Enhancing Construction Data Integration through Dynamic Ontology Alignment and Automated Attribute Mapping

Abstract: The construction industry struggles with data integration due to semantic heterogeneity across stakeholders and lifecycle phases. Users must handle complex formats, inconsistent naming conventions, and diverse standards, detracting from core tasks. This work presents a hybrid matching system combining lexical similarity with domain-specific pattern recognition to align terms across terminologies. Automated enrichment from standardized sources and semantic fingerprinting enable property mapping between stakeholders, domains, and formats. A Human-in-the-Loop component ensures continuous refinement via expert feedback. Real-world evaluation in facility management scenarios achieved 86.6% initial accuracy in attribute matching and significant query processing time reductions over manual approaches.