Automated tools for FAIRness assessment considering Chemical Digital Objects
Abstract
Enhancing knowledge discovery for human and computational agents is a challenge for data-intensive sciences involving accessing, integrating, and analyzing task-appropriate data. The FAIR principles stand for improving the Findability, Accessibility, Interoperability, and Reusability of digital resources. Applying the FAIR principles produces digital objects that ensure goals like transparency, reproducibility, and reusability, which are essential requirements in the context of using chemicals in scientific discovery.
Guidelines, processes, questionnaires, and semi-automated and automated tools aim to evaluate a digital object's FAIRness level, i.e., produce a percentage grade indicating how close a digital object is to abiding by the FAIR principles.
Manual and semi-automated mechanisms are essential to improve understanding and appreciation of the research life cycle. However, assessing FAIRness with them is time-consuming, requires experience, carries difficulties when inspection is needed, and does not scale when considering several digital objects. An automated tool is more appropriate to handle these issues.
This work analyzes automated tools for FAIRness assessment regarding requirements elicited from the academic literature. When executing their appraisal, we applied the tools to evaluate chemical digital objects, such as compound representations in PubChem and chemical works published in Zenodo.
As examples, all the tools partially fit the requirement: "The tool should be customizable according to the type of digital object and community" since they do not support a user-friendly configuration, requiring software development skills to develop and add new FAIRness evaluation tests in the tool. In this case, the main question is, "How to develop FAIRness assessment tools easily adaptable by non-dev users?"
Considering the requirement "The tool should give a FAIRness score/grad.", some tools present a number without details. In this case, "How do we compute the FAIRness grade for dimensions, principles, metrics, and tests so that the grade considers aspects like priorities defined by the community and is transparent to the users?"
Overall, no tools meet all requirements. Choosing the best tool for FAIRness assessment is challenging; existing appraisals concerning established requirements may help. There is room to solve the gaps by improving existing tools or developing a new tool to overcome them.