Building information model and schema cross-validation using semantics – conceptual framework

Ensuring consistency between Mechanical, Electrical, and Plumbing (MEP) schema drawings and Building Information Models (BIM) is essential for design accuracy and minimizing data discrepancies in construction projects. While BIM provides detailed 3D visualizations of building components, schematic drawings remain crucial for capturing the logical and functional relationships within early-stage designs. However, discrepancies between these two representations often arise, necessitating extensive manual verification. This study introduces a conceptual framework for automated cross-validation between MEP schema drawings and BIM models by leveraging semantic representations. The framework utilizes AI-driven technologies, particularly Large Language Models (LLMs), to extract structured knowledge from both schematics and BIM data, translating this information into machine-readable formats based on the Brick ontology. By integrating semantic web technologies and multimodal processing, the proposed framework effectively identifies inconsistencies in airflow distribution, system connectivity, and performance parameters. This approach significantly enhances the efficiency and accuracy of design validation, minimizes data discrepancies, and fosters interoperability among heterogeneous data sources. Initial findings demonstrate the scalability and effectiveness of semantic-based validation, suggesting substantial benefits for MEP-BIM integration. Future research will extend the framework to additional MEP domains, including electrical and plumbing systems, and further refine AI-based recognition methods.