From quantity take-off to audit-ready decision intelligence: A scientometric and integrative review of cost estimation with Building Information Modeling

Abstract

Building Information Modeling (BIM)-based cost estimation has moved beyond automated quantity take-off toward workflows that combine structured cost data, open data exchange, semantic models, machine learning, provenance, and decision support. This study maps that transition and identifies the reliability gap that separates automated calculation from defensible cost decisions. Three targeted Scopus RIS sets yielded 497 records, 393 unique records, and 147 studies for bibliometric mapping; a broader four-set synthesis identified 1,436 records, 1,299 unique records, and 167 studies for critical review. A claim-level audit of 100 published and verified sources was used to construct the conceptual framework. Publication output rose from 24 papers in 2022 to 41 in 2024, with 27 papers already recorded by 1 August 2026. Keyword co-occurrence revealed four dominant structures around BIM-quantity take-off, integration-machine learning, semantics-3D modelling, and cost estimation-5D BIM, while quality control remained peripheral. The synthesis shows that accuracy alone is insufficient because reliability also depends on completeness, consistency, reproducibility, agreement, traceability, semantic interoperability, and version integrity. The resulting Audit-Ready BIM Cost Estimation (ARBICE) architecture integrates seven layers from evidence grounding to decision assurance. The framework repositions BIM cost estimation as a traceable data-to-decision system and provides testable directions for cross-platform validation, explainable discrepancy diagnosis, and audit prioritization.