Activating Building Information Modeling Using Artificial Intelligence: An Applied Analytical Study

Abstract

This study introduces the development of an intelligent, cost-effective, and replicable system for the classification and analysis of Building Information Modeling (BIM) data through supervised machine learning. The primary aim is to enhance the interpretability and functional value of BIM metadata by embedding artificial intelligence (AI) techniques into the design evaluation process. The research focuses on classifying BIM elements using structured attributes—such as dimensions, materials, fire ratings, and load-bearing status—and contextualizing these classifications within specific application domains, including residential, industrial, and healthcare environments. To identify the most effective classification strategy, four machine learning algorithms were evaluated: Logistic Regression, XGBoost, Neural Network (MLP), and Random Forest. Among these, the Random Forest model demonstrated superior performance with 99% accuracy, 0.99 precision, 0.98 recall, and a 0.99 F1-score, and was thus adopted as the core model for the proposed system. Unlike conventional BIM tools that depend on manual labeling, the proposed system autonomously predicts element categories using raw numerical and categorical data, showcasing a practical approach to semantic enrichment and intelligent automation in digital design workflows. The application, developed using Streamlit, features an interactive interface that accepts BIM data in CSV format, processes and classifies elements, assesses compliance with intended use contexts, and calculates associated design risk scores. It also generates simplified 3D-like visualizations to support user comprehension. In addition to classification, the system provides descriptive feedback and actionable suggestions, thereby facilitating informed decision-making during early design stages. By bridging the gap between static, IFC-based BIM data and AI-powered design intelligence, this research presents a novel tool for automated classification, risk evaluation, and context-aware assessment. The findings underscore the feasibility and utility of integrating AI into BIM environments to support more efficient, intelligent, and responsive architectural and structural planning.