Massive Open Online Courses (MOOCs) are designed for large-scale participation and open access via the internet. They typically include video lectures, reading materials, quizzes, discussion forums, and assignments. While MOOCs offer open access and self-paced learning, their structure often limits flexibility in how learners engage with individual components. Currently, the smallest independent unit of learning remains an entire course, which may not always align with the needs of learners or educators seeking modular learning experiences.

Although MOOCs may contain high-quality educational materials, these components are often not independently searchable, structured, or interchangeable. The lack of meaningful metadata and interconnectivity between smaller learning units (e.g., quizzes, videos, lecture segments) hinders personalized learning pathways and adaptive course design.
Drawing from research in lecture recommendation systems, we explore how embedding-based retrieval combined with generative AI can identify and interlink related MOOC components. Furthermore, we develop and share web components out of MOOC content to make it more modular, technological agnostic, searchable, and adaptable to diverse learning needs.

Our work focuses on:
-Generating meaningful metadata for MOOC content
-Creating an adaptive learning framework where learners can stack smaller units into personalized study paths.
-Enhancing search and discovery by using retrieval models that understand semantic relationships between course elements.

Implications for open Education
In this presentation, we will share our experiences applying machine learning and web techniques to enrich, structure and offer MOOC content, discussing key challenges in its implementation, with the goal of achieving a more personalized, open, and flexible education system.