/14258398

Task Description

Given a set of arguments, the task is to return for each argument several images that help convey the argument. A suitable image could depict the argument or show a generalization or specialization. Participants can optionally add a short caption that explains the meaning of the image. Images can be either retrieved from the focused crawl or generated using an image generator.

Additional Information:
Image Retrieval for Arguments - Touché at CLEF 2025

Data

The task data consists of 128 arguments. As document collection we provide a focused crawl of at least 1000 images per argument. Following the idea of the infinite index, we also provide an API for an image generator.

Evaluation methodology

The submitted images are judged by human experts according to defined relevance criteria.

Participant registration

First, you need to register for the CLEF 2025 here. This task is organized together with Touché, so you will find it twice on the registration page, under the Touché and ImageCLEF section. It does not matter which option you select. Submissions are made via Tira, for which you must register separately.

Preliminary Schedule

Nov. 2024: CLEF Registration opened [register]
January-February 2025:Release of full dataset.
2025-05-23: Approaches submission deadline.
2025-05-30: Participant paper submission.
2025-06-10: Peer review notification.
2025-07-07: Camera-ready participant papers submission.
Sep. 2025: CLEF Conference in Madrid and Touché Workshop.

Submission Instructions

Results

CEUR Working Notes

For detailed instructions, please refer to this PDF file. A summary of the most important points:

All participating teams with at least one graded submission, regardless of the score, should submit a CEUR working notes paper.

Citations

Contact

Organizers:

Maximilian Heinrich - Bauhaus-Universität Weimar
Johannes Kiesel - Bauhaus-Universität Weimar
Moritz Wolter - Leipzig University
Martin Potthast - University of Kassel, hessian.Ai, scads.AI
Benno Stein - Bauhaus-Universität Weimar

Questions:
Write a mail to:
maximilian.heinrich@uni-weimar