Skip to main content Skip to navigation

Evaluating Automatic Difficulty Estimation of Logic Formalization Exercises

Project Overview

The document explores the role of generative AI in education, focusing on its application in assessing the difficulty of logic formalization exercises within mathematical logic. It underscores the necessity of recognizing student challenges in logic tasks to enhance pedagogical approaches. By utilizing the Grade Grinder corpus, the study evaluates a current algorithm designed for estimating exercise difficulty and identifies additional influential factors, including predicate complexity and pragmatic considerations. The findings indicate that incorporating these variables can lead to a more nuanced understanding of exercise difficulty, ultimately aiming to improve educational outcomes and support tailored teaching strategies. Through this analysis, the document highlights the potential of generative AI to inform and refine educational practices, ensuring they address the diverse needs of learners.

Key Applications

Difficulty estimation module in a tutoring system for teaching First Order Predicate Logic (FOPL)

Context: University-level education, particularly in logic classes

Implementation: The system assigns difficulty levels to formalization exercises based on logic and natural language features and provides feedback to students.

Outcomes: Moderate correlation between predicted difficulty and actual student performance, indicating the algorithm captures some important sources of difficulty.

Challenges: The algorithm does not account for all factors affecting difficulty, such as familiarity with content and pragmatic understanding, leading to misclassifications of exercises.

Implementation Barriers

Technical Barrier

The algorithm's inability to fully capture the complexity of student understanding and difficulty factors.

Proposed Solutions: Incorporate additional features into the algorithm, such as familiarity metrics and considerations for predicate complexity.

Pedagogical Barrier

Teachers may inaccurately assess the difficulty of exercises, leading to mismatches between expected and actual student performance. Use empirical performance metrics to evaluate difficulty, alongside expert assessments.

Proposed Solutions: Implement training for teachers on evaluating exercise difficulty and integrating performance metrics effectively.

Project Team

Alexandra Mayn

Researcher

Kees van Deemter

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Alexandra Mayn, Kees van Deemter

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

Let us know you agree to cookies