Exposía is a publicly available research dataset that captures the full, pedagogically grounded process of academic writing and feedback in higher education. The dataset originated from the undergraduate lecture “Introduction to Scientific Work” offered in the winter term 2024/2025 at the Technical University of Darmstadt, Germany, and reflects authentic educational practices in teaching and assessing academic writing.

Exposía comprises student research project proposals (exposés) in multiple stages, including draft and final versions provided as both PDF and LaTeX sources. In addition, the dataset contains rich feedback artifacts generated during the course, such as peer and instructor reviews, inline comments and highlights, comment votes, and structured assessment scores. Where consent was given, these materials are linked across stages, enabling the study of revision processes and feedback.

A key contribution of Exposía is its fine-grained, pedagogically motivated scoring schema, which is applied consistently to both student writing and feedback quality. This allows for systematic analysis of writing competence, feedback practices, and assessment reliability. The dataset is particularly relevant for research in learning sciences, natural language processing, and human-centered AI.