Interlinking the Brick Schema with Building Domain Ontologies
Celia Garrido-Hidalgo, Jonathan Fürst, Bin Cheng, Luis Roda-Sánchez, Teresa Olivares, Ernő Kovács · 2022
In the building context, there is a growing requirement for numerous data models and ontologies to coexist as a means to cover different perspectives and communities. While there are some well-known efforts, such as the Brick schema, to create an overall usable ontology, interlinking available ontologies and data models is still a challenge that can provide significant benefits towards interoperability of Building Information Models. To shed light on this matter, we provide a review of some of the most important ontologies in the building context, which we then match against Brick as a means to provide an interlinked data model. For finding matches, we propose TrioNet, an interactive ontology matcher utilizing weak supervision and active learning, which we compare in terms of precision and recall with two well-known state-of-the-art ontology matchers: AgreementMakerLight (AML) and LogMap. TrioNet outperforms them in finding more verified matches with only a few domain expert annotations, making it an ideal tool for the creation of interlinked data models to improve interoperability. With this paper, we contribute the following datasets: (i) the overall Brick data model interlinked to five other ontologies; (ii) the discovered pairwise ontology alignments; and (iii) the manually-annotated matches used for evaluation.