Towards Findable, Accessible, Interoperable and Reusable (FAIR) Data Repositories: Improving a Data Repository to Behave as a FAIR Data Point

Abstract

ABSTRACT Significant effort is required to find, make sense and reuse research data. To tackle this problem, the Findable, Accessible, Reusable and Interoperable (FAIR) data principles describe a minimal set of requirements for data management and stewardship, considered as the technological basis for the European Open Science Cloud. The FAIR data point (FDP) leverages linked data (LD) to expose data and metadata adhering to the FAIR data principles, specifying a set of standardized metadata that a data repository should implement. Data owners can expose datasets, and data users can reuse datasets through RESTful services, enabling interoperability in a web scale. Data repositories and their underlying software only recently started supporting LD, and their metadata are only available as key-value pairs. An open question in this context is how to enable an existing data repository software to be compliant with the FDP specification, i.e., how to add semantic descriptions to data repositories to ensure the semantic interoperability among data from different repositories? This paper describes a semantic proxy solution to enable a data repository software, the EUDAT B2share service to behave as an FDP in a non-invasive and non-intrusive way, enabling the semantic interoperability through semantic translations. Our solution describes a methodology for metadata mapping based on endogenous model-driven transformations from lexicon to semantic models. We show how metadata in key-value pairs from a general-purpose repository can be made compliant with LD technology without changing the repository software. The solution validation includes functional tests of the FDP metadata layers and a performance analysis of the impact of the semantic proxy on data exchange. The results show that B2share can be compliant to FDP specifications with a reduced impact on the data exchange performance. Therefore, the validation shows that the solution is feasible and adequate to transform a general-purpose data repository software in an FDP.

Keywords: FAIR Data; Data Reusability; Data Repository Software; FAIR Data Point.