BEA 2026 Shared Task on Rubric-based Short Answer Scoring for German

A shared task is a competition focused on developing particularly effective models to solve a specific problem. In this case, the task involves evaluating short free-text items from a science context. What makes this unique is that models must understand scoring rubrics in order to successfully evaluate the task. At the end, all participants may submit a paper describing their approach. The competition is being held as part of the ACL Workshop on Innovative Use of NLP for Building Educational Applications.

Since this is an open, collaborative task, participants may use whatever method they deem appropriate: from Large Language Models (LLMs) and Peripheral Language Models (PLMs) to neurosymbolic methods or statistical learning models.

The shared task is organized by EduTec Research Group and TBA Centre at DIPF, in collaboration with IPN | Leibniz Institute for Science and Mathematics Education.

More information on the exact procedure is available here: https://edutec.science/bea-2026-shared-task/

Registration is now open to all interested parties.