Metadata-aware RAG for Enforcing Access Control and Metadata-based Filtering: Proof of Concept and Evaluation
Abstract
Organizations are integrating Large Language Models with internal knowledge bases to enhance knowledge retrieval and decision support. However, applying Retrieval-Augmented Generation systems in domains with sensitive data presents challenges such as weak or insufficient access control, inadequate metadata filtering, and lower precision or answer relevancy in the retrieved responses. For example, without proper access control mechanisms, sensitive patient information could be inadvertently exposed to unauthorized users. Furthermore, inadequate metadata filtering can lead to the retrieval of irrelevant or incomplete information, which reduces the accuracy and relevance of the response. This paper proposes a novel framework that integrates Metadata Filtering with fine-grained Role-Based Access Control to ensure both secure and accurate retrieval. By allowing metadata to serve as both a filtering mechanism and an enforcement tool, the system narrows search results while maintaining compliance and accountability. A proof-of-concept implementation using patient profile data demonstrates its effectiveness in secure AI-driven knowledge management.
Read the paper
Related studies
- Knowledge Management Perspective of Generative Artificial Intelligence
- Navigating Generative Artificial Intelligence Promises and Perils for Knowledge and Creative Work
- Characterizing the evolving research on enterprise content management
- Large language models and multimodal foundation models for precision oncology
- From Generative AI to Extended Reality: Multidisciplinary Perspectives on the Challenges, Opportunities, and Future of Educational Computing
- Generative Artificial Intelligence: A New Engine for Advancing Environmental Science and Engineering
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.