From Standards to Application Matrices: Building FAIRagro Data Quality & Fitness-for-Use

Mahdi Hedayat Mahmoudi, Markus Möller Julius Kühn Institute, Digitalisation and Artificial Intelligence, Kleinmachnow, Germany From Standards to Application Matrices: Building FAIRagro Data Quality & Fitness-for-Use Impact •Extracted ISO based descriptors. •Generated machine-readable outputs (JSON, RDF). Benefits •Standard compliance. •Scalable, repeatable workflows. Data quality is key for reusability → must consider both producer and user perspectives. •Producers (Standards): ISO 19157-1 defines the producer view. •Gaps: PDFs not machine-actionable. •Users (Applications): No standards for user perspective; quality elements scattered across literature. •Opportunities: Scientific literature reveals user needs and data applications. •Key Question: How can we transform standards + literature into structured, queryable, FAIR-compliant descriptors? •Source ISO 19157 PDF. •Use LLMs to extract structured JSON (text, tables, flowcharts). •Extract machine-readable quality descriptors from ISO standards. •Preserve ISO compliance. •LLM-assisted extraction enables structured knowledge from ISO PDFs. •JSON representation allows systematic literature-based expansion. •Workflow preserves ISO compliance while enhancing domain-specific relevance. •See Breakout Sessions 3 (A) and https://doi.org/10.5281/zenodo.17198234 •. Objective •Extend with fitness-for-purpose elements from literature. •Link datasets ↔ applications via application–data matrices. Methodology •Expand JSON with literature-derived metrics. •Derive application matrices. •Produce FAIR-compliant machine-actionable outputs (JSON, RDF). The Challenge Impact & Benefits Users Producers Integration Conclusion Impact •Literature-derived descriptors added. •Developed demo app to showcase LLM-based extraction. Benefits •Extends domain relevance. •Enables automatic ontology/relationship extraction. Producers Producers Users Users