Finding and Reusing Metrics and Benchmarks

This tutorial explains how to discover FAIR metrics and benchmarks using two complementary routes: the FAIRassist registry and FAIRsharing’s own search and browse features. Once you have found a metric of interest, you can learn how to locate any tests that implement it in find-test-for-digital-object.rst.

What are metrics and benchmarks?

FAIRsharing registers the conceptual components of the FAIR assessment ecosystem within its FAIRassist registry. This includes:

  • Principles — are designed to be subject- and implementation-agnostic criteria or high-level goals that may be refined for particular communities using the Assess IF (and are stored within FAIRsharing’s Standards registry, e.g. the FAIR Principles).

  • Metrics — measurable criteria that interpret a Principle, specifying what must be true of a digital object for that Principle to be satisfied.

  • Benchmarks — curated sets of Metrics that together define what “FAIR” means in practice for a particular community, tool, or use case.

For more background on the FAIRassist registry and how it aligns with the FAIR Testing Resource (FTR) vocabulary, see the FAIRsharing documentation on registry types.

A note on object types

Both sections of this tutorial describe a number of ways to filter metrics and benchmarks, including by the type of digital object they are designed to assess. Before using this filter, it is worth reviewing the full controlled vocabulary of object types in FAIRsharing, available at https://fairsharing.gitbook.io/fairsharing/record-sections-and-fields/general-information/object-types.

The controlled vocabulary includes specific types such as dataset, software application, model, terminology artifact, protocol or workflow, and several others. Two general-purpose types are also available:

  • object type agnostic — use this when a metric or benchmark is relevant across all types of digital object.

  • other object type — use this only when the object type is not covered by any of the other terms in the vocabulary and the resource is not agnostic.

Note

The term object type not found appears on a small number of older deprecated records where the original object type could not be determined during curation. This is an administrative term and is not relevant when searching for metrics or benchmarks; you can safely ignore it.

Take time to review the full vocabulary before selecting a type to filter by, as the most appropriate choice may not always be immediately obvious. For example, a metric that is relevant to all digital objects should be found using object type agnostic rather than by listing individual types.

Section 1: Metric discovery via FAIRassist

The FAIRassist registry is the primary entry point for discovering metrics and benchmarks. Although FAIRassist is part of FAIRsharing, its interface is purpose-built for exploring these record types, and presents search results in a tabular format that gives you an overview of the full ecosystem of resources related to a given set of metrics or benchmarks.

FAIRassist is particularly useful for answering questions such as:

  • What assessments already exist for my subject area or for a particular tool?

  • Which standards and databases are referenced by assessments in my domain?

  • What benchmarks has a given organisation defined or contributed to?

  • Which metrics apply to a particular type of digital object?

This search and display method provides dynamic guidance that supplies extra context around FAIR benchmarks, and transparent definitions of metrics to support their correct reuse.

Using the filters

The FAIRassist registry provides five filters that can be used individually or in combination.

Record type

Select whether you want to retrieve Metrics or Benchmarks (or both). This is the broadest filter and must be used to ensure the appropriate usage of later filters.

Object type

Metrics can be filtered by the type of digital object they are designed to assess. Before using this filter, review the full controlled vocabulary of object types described in the note on object types above and in the FAIRsharing object types documentation, to ensure you select the term that most accurately reflects your needs. This may include object type agnostic for metrics that apply across all digital object types, or other object type for those covering object types not represented elsewhere in the vocabulary.

Tool

Metrics can be filtered by the assessment tool that utilises them. Currently available tool options include FOOPS! and FAIR Champion. This filter is useful when you are working with a specific tool and want to understand which metrics that tool draws on, or when you want to find metrics that are already implemented in a tool you trust.

Subject

Both metrics and benchmarks can be filtered by subject area. The subjects available reflect the FAIRsharing subject hierarchy, which you can browse at https://fairsharing.org/browse/subject. Use this filter to narrow your results to metrics and benchmarks that are relevant to your research domain.

Organisation

Both metrics and benchmarks can be filtered by the organisation associated with them. This is useful when you want to find all metrics or benchmarks developed or maintained by a particular institution, project, or working group.

A worked example: finding metrics linked to CESSDA

To illustrate how these filters work together, the following URL retrieves all metrics associated with the CESSDA organisation and linked to the FAIR Principles:

https://fairassist.org/registry?search=(principle=The+FAIR+Principles%26organisations=cessda%26recordType=metric_ids

The results are displayed in tabular sections, one per record type, allowing you to quickly scan the set of metrics that CESSDA has contributed to or is associated with, alongside contextual information about each one. From any row in the results you can navigate directly to the full FAIRsharing record for that metric.

Section 2: Metric discovery via FAIRsharing

FAIRsharing’s own search and browse tools can also be used to find metrics and benchmarks. This route is particularly useful when you want to combine FAIRassist record types with other FAIRsharing filters, such as status or object type, or when you want to use the advanced search to build a precise query. Full documentation on FAIRsharing’s search features is available at https://fairsharing.gitbook.io/fairsharing/how-to/searching-and-browsing.

Browse by subject

If you prefer to explore visually, the FAIRsharing subject browser at https://fairsharing.org/browse/subject allows you to navigate a hierarchical sunburst diagram of subject areas. Selecting a subject will show you all FAIRsharing records tagged with that subject or any of its child terms, across all registries including FAIRassist.

Next steps

Once you have identified a metric of interest, you may want to find any tests that implement it. See find-test-for-digital-object.rst for a step-by-step guide.