A new paper published on ArXiv discusses the use of data augmentation and resampling techniques to tackle class imbalance in AI scoring systems for scientific explanations.
The study emphasizes the importance of accurate feedback in educational settings, particularly in Next Generation Science Standards (NGSS) classrooms, where class imbalance can hinder effective assessment.
Published on April 23, 2026, the research aims to enhance the reliability of automated scoring methods, which are increasingly utilized to provide immediate feedback to students.