Graph representation of BIM models

Abstract

Even with modern BIM software, design collaboration across disciplines remains sequential and siloed. The prevailing file-based approach to information exchange cannot support advanced functions such as proactively maintaining design consistency. This PhD study aims to explore an object-based approach of representing building information as graphs and implementing intelligent applications on BIM graphs for enhancing collaboration and easing interoperability issues. Specifically, domain-specific models are compiled as BIM graphs with the supplementary of implicit information by semantic enrichment. BIM graphs are then properly linked with topology and design intent relationships by artificial intelligence (AI) techniques. Intelligent applications, such as across-domain consistency maintenance, are applied to the linked BIM graphs for tracking and processing design changes automatically. This research validates that graph representation is a suitable approach for storing and linking BIM models, and that intelligent functions can be achieved on the linked graphs to improve the data exchange and collaboration efficiency.