[Paper Review] CBIM: A Graph-based Approach to Enhance Interoperability Using Semantic Enrichment

Zijian Wang, Huaquan Ying+2 more|arXiv (Cornell University)|Apr 23, 2023

3D Surveying and Cultural Heritage4 citations

TL;DR

This paper proposes CBIM, a graph-based approach that enhances BIM interoperability by semantically enriching pure building object geometries using machine learning and rule-based methods. The method classifies objects, infers relationships (e.g., hosting, adjacency), computes attributes, and constructs a BIM knowledge graph, successfully enabling reconstruction of a Revit 2022 model from geometries exported from Revit 2023 with full consistency.

ABSTRACT

Interoperability remains a challenge in the construction industry. In this study, we propose a semantic enrichment approach to construct BIM knowledge graphs from pure building object geometries and demonstrate its potential to support BIM interoperability. Our approach involves machine learning and rule-based methods for object classification, relationship determination (e.g., hosting and adjacent) and attribute computation. The enriched results are compiled into a BIM graph. A case study was conducted to illustrate the approach for facilitating interoperability between different versions of the BIM authoring software Autodesk Revit. First, pure object geometries of an architectural apartment model were exported from Revit 2023 and fed into the developed tools in sequence to generate a BIM graph. Then, essential information was extracted from the graph and used to reconstruct an architectural model in the version 2022 of Revit. Upon examination, the reconstructed model was consistent with the original one. The success of this experiment demonstrates the feasibility of generating a BIM graph from object geometries and utilizing it to support interoperability.

Motivation & Objective

• To address persistent interoperability challenges in the construction industry caused by data silos and format incompatibilities.
• To enable semantic enrichment of pure building object geometries without relying on existing BIM metadata.
• To develop a graph-based framework that captures object semantics, relationships, and attributes for improved data exchange.
• To demonstrate feasibility of reconstructing a BIM model across different versions of Revit using only geometric data and semantic enrichment.
• To establish a foundation for lossless data migration and semantic interoperability in AEC workflows.

Proposed method

• Employing a hybrid approach combining machine learning and rule-based logic for object classification from pure geometries.
• Using spatial and topological heuristics to infer semantic relationships such as hosting and adjacency between objects.
• Computing attributes (e.g., function, material) through rule-based inference and geometric context analysis.
• Compiling all enriched data into a structured BIM knowledge graph using RDF/OWL-like graph representation.
• Exporting geometries from Revit 2023 as a starting point for the pipeline.
• Reconstructing a Revit 2022 model using extracted semantic and geometric information from the BIM graph.

Experimental results

Research questions

• RQ1 Can pure building object geometries be semantically enriched to reconstruct a functional BIM model?
• RQ2 To what extent can machine learning and rule-based systems infer semantic relationships (e.g., hosting, adjacency) from geometry alone?
• RQ3 Can a BIM knowledge graph enable accurate interoperability between different versions of Revit?
• RQ4 How well does the reconstructed model preserve semantic and geometric fidelity compared to the original?
• RQ5 What is the feasibility of using a graph-based representation to bridge BIM format and version gaps?

Key findings

• The proposed CBIM framework successfully reconstructed a Revit 2022 model from pure geometries exported from Revit 2023.
• The reconstructed model was found to be consistent with the original in terms of geometry, object hierarchy, and semantic relationships.
• The method demonstrated that semantic enrichment from geometry alone is feasible using a combination of ML and rule-based inference.
• The BIM knowledge graph effectively encoded object types, relationships, and attributes necessary for model reconstruction.
• The case study confirmed the potential of graph-based semantic enrichment to enable interoperability across BIM software versions.
• No loss of structural or semantic information was observed in the reconstructed model, validating the approach’s fidelity.