Seminar: “How Fair is FAIR? Understanding LOD Cloud FAIRness Through Correlation Patterns” by Gabriele Tuozzo

Abstract:
While the FAIR principles (Findability, Accessibility, Interoperability, and Reusability) and data quality dimensions are widely used to evaluate Linked Data, their interdependencies remain largely unexplored. This paper is grounded on a systematic integration of these two frameworks by mapping data quality dimensions to FAIR sub-principles, revealing how individual features—such as endpoint availability, metadata richness, or use of standard vocabularies—can simultaneously contribute to multiple FAIR goals. Building on this mapping, this paper reports a large-scale, data-driven, longitudinal study of 1,445 datasets from the LOD Cloud extending KGHeartBeat, an open-source quality assessment framework. This paper quantifies FAIRness at the sub-principle level and computes correlation patterns across five temporal snapshots and nine topical domains. The reported findings reveal that most correlations are positive and statistically significant but vary across time and domain, with only a few stable or persistent relationships. Strong inter-principle correlations (such as those linking metadata standards and security transparency) emerge over time, while intra-principle coherence is often weak. These insights offer concrete guidance for improving FAIR compliance, highlight the importance of domain-aware evaluation, and support the development of more holistic and reproducible FAIR assessment strategies for Linked Data ecosystems.