Efficient Feedback and Partial Credit Grading for Proof Blocks Problems
Project Overview
The document explores the innovative use of generative AI in education, particularly through the Proof Blocks tool designed to assist students in constructing mathematical proofs via a user-friendly drag-and-drop interface. This tool offers immediate feedback and assigns partial credit by employing an edit distance algorithm, which effectively measures the discrepancies between student submissions and the correct answers. This method not only enhances the accuracy and efficiency of grading but also facilitates classroom integration, thereby positively impacting thousands of students annually. The findings suggest that such AI-driven tools can significantly improve learning outcomes by providing personalized support and fostering a deeper understanding of mathematical concepts. Overall, the implementation of generative AI in educational settings demonstrates promising advancements in teaching methodologies and student engagement.
Key Applications
Proof Blocks
Context: Used in Discrete Mathematics courses at universities, targeting students learning to write mathematical proofs.
Implementation: Students use a drag-and-drop interface to arrange proof lines according to logical dependencies specified by instructors. The algorithm evaluates submissions against correct solutions using edit distance.
Outcomes: Provides instant feedback, allows for partial credit, and is effective in classroom settings, improving student learning and engagement.
Challenges: The initial algorithm was computationally expensive and not scalable for large classes or complex problems.
Implementation Barriers
Technical Barrier
Initial algorithm lacked the efficiency needed for large classrooms, leading to slow feedback times.
Proposed Solutions: A new algorithm based on minimum vertex cover was developed to optimize performance and allow for rapid grading.
Project Team
Seth Poulsen
Researcher
Shubhang Kulkarni
Researcher
Geoffrey Herman
Researcher
Matthew West
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Seth Poulsen, Shubhang Kulkarni, Geoffrey Herman, Matthew West
Source Publication: View Original PaperLink opens in a new window
Project Contact: Dr. Jianhua Yang
LLM Model Version: gpt-4o-mini-2024-07-18
Analysis Provider: Openai