Leveraging LLMs for Adaptive Course Generation in Artemis: An Automated Pipeline for Personalized Exercise Creation

Introduction: This research introduces a novel approach to adaptive course design by using LLMs to generate personalized exercises that adjust content and difficulty according to user preferences within Artemis.

Methodology:

• Literature Review: Investigate existing LLM-based exercise generation and evaluation frameworks.
• Prompt Engineering: Develop precise strategies for directing the LLM to produce targeted exercises.
• Pipeline Implementation: Create an integrated system for data collection, exercise generation, filtering, and evaluation.
• Feedback Integration: Incorporate instructor evaluations to iteratively improve exercise quality.
• Deployment: Test the pipeline by generating and exporting an updated Artemis course with adaptive tasks.

Back to Theses Topics