Data Atomization: A Framework for On-Demand Association and Access Control of Sensitive Data

Traditional data models often emphasize structure over context-dependent reality, leading to a loss of control by data creators. Consequently, protecting and providing on-demand access control to sensitive data remains a critical challenge in modern data management. This paper proposes Data Atomization, a framework to address these challenges by decoupling data from predefined structures, as seen in traditional data models, such as relational databases. Instead of rigid associations, our approach uses flexible ontologies to define data context, making it more adaptable. These ontologies can exist at different levels as organizational, regional, or authority-based, allowing data elements to be interconnected. This ensures data is only linked to its intended use cases, enabling independent growth of both data and its structure. By leveraging ontological querying, data atomization, and hybrid encryption, our framework provides a secure and dynamic way to manage sensitive data with context-specific access. Unlike existing models for data privacy preservation, our approach differs in representation, process, and data residency.