Automated formalization and verification of building code requirements using knowledge graph construction

Purpose This study develops and evaluates a methodology for automated formalization and verification of building code requirements using ontology-based knowledge graphs aligned with building information modeling (BIM) data, with the goal of producing executable, traceable, and audit-ready compliance outcomes. Design/methodology/approach A phased, verification-driven pipeline is proposed that combines ontology engineering with the Lemon/OntoLex specification, direct alignment with the Industry Foundation Classes (IFC) ontology, controlled use of compact large language models (LLMs) for linguistically intensive subtasks, and deterministic generation of executable SPARQL Constraints, Rules, and Functions. End-to-end traceability is implemented through PROV Ontology (PROV-O). Findings Quantitative and structurally explicit provisions involving numeric thresholds, spatial relations, and aggregations of BIM-derived properties were translated into executable rules. The proof-of-concept evaluation, conducted on accessibility and selected fire-safety clauses from two building codes of the Republic of Kazakhstan, confirms end-to-end executability. Requirements depending on qualitative design intent or modeling conventions not encoded in IFC (e.g. kitchen equipment arrangement) were deliberately flagged as not formalizable rather than translated into speculative rules. Research limitations/implications The evaluation is illustrative and based on two regulatory clauses; broader empirical validation, quantitative coverage metrics, and cross-jurisdictional generalization remain open. Practical implications The approach supports transparent digital compliance workflows, enables earlier detection of violations during design, and produces auditable evidence suitable for permitting, supervision, and sustainability reporting. Social implications Automated and transparent verification of regulatory requirements can improve accountability and consistency in construction project approvals. By enabling auditable compliance checking directly on BIM data, the proposed approach supports safer, more accessible and more inclusive built environments. The methodology can assist public authorities, designers and developers in identifying non-compliant design solutions earlier in the design process, reducing delays and improving regulatory transparency. In the longer term, such digital compliance infrastructures may contribute to more sustainable urban development by strengthening governance mechanisms and supporting evidence-based decision-making in the construction sector. Originality/value The study contributes a pipeline that combines explicit conceptual–lexical separation, direct IFC alignment, strict confinement of LLMs to interpretation-oriented subtasks, and SPARQL-based executability with provenance traceability. An explicit boundary analysis distinguishes formalizable from non-formalizable requirements–a contribution rarely made explicit in the literature.