Summary of Learning to Love Edge Cases in Formative Math Assessment: Using the Ammore Dataset and Chain-of-thought Prompting to Improve Grading Accuracy, by Owen Henkel et al.
Learning to Love Edge Cases in Formative Math Assessment: Using the AMMORE Dataset and Chain-of-Thought Prompting to Improve Grading Accuracy
by Owen Henkel, Hannah Horne-Robinson, Maria Dyshel, Nabil Ch, Baptiste Moreau-Pernet, Ralph Abood
First submitted to arxiv on: 26 Sep 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: None
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper introduces AMMORE, a dataset of 53,000 math question-answer pairs from Rori, a learning platform used by students in several African countries. The researchers evaluate the use of large language models (LLMs) for grading challenging student answers. In experiment 1, various LLM-driven approaches are used to grade edge cases that a rule-based classifier fails to grade accurately. The best-performing approach, chain-of-thought prompting, achieves an accuracy rate of 92%. Experiment 2 passes grades generated by the best-performing LLM-based approach to a Bayesian Knowledge Tracing (BKT) model, which estimates student mastery of specific lessons. The findings suggest that LLMs can improve grading accuracy and reduce misclassification rates in K-12 mathematics education. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about using big language models to help grade math problems for students. They made a special dataset with 53,000 examples from a learning platform used by students in Africa. The researchers tried different ways of using these language models and found that one method was very good at grading tricky questions correctly. They also tested how well this method worked with another tool that helps teachers understand what their students have learned. The results show that using big language models can help reduce mistakes when grading math problems, which could make it easier for students to learn and grow. |
Keywords
» Artificial intelligence » Prompting
