Network-Based Information Extraction from IFC Files to Support Intelligent BIM Companion (iBcom) Technology

摘要

Architecture, engineering, and construction (AEC) industry possesses a diverse workforce involving multi-stakeholder participation. Building information modelling (BIM) technology appears as a cutting-edge platform for collaboration throughout the construction life-cycle. However, many barriers, such as high cost of implementation, inefficient data interoperability, and lack of skilled personnel have diminished the adoption of BIM technology to be slower than anticipated. One of the hidden and overlooked problems is that the current BIM technology is only able to offer its functionality in a strictly uniform manner regardless of workforce's knowledge levels, information needs, and the level of development (LOD) requirements. To address this gap, we are proposing a new intelligent BIM companion (iBCom) technology that can assist diverse workforce to actively seek and efficiently access and use BIM information through interactive and context-sensitive natural language conversations. The ability to efficiently and accurately retrieve information from BIM models based on user's information needs is critical to support the iBCom technology. This paper thus proposes a novel network-based BIM information extraction approach that automatically extracts building element information from the industry foundation classes (IFC) files. The proposed methodology is based on graph theory which first creates building element networks among principle (parent), associated (children), and terminal nodes of IFC data instances, and extracts the information of a particular building element directly from the terminal nodes of its network. The proposed methodology achieved 100% precision and recall when extracting geometry information from two test case models.