Findable. Accessible. Interoperable. Reusable.
Wilkinson et al., https://www.nature.com/articles/sdata201618
Since their introduction in 2016, the FAIR data principles have defined the standards by which scientific researchers share data. However, modern research in data editing and management consistently shows that while the FAIR data principles are widely accepted in theory, they can be much more difficult to understand and implement in practice. In this Knowledge Discovery and Data Mining 2024 tutorial, we explore some of the simple, realistic steps scientists can take to FAIRly release open data.
We also explore areas where the current FAIR guidelines fall short and offer practical suggestions for making open data FAIR(ER): more Equitable and Realistic. This first involves ways to make datasets themselves more equitably accessible for researchers with disabilities. While equitably accessible data design has some research overlap with paper, presentation, and website design, we suggest several unique distinctions specific to datasets. The “Realistic” aspect of FAIR(ER) data facilitates a path to translate open data (and research on that data) back to true applications. Driven by national security applications pipelines, we call out important considerations for balancing data editing against data realism.