Projects

Project description: The rapid expansion of online and large-scale educational platforms has significantly increased the volume of student assignments, creating a growing demand for efficient and effective grading systems. To address this challenge, our project proposes the implementation of an Automatic Short Answer Grading (ASAG) system, utilizing advancements in Large Language Models (LLMs) to streamline the grading process.

This project enables instructors, irrespective of their familiarity with LLM technology, to utilize the ASAG system for grading free-form textual short answers. The system leverages pretrained models such as OpenAI's GPT variants, as well as specialized locally trained models, to provide accurate, fair, and timely feedback on student submissions. This initiative will not only reduce the grading burden on educators but also enhance the educational experience by delivering immediate and constructive feedback to students, thereby promoting deeper engagement and continuous learning.

Key research questions addressed in this project include the optimization of prompt design, the necessity and structure of training data for LLMs, and the system's ability to function in real-time environments. Additionally, we explore the system's capacity to provide equitable and unbiased feedback across diverse student populations, as well as study human perceptions regarding unreliable systems.

Project codebase: ai-grading.ts, ai-grading.sql

Project description: The CS 101 course, required by most non-CS majors in the Grainger College of Engineering, has historically equipped students with Python and Matlab programming skills essential for solving engineering problems. However, the course's effectiveness has been diluted over time due to an overemphasis on diverse engineering applications and outdated content material, undermining its programming rigor.

In this project, we propose a comprehensive redesign of CS 101,focusing on re-establishing basic programming fundamentals (CS1) during lectures while integrating diverse engineering applications into lab sections and bi-weekly mini-projects. Drawing on qualitative feedback from interviews with over 10 faculty members across both Computer Science and other Grainger Departments, we propose to develop a curriculum that balances CS1 topics with practical engineering applications, create new lecture content reflecting CS1 learning objectives, create auto-graded and randomized questions on PrairieLearn to support mastery learning during homework, enhance the lab sections through Process-Oriented Guided Inquiry Learning, develop mini-projects applied to engineering applications, and design concept inventory questions to measure the impact of our course redesign. Our goal is to ensure that engineering students are better equipped with the computational skills necessary to navigate and innovate within their respective engineering fields.

A built-in web calculator for PrairieLearn

Project description: In the University of Illinois's Computer-Based Testing Facility, there has historically only been physical TI calculators offered as the calculator of choice during various exams. However, the complexity of the functions and the lack of clear documentation on TI calculators often result in students being unsure of how to use them effectively.

Following a Universal Design for Learning approach, we build a web-based calculator inside PrairieLearn that has similar functions to the real TI calculators with a more user-friendly interface. This would allow students to access the same calculator interface both inside and outside the testing environment to gain the necessary familiarity.