Semantic ontologies like SAREF provide a standardized framework for interoperability in smart buildings. However, its potential for metadata extraction remains underexplored. This work introduces a novel approach that utilizes SAREF for structuring data to extract metadata from simulation-based IoT data. A data extraction pipeline processes EnergyPlus output and IoT sensor data, identifying hidden relationships among temperature, humidity, CO2 levels, occupancy, and energy consumption. The extracted metadata is mapped into a SAREF-based knowledge graph, enabling semantic reasoning and advanced querying. This framework leverages SAREF for relationship extraction, bridging the gap between simulation data and a semantically rich knowledge graph, laying the foundation for reinforcement learning-based optimization in smart buildings. It also provides a basis for future AI-driven control strategies to enhance energy efficiency and intelligent automation.