Partial Credit Grading

Partial credit grading research aims to automate and improve the assessment of student work, particularly for complex tasks like open-ended questions and coding assignments, reducing grading time and increasing consistency. Current research explores various AI models, including transformer networks, neural additive models, and Bayesian inference methods, to achieve accurate and explainable grading, often incorporating techniques like edit distance calculations and latent space manipulation for image-based assessments. These advancements hold significant potential for improving educational efficiency and fairness, as well as for applications beyond education, such as automated quality control in industrial settings.