Metadateren kost tijd. Maar het niet, of op een foutieve manier metadateren van gegevens kost nog veel meer tijd. Om de bruikbaarheid van metadata te vergroten zijn de FAIR-principes opgesteld. Deze principes staan voor het Findable, Accessible, Interoperable en Reusable, maken van (wetenschappelijke) gegevens (Directorate-General for Research and Innovation, European Commission 2018).
FAIR data in 2 minutes
In de video van de Universiteit van Maastricht worden de FAIR-principes kort uitgelegd.
Findable (Vindbaar)
Metadata speelt een cruciale rol bij het vinden van data. Het toepassen van gestructureerde en consistente metadata zorgt ervoor dat data gemakkelijk te vinden is, zelfs in grote en complexe datasets.
Het gebruik van uniforme metadatastandaarden en -praktijken maakt het mogelijk dat zoekmachines en gegevenscatalogi effectief en efficiënt door gegevens kunnen navigeren.
Het vindbaarheidsprincipe kan in vier onderdelen opgebroken worden, die hieronder worden genoemd. Klik op de link voor meer informatie.
Metadata biedt informatie over de toegangsrechten en mogelijkheden voor gegevens. Het helpt gebruikers te begrijpen wie toegang heeft tot de gegevens, onder welke voorwaarden en hoe deze kunnen worden benaderd. Het FAIR-principe bevordert transparante en gestandaardiseerde metadata, waardoor gebruikers snel kunnen vaststellen of ze gemachtigd zijn om toegang te krijgen tot de gegevens.
Het toegankelijkheidsprincipe kan in vier onderdelen opgebroken worden, die hieronder worden genoemd. Klik op de link voor meer informatie.
Het vaststellen van consistente metadatastandaarden vergemakkelijkt de interoperabiliteit tussen verschillende systemen en platforms. Dit is essentieel voor het uitwisselen van gegevens tussen verschillende organisaties en disciplines. Standaardisatie van metadata maakt het mogelijk om gegevens op een begrijpelijke en uniforme manier te interpreteren, waardoor ze gemakkelijker kunnen worden geïntegreerd in diverse onderzoeks- en analysetoepassingen.
Het interoperabiliteitsprincipe kan in vier onderdelen opgebroken worden, die hieronder worden genoemd. Klik op de link voor meer informatie.
We krijgen in projecten steeds meer te maken met verschillende soorten data en datastromen. Dan moet er goed gekeken worden naar wat voor standaarden er bestaan, en hoe hieruit te kiezen of combineren. Daarbij is het belangrijk om er ook de verschillende (type) gebruikers uit de verschillende (onderzoeks)disciplines te betrekken.
Reusable (Herbruikbaar)
Metadata speelt een sleutelrol bij het beoordelen van de bruikbaarheid en geschiktheid van gegevens voor hergebruik. Duidelijke metadata geeft informatie over de context, kwaliteit en inhoud van de gegevens. Het FAIR-principe benadrukt het belang van gestructureerde metadata die de herbruikbaarheid van gegevens vergroten door anderen in staat te stellen de gegevens correct te begrijpen en te interpreteren.
Het herbruikbaarheid principe kan in vier onderdelen opgebroken worden, die hieronder worden genoemd. Klik op de link voor meer informatie.
Het gebruik van de FAIR-principes kan de impact van je werk vergroten, omdat meer onderzoekers je gegevens kunnen vinden en hergebruiken. Dit kan resulteren in meer samenwerking met onderzoeksinstellingen en de industrie en erkenning van je gegevens in andere publicaties. Ook onderzoeksgemeenschappen, infrastructuurfaciliteiten voor onderzoek en onderzoeksorganisaties hebben er baat bij (Wilkinson et al. 2016).
3.2 FAIRification proces
Sinds de publicatie van de principes in 2016 zijn er verschillende workflows voorgesteld om het proces om gegevens FAIR te maken te ondersteunen (Bernabé et al. 2023). Figuur 3.1 uit Welter et al. (2023) laat een template zien met daarin 8 belangrijke stappen richting de fairificatie van de data. In de bijlage van dezelfde publicatie (zie hier) worden deze stappen verder uitgewerkt.
