Thursday 23 January 2025
The quest for personalized learning recommendations has long been a challenge in higher education. Researchers have been exploring ways to develop intelligent systems that can learn and adapt to individual students’ needs, but it’s crucial to consider the complexities of domain models and learning contexts. A recent study proposes an innovative approach to address this issue by utilizing large language models (LLMs) for knowledge graph completion.
The researchers created a comprehensive ontology that represents university subjects, linking them to corresponding domain models. This allows for the integration of learning modules from different faculties and institutions into a student’s learning path. The LLM-assisted pipeline extracts fine-grained topics from lecture materials, which are then validated by human experts.
The evaluation results show that the proposed approach can effectively identify relevant course contents and connect them to the domain models. The LLM-assisted extraction of topics and subtopics demonstrated high precision and recall rates, with an average degree centrality increase of 0.13. Although the modularity score decreased slightly, this is attributed to the small sample size.
The study highlights the importance of human-AI collaboration in creating a robust knowledge graph. The LLM serves as a powerful tool for extracting relevant concepts from large amounts of educational content, while human experts validate and refine the results. This synergy enables the development of high-quality learning-path recommendations that cater to individual students’ needs.
The researchers also emphasize the need for diversifying input data to account for differences in lecturers’ recording styles and to ensure a good quality extraction. Furthermore, they suggest using the knowledge graph as a supporting data source for retrieval-augmented generation approaches.
This innovative approach has significant implications for personalized learning recommendations in higher education. By integrating LLMs with human expertise, educators can create more effective learning pathways that adapt to individual students’ strengths, weaknesses, and goals. As the field of educational technology continues to evolve, this study provides a valuable contribution towards developing intelligent systems that support students’ lifelong learning journeys.
Cite this article: “Personalized Learning Pathways through Human-AI Collaboration”, The Science Archive, 2025.
Personalized Learning Recommendations, Higher Education, Large Language Models, Knowledge Graph Completion, Domain Models, Learning Contexts, Educational Technology, Lifelong Learning Journeys, Human-Ai Collaboration, Ontology Development
