Proposing Big Data Architecture for Addressing Dropout Problem in MOOC Platforms

Accessing academic knowledge on Massive Open Online Courses (MOOCs) has made learning more convenient with flexible schedules and a vast array of course options. The common weakness of these platforms, however, lies in the difficulty of controlling learners' behaviour. Noone can know for certain whether the level of engagement during learning is sufficient for learners to fully grasp the knowledge, or whether learners may fail to complete the courses they have enrolled in, leading to dropout behaviour. In this study, we also ap-plied an artificial intelligence model to predict whether current students can complete the course, enabling quick detection of dropout behaviour and timely preventive measures. Based on the proposed architecture, a real-time monitoring, analysis, and management application system for learner behavior can be developed. This empowers course managers to detect which learners might drop out of which courses, enabling timely alerts to learners for adjusting their study plans or, on a broader scale, restructuring the organization of courses with excessively high dropout rates. To maximize scalability with the increasing volume of MOOC data and applicability across different MOOC platforms, our architecture will be built on the Microsoft Azure Cloud computing service, utilizing modern and renowned big data technologies to perform tasks ranging from streaming data collection, batch data processing, distributed processing of large-scale data, to storing data in any format with the latest data lakehouse storage architecture, and real-time data visualization and anomaly detection through integrated models.