Adapting Pedagogical Frameworks to Online Learning: Simulating the Tutor Role with LLMs and the QUILT Model

This paper presents SoloScholar, a novel asynchronous online learning platform designed to deliver personalized education to large undergraduate student groups by integrating established pedagogical frameworks with Large Language Models (LLMs). Leveraging Bloom's Revised Taxonomy, learners are categorized into beginner, intermediate, and advanced levels to personalize the educational content. The platform employs the Questioning and Understanding to Improve Learning and Thinking (QUILT) model to structure tutorial sessions and simulate the teaching role of a tutor at a Higher Education Institute (HEI). To generate contextually relevant and cognitively appropriate questions, carefully crafted prompts are fed into LLMs, synthesizing information from reliable knowledge sources. The efficacy of SoloScholar was evaluated through a dual assessment approach: a technical evaluation using the Retrieval Augmented Generation Assessment (RAGAS) framework and a human evaluation involving academic experts. The technical evaluation demonstrated high scores in answer relevancy and context utilization across all learning levels. Similarly, quantitative responses from academic experts indicated strong satisfaction with the quality and appropriateness of the generated content. The findings underscore the potential of integrating LLMs with pedagogical models to enhance personalization in online education. SoloScholar effectively addresses challenges associated with large-group teaching by providing adaptive learning experiences that cater to diverse cognitive needs.