Managing interrelated project information in AEC Knowledge Graphs

In the architecture, engineering and construction (AEC) industry stakeholders from different companies and backgrounds collaborate in realising a common goal being some physical structure. The exact goal is typically not known from the beginning, and throughout all design stages, new decisions are made - similarly to other design industries [1]. As a result, the design must adapt and subsequent consequences follow. With working methods being predominantly document-centric, highly interrelated and rapidly changing design data in a complex network of decisions, requirements and product specifications is primarily captured in static documents. In this paper, we consider a purely data-driven approach based on semantic web technologies and an earlier proposed Ontology for Property Management (OPM). The main contribution of this work consists of extensions for OPM to account for new competency questions including the description of property reliability and the reasoning logic behind derived properties. The secondary contribution is the specification of a homogeneous way to generate parametric queries for managing an OPM-compliant AEC Knowledge Graph (AEC-KG). A software library for operating an OPM-compliant AEC-KG is further presented in the form of an OPM Query Generator (OPM-QG). The library generates SPARQL 1.1 queries to query and manipulate construction project Knowledge Graphs represented using OPM. The OPM ontology aligns with latest developments in the W3C Community Group on Linked Building Data and suggests an approach to working with design data in a distributed environment using separate graphs for explicit facts and for materialised, deduced data. Finally, we evaluate the suggested approach using an open-source software artefact developed using OPM and OPM-QG, demonstrated online with an actual building Knowledge Graph. The particular design task evaluated is performing heat loss calculations for spaces of a future building using an AEC-KG described using domain- and project specific extensions of the Building Topology Ontology (BOT) in combination with OPM. With this work, we demonstrate how a typical engineering task can be accomplished and managed in an evolving design environment, thereby providing the engineers with insights to support decision making as changes occur. The application uses a strict division between the client viewer and the actual data model holding design logic, and can easily be extended to support other design tasks.