A dataset of scanned examination answers has been created for research purposes. The dataset includes 485 answer submissions from students at four institutions, paired with expert-designed grading metadata. The data is multimodal, containing various types of content such as handwriting, printed text, equations, and diagrams. The scans were taken using different scanners to introduce variation in factors like illumination and resolution. The dataset can be used for research on rubric-aware automated evaluation, document understanding, and criterion-level feedback. --- Why it matters: This dataset matters because it provides a large-scale, multimodal collection of examination answers that can be used to improve the accuracy of automated grading systems and provide more detailed feedback to students. Source: https://arxiv.org/abs/2608.22346