Norbert Kockmann | Search arXiv

arXiv2026

ChemDCAT-AP: Enabling Semantic Interoperability with a Contextual Extension of DCAT-AP

Cross-domain data integration drives interdisciplinary data reuse and knowledge transfer across domains. However, each discipline maintains its own metadata schemas and domain ontologies, employing distinct conceptual models and application profiles, which complicates semantic interoperability. The W3C Data Catalog Vocabulary (DCAT) offers a widely adopted RDF vocabulary for describing datasets and their distributions, but its core model is intentionally lightweight. Numerous domain-specific application profiles have emerged to enrich DCAT's expressivity, the most well-known DCAT-AP for public data. To facilitate cross-domain interoperability for research data, we propose DCAT-AP PLUS, a DCAT Application Profile (P)roviding additional (L)inks to (U)se-case (S)pecific context (DCAT-AP+). This generic application profile enables a comprehensive representation of the provenance and context of research data generation. DACT-AP+ introduces an upper-level layer that can be specialized by individual domains without sacrificing compatibility. We demonstrate the application of DCAT-AP+ and a specific profile ChemDCAT-AP to showcase the potential of data integration of the neighboring disciplines chemistry and catalysis. We adopt LinkML, a YAML-based modeling framework, to support schema inheritance, generate domain-specific subschemas, and provide mechanisms for data type harmonization, validation, and format conversion, ensuring smooth integration of DCAT-AP+ and ChemDCAT-AP within existing data infrastructures.

cs.DB↗

arXiv2023

Using Artificial Neural Networks to Determine Ontologies Most Relevant to Scientific Texts

This paper provides an insight into the possibility of how to find ontologies most relevant to scientific texts using artificial neural networks. The basic idea of the presented approach is to select a representative paragraph from a source text file, embed it to a vector space by a pre-trained fine-tuned transformer, and classify the embedded vector according to its relevance to a target ontology. We have considered different classifiers to categorize the output from the transformer, in particular random forest, support vector machine, multilayer perceptron, k-nearest neighbors, and Gaussian process classifiers. Their suitability has been evaluated in a use case with ontologies and scientific texts concerning catalysis research. From results we can say the worst results have random forest. The best results in this task brought support vector machine classifier.

cs.NE↗