Introduction
In the Architecture, Engineering, Construction, and Operation (AECO) industry, defining "Information Requirements" remains a significant bottleneck. While Building Information Modeling (BIM) aims for seamless data exchange, the actual requirements—specifying what data is needed, by whom, and at what project stage—are frequently trapped in semi-structured formats like PDF or Word documents. This lack of machine-readability prevents the automated validation of project deliverables. This paper introduces a harmonized ontology network designed to transform these requirements into machine-readable Linked Data by aligning three pivotal international standards.
Normative Framework and Methodology
The research addresses fragmentation in the industry by bridging the gap between high-level project requirements and granular technical properties. The authors analyzed three core standards to create a unified framework:
- ISO 7817 (LOIN): Specifies the "Level of Information Need."
- ISO 23387 (Data Templates): Defines the structure for construction objects.
- ISO 23386 (Properties): Provides a methodology for describing properties in data dictionaries.
The development followed the LOT (Linked Open Terms) methodology and adhered to FAIR (Findable, Accessible, Interoperable, Reusable) principles. The process involved creating a new ontology for Data Templates (DT) and updating existing ones for LOIN and Properties (ISOProps).
The Ontology Network Architecture
The core innovation lies in the formal alignment of these ontologies. A critical technical solution was the introduction of a "pseudo-root" element, dt:LibraryComponent. This element ensures that subclasses across different domains—such as construction objects or specific properties—inherit necessary references to external dictionaries and documents without redundant coding.
By establishing these links, the network connects high-level project milestones defined in LOIN to specific physical quantities in Data Templates, and finally to standardized definitions within a property dictionary. Notably, the LOIN specification is designed to be "agnostic," supporting various classification systems like IFC or LandXML.
Validation and Results
The resulting ontology network was validated using a demonstration dataset focused on fire safety regulations for partition walls. The authors demonstrated that the network could be effectively queried using SPARQL.
In a practical scenario, a structural engineer (the "sending actor") could execute a single SPARQL query to retrieve all alphanumerical requirements for a partition wall at a specific project milestone. For a "non-load-bearing partition wall" at the "detailed design" stage, the query successfully identified required properties such as Wall_Thickness and Fire_Resistance, both linked to the European standard DIN EN 1992-1-2.
Conclusion and Future Work
While the ontology network provides a robust formal framework for automating information exchange, several challenges remain. Currently, the validation is limited to a single fire safety use case, and the reliance on SPARQL queries makes the system accessible primarily to experts.
Future research will focus on:
- Broadening Scope: Applying the network to a wider range of construction objects to identify edge cases.
- User Accessibility: Developing user interfaces to make the technology accessible to non-technical end-users.
- BIM Integration: Creating interfaces for common BIM modeling environments to enable real-time requirement checking during the design process.
The ontologies and demonstration data are available as open-source deliverables on GitHub.
Summary updated