The International Association for Automation and Robotics in Construction
Publications / CSCE/CRC 2025 - Montreal, Canada
Automating Sustainability Assessment in Construction Contracts Using Machine Learning
The construction industry is a major contributor to environmental degradation, resource consumption, and greenhouse gas emissions. This research proposes an innovative, automated system for the sustainable evaluation of construction contracts by integrating advanced machine learning (ML) techniques and large language models (LLMs). The proposed framework automates extracting, classifying, and evaluating sustainability-related clauses from diverse contract documents. The system transforms contractual text into semantic embeddings stored in a dedicated vector database by leveraging Natural Language Processing (NLP) tools such as LangChain for text segmentation and FAISS for efficient vector retrieval. A locally deployed LLM then processes these embeddings to identify key sustainability factors, including energy efficiency, waste management, environmental impact, and worker safety, providing comprehensive insights in real time. We tested the system on a set of construction contracts, which demonstrated the system's capability to accurately classify sustainability clauses into environmental, social, and economic categories. This capability contributes to a reduction in human error and a significant enhancement in the efficiency of the evaluation process. The results underscore the potential of ML-driven approaches to standardize and enhance sustainability assessments during the contract formulation stage, a critical yet under-explored phase in construction management. Moreover, integrating keyword-scoring mechanisms further refines the system's analytical precision, ensuring that sustainability objectives are clearly defined and met. This research contributes to the development of data-driven sustainable procurement practices and sets the stage for future enhancements, including deeper integration with external sustainability benchmarks and broader industry adoption.