Ontology-based harmonization of disparate datasets for early-phase building life cycle assessment

The building sector faces increasing pressure to reduce life-cycle environmental impacts, particularly during early design stages when key decisions about materials, systems, and construction methods remain open. Although Environmental Product Declarations (EPDs) and life-cycle assessment (LCA) data are available through repositories such as ÖKOBAUDAT, their use in early design remains limited. The analyzed repositories, which are based on the International Life Cycle Data system (ILCD), differ in classification schemes, metadata structures and access formats, making cross-repository search, comparison and integration difficult. This thesis investigates how semantic technologies can improve the interoperability and usability of ILCD-based environmental data for early-stage building design. It develops an ontology-based harmonization workflow that transforms selected repository data into a shared knowledge graph. The workflow combines LinkML-based schema modeling, SKOS-based classification alignment, SHACL-based validation and rule-based enrichment and retrieval-augmented language-model methods for semi-automated categorization. The resulting graph enables more consistent querying across heterogeneous datasets and supports structured filtering by material properties, classifications and environmental indicators. The work is demonstrated through a prototype query interface for early-phase BIM-LCA workflows, using ready-mix concrete as the primary material domain. The prototype illustrates how planning-oriented filters can be exposed without requiring SPARQL knowledge and how fragmented ILCD-based data can be reorganized into a more accessible semantic query space. Overall, the thesis contributes a semantic harmonization approach for making ILCD-based environmental data more transparent, comparable and reusable in early-phase design.