🤖 AI Summary
This study addresses the limitation of traditional encoding schemes—such as one-hot encoding—in capturing fine-grained semantic relationships among architectural component subtypes, which hinders semantic understanding in artificial intelligence applications within the AECO (Architecture, Engineering, Construction, and Operations) domain. To overcome this, the authors propose a novel semantic encoding approach that integrates large language model (LLM) embeddings with Matryoshka-compressed representations. Specifically, they combine LLM-generated semantic vectors—derived from GPT and LLaMA—with GraphSAGE, a graph neural network, for semantic classification across 42 architectural component categories. Experimental results demonstrate that the proposed method substantially outperforms the one-hot baseline, with the compressed LLaMA-3 embeddings achieving a weighted F1 score of 0.8766—an improvement of approximately 2.9%—while effectively preserving fine-grained semantic relationships and enabling efficient modeling.
📝 Abstract
Accurate representation of building semantics, encompassing both generic object types and specific subtypes, is essential for effective AI model training in the architecture, engineering, construction, and operation (AECO) industry. Conventional encoding methods (e.g., one-hot) often fail to convey the nuanced relationships among closely related subtypes, limiting AI's semantic comprehension. To address this limitation, this study proposes a novel training approach that employs large language model (LLM) embeddings (e.g., OpenAI GPT and Meta LLaMA) as encodings to preserve finer distinctions in building semantics. We evaluated the proposed method by training GraphSAGE models to classify 42 building object subtypes across five high-rise residential building information models (BIMs). Various embedding dimensions were tested, including original high-dimensional LLM embeddings (1,536, 3,072, or 4,096) and 1,024-dimensional compacted embeddings generated via the Matryoshka representation model. Experimental results demonstrated that LLM encodings outperformed the conventional one-hot baseline, with the llama-3 (compacted) embedding achieving a weighted average F1-score of 0.8766, compared to 0.8475 for one-hot encoding. The results underscore the promise of leveraging LLM-based encodings to enhance AI's ability to interpret complex, domain-specific building semantics. As the capabilities of LLMs and dimensionality reduction techniques continue to evolve, this approach holds considerable potential for broad application in semantic elaboration tasks throughout the AECO industry.