{"ID":2827972,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2512.15281","arxiv_id":"2512.15281","title":"Semantic Grounding of Digital Twin Metamodels Using RDF Graphs","abstract":"Digital Twins (DTs) represent digital counterparts of physical systems, assets, or processes, referred to as the actual twin (AT). DTs integrate heterogeneous data, models, and semantic technologies to support monitoring, simulation, prediction, and optimization, enabling informed decision-making while maintaining a dynamic and accurate reflection of the AT. A key challenge is aligning heterogeneous models, which can cause semantic mismatches, inconsistencies, and synchronization issues. Existing approaches relying on static mappings and manual updates are often inflexible and error-prone. In this study, we address heterogeneity challenge in multi-layered DT, by introducing semantic grounding pipeline for multi-layered DTs that enables consistent and reliable interoperability between abstraction layers. We make three contributions. First, we design and implement multi-layered DT using flexible modelling framework, to organize data, model and metamodel layers. Second, we semantically lift DT metamodel to RDF graph for unified representation. Finally, we present a graph-based alignment approach (SSM-OM), which leverages semantic embeddings, lexical similarity, and large language model (LLM) reasoning to accurately establish and validate correspondences between the lifted metamodel and ontology. We validate correctness, interoperability, cross-layer traceability, domain applicability and general empirical performance through RDF tests, a DT usecase, and ontology alignment evaluation initiative (OAEI) benchmarks, demonstrating semantic consistency in multi-layered DT.","short_abstract":"Digital Twins (DTs) represent digital counterparts of physical systems, assets, or processes, referred to as the actual twin (AT). DTs integrate heterogeneous data, models, and semantic technologies to support monitoring, simulation, prediction, and optimization, enabling informed decision-making while maintaining a dy...","url_abs":"https://arxiv.org/abs/2512.15281","url_pdf":"https://arxiv.org/pdf/2512.15281v2","authors":"[\"Faima Abbasi\",\"Jean-Sébastien Sottet\",\"Cedric Pruski\"]","published":"2025-12-17T10:36:55Z","proceeding":"cs.SE","tasks":"[\"cs.SE\"]","methods":"[\"Large Language Model\",\"Language Model\",\"Generative Adversarial Network\"]","has_code":false}