Bridging Building Information Systems: A Parameter-Efficient Semantic Approach to Construction Data Interoperability

This paper investigates the application of semantic text similarity embedding models to address data interoperability challenges in the construction and civil engineering sectors.We focus on improving embedding models to produce high-quality domain-specific embeddings that precisely capture complex terminology and relationships among various building systems, including Building Information Modeling, Building Energy Modeling, and Building Management Systems.Our methodology includes a comprehensive data collection and preprocessing pipeline, followed by a multi-stage fine-tuning approach that employs continuous pre-training, task-specific fine-tuning, and parameter-efficient adaptation algorithms.We assess three parameter-efficient fine-tuning techniques: Low-Rank Adaptation, Rank-Stabilized Low-Rank Adaptation, and Weight-Decomposed Low-Rank Adaptation.Our evaluation demonstrates that domain-adapted models, especially, substantially surpass baseline language models in domain-specific tasks, including understanding technical terms and data alignment across domains.The results show significant performance enhancements with minimal resource investment, making these approaches practical for real-world applications in the building industry.