ECLAIR: A Centralized AI-Powered Recommendations System in a Multi-Node EXAIT System

Abstract

Educational recommender systems are increasingly becoming a core feature of modern educational systems. Often the recommender component of a system is tightly integrated, or might be remotely located without accessing data from other local systems. This paper proposes a framework called ECLAIR in which local educational systems can work and share data with a global recommender system that spans multiple educational institutions. In particular, an AI-driven recommendation system is intricately integrated within a multi-node learning management system. Situated at the intersection of large-scale data analysis and personalized education, ECLAIR efficiently processes heterogeneous data from diverse LMS, while ensuring data security and privacy. The proposed ECLAIR's architecture, data pipeline, and processing mechanisms are explored in detail, focusing on its seamless integration with the existing infrastructure. By leveraging MongoDB change streams and relational databases, ECLAIR guarantees real-time data synchronization, secure storage, and efficient processing. Its unique Ingestor tool transforms selected xAPI data points into a relational table format, bolstering system functionality. The successful launch of ECLAIR serves as a testament to the potential of AI in enhancing personalized learning experiences, improving data security, and bolstering system efficiency. Nevertheless, the paper emphasizes the need for ongoing research, specifically concerning privacy-preserving mechanisms and efficient management of data heterogeneity. It demonstrates ECLAIR's pivotal role in the rapidly evolving landscape of eLearning, and its potential for future advancements, scalability, and adaptability, setting a new precedent for AI-powered eLearning recommendation systems.