We are pleased to announce the Long-Form Analogy Evaluation Challenge, a shared task focused on automatically evaluating multi-modal analogies for explaining Computer Science concepts. The shared task will be co-located with INLG 2026 (https://2026.inlgmeeting.org/), and participants will have the opportunity to write a paper describing their system and approach.
Submission Deadline: August 10, 2026
Prizes: The top 3 participants will receive $100, $50, and $25, respectively.
About the Challenge
Participants are asked to develop a method that predicts human quality ratings for long-form analogies. Given a concept, its description, a textual analogy, and an accompanying animation video, systems must predict scores across textual and video-based criteria. Submissions are evaluated using rank correlation with human judgments.
Links
Competition: https://huggingface.co/spaces/analogy-evaluation/Analogy-Evaluation-Chal...
Dataset: https://huggingface.co/datasets/analogy-evaluation/challenge-dataset
Baseline code: https://github.com/yurinoviello/analogy-eval-baseline-INLG
We warmly encourage members of this community to participate and share this announcement with colleagues and students who may be interested.
Best regards,
Bhavya
Research Scientist, IBM
Yuri Noviello
PhD Student, Delft University of Technology