UK Data Service

FAIR-IMPACT aims to realise a FAIR EOSC using proven solutions, tools and methods developed during the FAIRsFAIR and other initiatives. One of the goals of the project is to enable the ‘FAIRification’ of different research objects such as datasets, software and semantic artefacts originating from a large range of scientific disciplines. This includes the provision and extension of FAIR assessment metrics and associated tools and their adoption to the needs and requirements of a variety of research communities. In particular FAIRsFAIR data object assessment metrics as well as the F-UJI tool are intended to become more disciplinary-context aware and to include more discipline-specific tests in cooperation with FAIR-IMPACT use case partners, domain data repositories, research infrastructures and e-infrastructures. This deliverable provides the first set of discipline specific tests and metrics developed in cooperation with FAIR-IMPACT Social Sciences and Humanities (SSH) use case partners. We present an analysis of SSH community FAIR-aligned habits and practices carried out using available literature and whitepapers, data collected using standard interfaces provided by the community, as well as FAIR Implementation Profiles (FIPs) from a number of SSH data repositories. Based on this analysis we identified an appropriate SSH sub-community, the social sciences, for which we defined a set of discipline specific metrics and tests derived from the FAIRsFAIR data assessment metrics which are also presented in this deliverable.

This iterative working paper presents a set of high-level interactive components necessary for a functioning FAIR ecosystem. Operational repositories, and other (meta)data related services are supported by common reference information managed in registries, and by evaluation processes consisting of standards and assessment methods. <br> Findable, Accessible, Interoperable and Reusable digital objects (data and metadata) are a necessary input to, and output from, ecosystems that integrate (one provider serves many) or federate (many providers interoperate) services. <br> Users of all kinds must be empowered to find and access (meta)data that supports interoperability and reuse. All ecosystem actors depend on machine and human-mediated evaluation of digital objects’ FAIRness.

Machine learning (ML) is increasingly useful as data grow in volume and accessibility. ML can perform tasks (e.g., categorisation, decision making, anomaly detection, etc.) through experience and without explicit instruction, even when the data are too vast, complex, highly variable, full of errors to be analysed in other ways. Thus, ML is great for natural language, images, or other complex and messy data available in large and growing volumes. Selecting ML models for tasks depends on many factors as they vary in supervision needed, tolerable error levels, and ability to account for order or temporal context, among many other things. Importantly, ML methods for tasks that use explicitly ordered or time-dependent data struggle with errors or data asymmetry. Most data are (implicitly) ordered or time-dependent, potentially allowing a hidden ‘arrow of time’ to affect ML performance on non-temporal tasks. This research explores the interaction of ML and implicit order using two ML models to automatically classify (a non-temporal task) tweets (temporal data) under conditions that balance volume and complexity of data. Results show that performance was affected, suggesting that researchers should carefully consider time when matching appropriate ML models to tasks, even when time is only implicitly included.

Machine learning (ML) is increasingly useful as data grows in volume and accessibility as it can perform tasks (e.g. categorisation, decision making, anomaly detection, etc.) through experience and without explicit instruction, even when the data are too vast, complex, highly variable, full of errors to be analysed in other ways , . Thus, ML is great for natural language, images, or other complex and messy data available in large and growing volumes. Selecting a ML algorithm depends on many factors as algorithms vary in supervision needed, tolerable error levels, and ability to account for order or temporal context, among many other things. Importantly, ML methods for explicitly ordered or time-dependent data struggle with errors or data asymmetry. Most data are at least implicitly ordered, potentially allowing a hidden `arrow of time&amp;rsquo; to affect non-temporal ML performance. This research explores the interaction of ML and implicit order by training two ML algorithms on Twitter data before performing automatic classification tasks under conditions that balance volume and complexity of data. Results show that performance was affected, suggesting that researchers should carefully consider time when selecting appropriate ML algorithms, even when time is only implicitly included.

This draft working paper has been prepared by members of the SSHOC, EOSC Nordic and FAIRsFAIR projects working on related topics. It seeks to present some key concepts and expand on them to specify the standards and assessments required for an interoperable ecosystem of FAIR (findable, accessible, interoperable and reusable) data preserved for the long term in generalist and specialist FAIR-enabling trustworthy digital repositories (TDR) for a defined designated community of users. It seeks to provide context and define these concepts for audiences familiar with research data and technical data management systems, but with less direct experience of digital preservation and trustworthy digital repositories. This is intended to help clarify which organisations are potential candidates to receive CoreTrustSeal TDR status and to identify and support the types of organisations that may not be candidates but also play a vital role in the data ecosystem. A public comment version is available in Google Document Format.

The European Language Social Science Thesaurus (ELSST) is a broad-based, multilingual thesaurus for the social sciences. It is owned and published by the Consortium of European Social Science Data Archives (CESSDA). ELSST will be used for data discovery within CESSDA and will thus facilitate access to data resources across Europe, independent of domain, resource, language or vocabulary. The thesaurus covers the core social science disciplines: politics, sociology, economics, education, law, crime, demography, health, employment, information and communication technology and, increasingly, environmental science. This guide covers the first release of the thesaurus to the CESSDA ELSST platform in November 2020.

Accompanying the growing importance of research data management, the provision and maintenance of metadata – understood as data about (research) data – have obtained a key role in contextualizing, understanding, and preserving research data. Acknowledging the importance of metadata in the social sciences, the Consortium of European Social Science Data Archives started the Metadata Office project in 2019. This project report presents the various activities of the Metadata Office (MDO). Metadata models, schema, controlled vocabularies and thesauri are covered, including the MDO’s collaboration with the DDI Alliance on multilingual translations of DDI vocabularies for CESSDA Service Providers. The report also summarizes the communication, training and advice provided by MDO, including DDI use across CESSDA, illustrates the impact of the project for the social science and research data management community, and offers an outline regarding future plans of the project.

This document represents the third alignment of CoreTrustSeal to FAIR requirements to inform repositories seeking to enable FAIR data. Further context is presented in the FAIRsFAIR project milestone: https://doi.org/10.5281/zenodo.4003598.

Aligning the CoreTrustSeal Requirements with an assessment of repositories' ability to enable FAIR data is an important part of delivering an EOSC. Trustworthy Digital Repositories (TDR) which enable FAIR data are a dependency for many components of modern, open, distributed research. This paper sets the work within the wider context of data infrastructures, describes the co-dependencies between (meta) data objects and their repository environment, and presents the developing mapping between requirements and principles. The evolving capability/maturity approach is explained and the design of a governed assessment and certification process is defined. This work will iterate alongside the wide range of ongoing data infrastructure initiatives to support a range of stakeholders on their journey towards trustworthy repository services that enable FAIR data. Extensive engagement and feedback are planned to allow us to reach this goal.

&lt;p&gt;This paper is milestone 4.1 of the FAIRsFAIR task 4.1 (Capability Maturity models towards FAIR Certification) within the FAIR Certification work package (WP4).&lt;/p&gt; &lt;p&gt;This document presents the first iterative step in aligning the characteristics of FAIR digital objects with the repositories that &lsquo;enable&rsquo; FAIRness, through the CoreTrustSeal Trustworthy Data Repository Requirements&nbsp;and the application of a capability/maturity evaluation approach.&lt;/p&gt; &lt;p&gt;&nbsp;&lt;/p&gt;