The ability for computers to interpret and utilise computerised building regulations is crucial for automated compliance checking of buildings, enabling a faster and more objective building permit review. Since natural language regulations are often not designed to be represented formally, integrating formal representations into the regulation drafting process could drive this change. Through in-depth qualitative expert interviews with regulators worldwide, this study examines the drafting process of building regulations. It aims to establish a methodology for developing formal representations in parallel with natural language versions. To be able to do so, the first step is to understand the current workflow and its requirements. The interviewees in this study highlighted several benefits of parallel drafting, including improved regulation quality and reduced inconsistencies. However, they also acknowledged challenges such as the difficulty of modifying existing laws and the significant time and costs required for such changes. This study explores various strategies for integrating formal representations into the drafting process of building regulations. Notably, in many countries, significant progress has been made towards automated compliance checking. However, much of this progress has taken place independently of legislative bodies, limiting the potential of automation in the building permit process. Nonetheless, early efforts by legislators to adopt formal representations are emerging. This study aims to raise awareness of these developments and contribute to the broader understanding of how formal representations can enhance the regulation drafting process.

Recent developments show that large transformer-based language models have the capability to generate coherent text and source code in response to user prompts. This capability can be used in the construction domain to interpret building regulations and convert them into a formal representation usable for automated compliance checking. While base-size models can already be taught to perform semantic parsing with decent quality, this paper shows how Intermediate Representations (IRs) can be used to improve the semantic parsing quality. With reversible IRs, the training time was reduced to almost a quarter of the initial duration, and through adding a hierarchical parsing step, improvements of up to 6.6% on F1 scores were reached. Furthermore, intermediate representations provide a novel and interpretable method towards a human-in-the-loop approach for translating building regulations into a formal representation.