Automated I-ADOPT service¶
The I-ADOPT Variable Service transforms natural-language descriptions of scientific variables into machine-interpretable representations aligned with the I-ADOPT Framework Ontology.
Role in the FAIRification Framework¶
Scientific variables are frequently described only with free-text labels. This makes them difficult to compare, integrate and process automatically across datasets and infrastructures. I-ADOPT addresses this problem by decomposing a variable into semantic components such as the property, object of interest, matrix, contextual objects, statistical modifiers and constraints.
The service combines LLM-assisted decomposition with a human-in-the-loop workflow. A user supplies a plain-language variable definition, reviews the proposed decomposition and can correct it before producing an I-ADOPT-conformant representation.
Response to user requirements¶
The FAIR2Adapt requirements process explicitly identifies variables from case-study digital objects so that they can later be enriched and aligned with I-ADOPT. This affects the service in several ways.
Lower the semantic-modelling barrier¶
CCA researchers and data providers need machine-actionable variable metadata, but they cannot be expected to be experts in semantic-web modelling. The service therefore starts from a natural-language definition and proposes the decomposition automatically.
Keep domain experts in control¶
Automatic decomposition cannot replace domain knowledge. The proposed representation is presented to the user for review and correction. This human-in-the-loop approach responds to the need for automation while preserving semantic quality and scientific oversight.
Produce reusable machine-actionable variables¶
The result is not only a textual description. The service generates standards-compliant RDF/Turtle aligned with I-ADOPT, can link components to external semantic concepts and supports persistent publication of variable descriptions.
Evaluation in the CCA case studies¶
The service is directly relevant to CS2 RiOMar, where oceanographic variables associated with the coastal-water-quality model outputs need interoperable descriptions. I-ADOPT variable descriptions form part of the RiOMar FAIR Solution together with the data-processing/publishing pipeline, RO-Crate profiles and FAIR publication mechanisms.
The requirements analysis also envisages I-ADOPT enrichment across the case studies. This makes the service a semantic interoperability layer that can be reused wherever variables extracted from datasets or publications need to be represented in a machine-actionable form.
The service is designed for researchers, data stewards and infrastructure developers who need FAIR-compliant variable metadata without requiring deep semantic-web expertise.
CCA evaluation
See Evaluation in the CCA case studies for the online summary of how the I-ADOPT service and the other FAIR supporting services respond to the case-study requirements.