AI-augmented construction cost estimation: an ensemble Natural Language Processing (NLP) model to align quantity take-offs with cost indexes
Peyman Jafary, Davood Shojaei, Abbas Rajabifard, Tuan Ngo
Abstract
Accurate construction cost estimation is crucial for the financial success of construction projects. Effective financial management in construction heavily relies on a robust and reliable estimation system. Building Information Modeling (BIM) has emerged as a powerful tool, providing precise quantities for various building elements through Quantity Take-Offs (QTOs). Traditionally, Quantity Surveyors (QSs) match these QTOs with cost indexes, a task that is both labor-intensive and prone to errors due to subjectivity and inconsistencies in classification systems and software structures. This paper presents an ensemble Natural Language Processing (NLP)-based method designed to automatically align QTOs with corresponding cost indexes across different building classifications and works. Our proposed method leverages advanced NLP techniques to analyze and align textual descriptions in QTOs with corresponding items in cost indexes. The system was rigorously tested on a high-rise residential building project, where it demonstrated an 82.96% agreement rate with QS-based estimations, highlighting its semantic alignment accuracy. The model also achieved only −2.05% total cost deviation from QS estimates. The proposed Artificial Intelligence (AI)-driven system complements QSs by serving as a notification tool, highlighting discrepancies between human and system-generated estimates. This approach aids QSs in refining their cost assessments, enhancing accuracy and mitigating potential financial risks.