KnowAIDE: a FAIR-Compliant Data Environment to Accelerate AI Research

The increasing complexity and volume of data from heterogeneous sources in scientific research require an infrastructure that ensures data is Findable, Accessible, Interoperable, and Reusable (FAIR). This paper presents KnowAIDE, an Artificial Intelligence and data environment designed to meet these requirements for large-scale research projects. The architecture integrates the iRODS data management system for file-based assets with a FROST-Server implementation of the OGC SensorThings API for time-series sensor data. We describe the system architecture, evaluate its alignment with FAIR principles using standardized assessment tools, and introduce a pre-materialized Semantic Fusion Index to resolve the architectural impedance mismatch between these storage technologies. The infrastructure is deployed and validated within the EDIAQI project, a multi-site study on indoor air quality. The evaluation confirms a high degree of FAIR compliance, and comparative benchmarking demonstrates that the Semantic Fusion Index maintains near-constant retrieval times, achieving a 14-fold speedup over standard API chaining for multi-year datasets. These findings establish the technical viability of the KnowAIDE architecture for managing complex, heterogeneous data in collaborative scientific environments.