A Zero-Shot LLM Framework for Automatic Assignment Grading in Higher Education

TL;DR

The paper tackles scalable assignment grading in higher education by addressing the data demands of few-shot methods and the need for personalized feedback. It introduces a zero-shot LLM-based Automated Assignment Grading (AAG) framework that relies on prompt engineering to evaluate both calculations and explanations without task-specific training. Through experiments on open-ended STAT1011 items and a student survey, the approach achieves strong alignment with human grading (e.g., $r=0.75$ and $r=0.82$) and yields significantly positive, personalized feedback that enhances understanding and motivation. The findings suggest AAG can provide consistent, high-quality feedback at scale and highlight future work in LMS integration, iterative feedback, and broader disciplinary applicability.

Abstract

Automated grading has become an essential tool in education technology due to its ability to efficiently assess large volumes of student work, provide consistent and unbiased evaluations, and deliver immediate feedback to enhance learning. However, current systems face significant limitations, including the need for large datasets in few-shot learning methods, a lack of personalized and actionable feedback, and an overemphasis on benchmark performance rather than student experience. To address these challenges, we propose a Zero-Shot Large Language Model (LLM)-Based Automated Assignment Grading (AAG) system. This framework leverages prompt engineering to evaluate both computational and explanatory student responses without requiring additional training or fine-tuning. The AAG system delivers tailored feedback that highlights individual strengths and areas for improvement, thereby enhancing student learning outcomes. Our study demonstrates the system's effectiveness through comprehensive evaluations, including survey responses from higher education students that indicate significant improvements in motivation, understanding, and preparedness compared to traditional grading methods. The results validate the AAG system's potential to transform educational assessment by prioritizing learning experiences and providing scalable, high-quality feedback.

Figures (6)

• Figure 1: Overview of the Zero-Shot LLM-Based AAG System. The bolded box highlights the core innovation of this study: an AAG system integrating prompt engineering, zero-shot LLM capabilities, and tailored feedback for students. The dashed line box indicates non-complementary input.
• Figure 2: Evaluation prompt for the AAG system. The angle brackets < > indicate placeholders for specific content. The question and answer placeholders can represent multiple questions and answers, respectively, as a single question may include several subquestions and rely on previous answers. This prompt is designed to incorporate all prior context while specifically evaluating the last included question and its corresponding answer.
• Figure 3: AAG system student feedback example.
• Figure 4: AAG system teachers feedback example. (A) and (B) provide summaries based on the student's submission and the AAG score, while (C) presents a summary of the AAG feedback generated using LLM.
• Figure 5: Human and AAG system grading distribution of question 1 (left) and question 2 (right) in STAT1011.