Architecting a Federated Semantic Knowledge Lake Framework: Knowledge Graph-as-a-Platform for Distributed Data Ecosystem

With the exponential growth of heterogeneous data across distributed information systems (IS) and the influx of raw data with descriptions, organizations emphasized building virtual systems dealing with data silos across multiple data sources ranging from structured relational databases to unstructured data. Traditional data integration methods face scalability, freshness, and interoperability limitations. We propose a federated semantic knowledge lake (FSKL) framework built upon ontology-based data access (OBDA) and the FedX federation engines, enabling real-time integration and seamless interoperability across heterogeneous healthcare IS without requiring centralized data migration. The framework establishes semantic data pipelines for unstructured, semi-structured, and structured data sources, transforming them into ontological knowledge graphs. As a result, these are integrated into a federated virtual knowledge graph (FVKG) to enable seamless, real-time data access using SPARQL endpoints. The proposed knowledge-graph-as-a-platform (KGaaP) resolves semantic interoperability and supports service-oriented healthcare applications through SPARQL-based federated querying, promoting dynamic, scalable, and interoperable data ecosystems.