Converting Fire Safety Regulations to SHACL Shapes Using Natural Language Processing

Compliance with fire safety requirements is essential in building design and engineering. As such, fire safety regulations are a core part of building permit regulations. Both research and practice have been addressing fire safety over the past decades, leading to a decline in building-related fire deaths [1]. However, these regulations often come in non-machine-understandable formats, making it hard to perform automated compliance checking on these regulations. This results in fire safety checks being performed rather late in the design process, potentially causing significant delays, cost overruns and health hazards.
Fire safety regulations are typically complex sets of descriptive rules (codes) that often require expert assessment and interpretation. These regulations are highly specific and differ for various locations, building types, functions, and elements. The information necessary to assess the regulations or ensure compliance with them is often not properly digitized or is scattered across different local files by the various stakeholders involved in the design process. Contextualizing the regulations also requires expert knowledge or simulation results, which are difficult to capture in formal representations due to their complexity and the need for a deep understanding of the subject matter. Furthermore, the use of technical terms varies between stakeholders. For instance, the definitions and semantics of core concepts such as ‘height’ or