Abstract
Modern buildings generate large volumes of heterogeneous data from building automation systems, sensors, meters, and IoT devices, yet much of it remains underused because of poor interoperability, inconsistent naming conventions, and the absence of explicit context about what data represent and how building entities relate. Ontologies and semantic models address this gap by providing structured, machine-interpretable descriptions of spaces, equipment, sensors, and their interconnections. This enables heterogeneous data to be integrated, queried uniformly across vendors, and reused for data-driven services such as energy management, fault detection, predictive control, and digital twins. From a portfolio perspective, they allow analytics pipelines to be transferred across buildings with minimal adaptation, substantially lowering the cost of scaling smart-building solutions.
This report presents the concepts behind ontologies and knowledge graphs and reviews the current landscape of semantic models for building applications, including Brick, Project Haystack, REC, BOT, SAREF, SSN/SOSA, DBO, ASHRAE Standard 223p, ifcOWL, CIM, EFOnt, and OpenADR. A structured comparison of their modelling approach, maturity, adoption, interoperability, and licensing shows a clear shift from domain-specific data models toward modular, composable semantic ecosystems built on RDF/OWL. No single ontology covers the whole built environment; instead, frameworks are increasingly combined, with topology, system, observation, operational, and grid ontologies playing complementary roles.
The report also examines cross-cutting challenges: interoperability through composability and extensibility, computational scalability for large portfolios (favouring lightweight ontologies and hybrid graph/time-series architectures), AI-assisted and automated generation of semantic models from BIM, metadata, documents, and operational data, and verification and validation using SHACL and emerging AI methods. It concludes that Brick is a very suitable core ontology for buildings. It is mature, widely adopted, open-source, and highly interoperable. It can be deployed as a foundation and is hybridized with aligned ontologies (notably BOT and REC) and progressively extended toward advanced control, demand response, and automated fault detection.