arXiv 建筑语义标准论文· Daniel Ma·· 1 天前精选AI 评分82
「英中对照」研究称首次将知识图谱补全用于跨建筑 Brick 模型自动生成
Automated Brick Schema Knowledge Graph Completion Across Buildings
AI 导读
一篇 arXiv 论文将知识图谱补全用于自动生成 Brick 建筑模型，作者称这是该方法首次应用于 Brick 模型生成。
推荐理由
Brick 模型依赖人工识读图纸和点位表是楼宇语义建模落地的成本瓶颈，该研究把缺失关系推断形式化为知识图谱补全问题，为减少手工建模提供了可比较的技术路径；但作者自己的评估显示跨数据集和跨关系泛化不稳，工程可用性仍需在更多真实建筑数据上验证。
正文 · 中英逐段对照
View PDF HTML (experimental)查看 PDF HTML（实验性）
Abstract:The Brick schema represents building equipment, sensors, spaces, and the relationships between them as typed, multi-relational graphs, providing machine-readable metadata that supports generalizable control applications. Current methods for producing complete Brick models require extensive manual examination of building schematics and point lists. We observe that as Brick models are knowledge graphs, inferring missing relations is naturally posed as a knowledge graph completion problem. To the best of our knowledge, this is the first application of knowledge graph completion to Brick model generation. We compare three methods, gradient-boosted trees, Pairwise Brick Relation Ranking, and CompGCN, on the relations feeds, hasPart, and hasPoint, which account for the majority of the schematic-reading work in building a Brick model by hand. Models are evaluated on held-out MORTAR buildings and out-of-distribution Kaggle and Building Time Series data. Although CompGCN is the strongest performer on held-out buildings under metrics resistant to class imbalance, leading macro-F1 and AUPRC, no method generalizes uniformly across datasets or relations. 摘要：Brick 模式将建筑设备、传感器、空间以及它们之间的关系表示为类型化的多关系图，提供支持可泛化控制应用的机器可读元数据。目前生成完整 Brick 模型的方法需要大量人工检查建筑图纸和点位清单。我们注意到，由于 Brick 模型本质上是知识图谱，推断缺失关系自然可以表述为知识图谱补全问题。据我们所知，这是知识图谱补全首次应用于 Brick 模型生成。我们在 feeds、hasPart 和 hasPoint 三种关系上比较了三种方法——梯度提升树、Pairwise Brick Relation Ranking 和 CompGCN，这些关系占据了手工构建 Brick 模型时图纸阅读工作的大部分。模型在留出的 MORTAR 建筑以及分布外的 Kaggle 和 Building Time Series 数据上进行评估。尽管 CompGCN 在留出建筑上、在抗类别不平衡的指标（macro-F1 和 AUPRC 领先）下表现最佳，但没有任何方法能在所有数据集或关系上一致地泛化。
| Comments:备注： | 9 pages, 2 figures9 页，2 幅图 |
| Subjects:主题： | Machine Learning (cs.LG)机器学习 (cs.LG) |
| Cite as:引用格式： | arXiv:2610.27339 [cs.LG] |
| (or arXiv:2610.27339v1 [cs.LG] for this version) （或 arXiv:2610.27339v1 [cs.LG] 对应此版本） | |
| https://doi.org/10.48550/arXiv.2610.27339 arXiv-issued DOI via DataCite (pending registration) arXiv 通过 DataCite 签发的 DOI（待注册） |
Submission history提交历史
From: Daniel Ma [view email]
[v1]
Mon, 19 Oct 2026 22:06:34 UTC (1,509 KB)
发件人：Daniel Ma [查看邮箱]
[v1]
Mon, 19 Oct 2026 22:06:34 UTC (1,509 KB)
来源：arXiv 建筑语义标准论文 · arxiv.org