PETRA: A Personalized Educational Tutoring Tool for Recursive Assistance Leveraging Multi-model LLMs for Dynamic Programming Learning

Students often struggle to grasp abstract concepts of recursion and dynamic programming as they experience cognitive overload when tracking multiple recursive calls. Additionally, they often lack the personalized feedback needed to identify and correct their misconceptions. This study proposes PETRA...

Full description

Saved in:
Bibliographic Details
Published in2025 15th International Conference on Electrical Engineering (ICEENG) pp. 1 - 6
Main Authors Hewedy, Dalia Walid, Abdelhady, Aya Salama
Format Conference Proceeding
LanguageEnglish
Published IEEE 12.05.2025
Subjects
Adaptation models
Adaptive systems
Codes
Cognition
Dynamic programming
Education
Multi-model LLMs
Personalized Learning
Problem-solving
Recursion
Reviews
Testing
Visualization
Online AccessGet full text

Cover

More Information
Summary:Students often struggle to grasp abstract concepts of recursion and dynamic programming as they experience cognitive overload when tracking multiple recursive calls. Additionally, they often lack the personalized feedback needed to identify and correct their misconceptions. This study proposes PETRA (Personalized Educational Tutoring Tool for Recursive Assistance), a novel framework that utilizes multi-agent large language models (LLMs) to provide an adaptive tutoring system. The framework achieves comprehensive learning support through a coordinated system of six interconnected agents. The workflow begins with the Code Generator transforming the student's English description into executable code, which is then evaluated by the Correctness Scorer. This code undergoes analysis in the Compiler Emulator for potential errors before being processed by the Reasoning Model, which strategically generates hints that guide students without revealing complete solutions. The Visualizer then creates interactive recursion trees to illustrate the solution's behavior, while the Adaptive Orchestrator coordinates these components to deliver personalized exercises and guidance based on the student's progress. This paper provides a comprehensive review of the literature relevant to the subject. Furthermore, we propose a structured framework of an interactive multi-modal LLM for tailored assistance to serve as a foundation for future improvements and iterations.
DOI:10.1109/ICEENG64546.2025.11031311