Mitigate transcription failures through image-quality supports: quality review before submission, preview transcriptions, and multiple submission modes¶

Strategy · All strategies
Evidence · no claims cited

Description¶

Because most grading errors stem from transcription rather than rubric misapplication, the article recommends mitigations targeting image capture and preprocessing. These include "encouraging quality review before submission or providing preview transcriptions," adding direct screenshot uploads alongside webcam and mobile photo capture, and LLM-controllable rotation tools as future work. The article reports the team has already added direct screenshot uploads supporting three submission modes.

Design Implications¶

Context¶

Requirements¶

• Students must be able to preview their submission to ensure readability and crop the image to relevant work

Constraints¶

• LLM-controllable rotation tools increase token usage and computational cost
• Additional data collection is needed to assess improvements in transcription accuracy

Target Learners¶

• university STEM students submitting photographed handwritten solutions

Target Learning Goals¶

• reducing transcription errors in automated grading of handwritten mathematics

Related Strategies¶

• Mitigate extraction errors by requiring high-resolution PDFs, equation-editor equations, and legible graph axes, units, and legends

Examples¶

-

Key Sources¶

• Jacob Levine, Miguel Aenlle, Craig Zilles, Matthew West, and Mariana Silva. (2026). Automated Grading of Handwritten Mathematics Using Vision-Capable LLMs. https://arxiv.org/abs/2605.19043