Using support vector machines to classify building elements for checking the semantic integrity of building information models

Abstract

Abstract The Industry Foundation Classes (IFC), an open and neutral ISO standard, plays a key role in enabling interoperability, allowing entity and relationship data to be exchanged seamlessly between Building Information Modeling (BIM) applications. However, due to the lack of formal rigidity, data exchanges can often be arbitrary and susceptible to errors, omissions and misrepresentations. This research applied support vector machines (SVM), a technique of machine learning, to check the semantic integrity of mappings between BIM elements and IFC classes. The SVM was trained to distinguish model elements from a dataset of 4187 unique elements collected from six architectural BIM models, based on their geometric and relational features. Using a two staged approach, the SVM was first tested to classify the elements with respect to eight IFC classes. Secondly, the SVM was further tested to distinguish between the element subtypes within individual IFC classes. Results of high accuracy (ACC) and F1 scores in both stages attested to the power and generality of the algorithm. The developed approach provides a way to verify BIM models for data consistency, as well as add semantics required for domain-specific analysis. Practically, the approach is envisioned to be of value for automating the quality checks of BIM deliverables, which is still largely a manual process.