OpenRAG: Open-source Retrieval-Augmented Generation Architecture for Personalized Learning

This paper introduces OpenRAG, an open-source Retrieval-Augmented Generation (RAG) system architecture designed to enhance GenAI applications in personalized learning. The architecture is modular with loosely coupled components: Generator, User Interface, Indexing subsystem, Retriever, and Orchestration module. The research applies cutting-edge design patterns for retrieval, generation, indexing, storage, user interactions, and performance optimization, whereas the data flow and processing pipeline are seamlessly integrated for customization and adaptability. A case study demonstrates a proof-of-concept implementation of OpenRAG used in an online learning platform, resulting in significant increases in learner engagement and education efficiency. Key design concerns such as computational efficiency, content accuracy, learning styles, and user experience are discussed along with their solutions.