Figuur 3.1: De stappen van het FAIRification template. Bij elke stap wordt aangegeven of het hier gaat om de inhoud (data), representatie en formaat (metadata) of hoe de data wordt gehost (infrastructuur). Bron: Welter et al. (2023), beschikbaar onder de CC BY 4.0 Deed Attribution 4.0 International licentie.
Figuur 3.1: De stappen van het FAIRification template. Bij elke stap wordt aangegeven of het hier gaat om de inhoud (data), representatie en formaat (metadata) of hoe de data wordt gehost (infrastructuur). Bron: Welter et al. (2023), beschikbaar onder de CC BY 4.0 Deed Attribution 4.0 International licentie.
Het stappenplan laat zien dat om een goede metadata strategie te ontwikkelen (stap 3) er eerst een aantal andere stappen gezet moeten worden. En er zijn ook een aantal belangrijke vervolgstappen. Let wel op, de stappen worden hier als een eenrichtingsverkeer weergegeven. Maar vooral in projecten waarin verschillende domeinen samenkomen, en waarin verschillende disciplines samenwerken zal dit vaak een iteratief proces zijn, waarin data sharing strategieën en mogelijkheden (stap 7, 8) ook weer van invloed zijn op je metadata strategie (stap 4).
3.3 FAIR zelfevaluatie
Je hebt nu gelezen over FAIR data en de rol die metadata daarin speelt. Hoe staat het eigenlijk met de ‘FAIR-heid’ van de data die jij of jouw organisatie creëert en beheert?
Met de FAIR Data Zelfevaluatietool kun je dit toetsen in hoeverre je data FAIR is en hoe je dit eventueel kunt verbeteren. De tool is ontwikkeld door ARDC en beschikbaar gemaakt onder de GNU GPLv3 licentie. Je kunt de oorspronkelijke code op GitHub vinden.
De tool richt zich op databibliothecarissen en IT-medewerkers, maar kan ook worden gebruikt door software-engineers die FAIR-data-tools en -diensten ontwikkelen, en door onderzoekers en onderzoeksondersteuners.
Er worden vragen gesteld over de principes die ten grondslag liggen aan Vindbaar, Toegankelijk, Interoperabel en Herbruikbaar. Als je alle vragen in elk onderdeel hebt beantwoord, krijg je een ‘groene balk’-indicator op basis van je antwoorden in dat onderdeel, en als alle onderdelen zijn ingevuld, krijg je een algemene ‘FAIRness’-indicator. Elke vraag voorzien van een link naar extra uitleg en informatie over gerelateerde onderwerpen.
Making data Findable includes assigning a
persistent identifier (like a DOI
or Handle ),
having rich metadata
to describe the data and making sure it is findable through
disciplinary and generalist discovery portals (local and international).
About Accessible
To make data accessible may include making the data
open using a standardised protocol.
However the data does not necessarily have to be open. There are sometimes good reasons
why data
cannot be made open, for example privacy concerns, national security or commercial
interests. If it is not open there should be clarity and transparency around the conditions
governing access and
reuse
About Interoperability
To be interoperable (i.e. data that is interpretable by a computer,
so that they can be automatically combined with other data) the data will
need to use community agreed formats,
language and
vocabularies. The metadata
will also need to use a community agreed standards and vocabularies, and
contain links to related information using identifiers.
About Reusable
Reusable data should maintain its initial richness. For example,
it should not be abridged for the purpose of explaining the findings
in one particular publication. It needs a clear machine-readable
licence
and provenance information on how the data was formed. It should also use
discipline-specific data and metadata standards to give it rich contextual
information that will allow for accurate interpretation and reuse.
About identifiers
Identifiers are essential for identifying, finding, retrieving, linking and citing datasets.
A Web address (URL) can be used to specify the online location of a resource but over time URLs
tend to change which leads to broken links to the data. To be useful, identifiers need to be
persistent and unique. Digital Object Identifiers (DOIs) and other persistent identifiers (PIDs)
provide a permanent citable reference to a particular dataset. The DOI is a permanent fixed
reference to the dataset no matter where it is located online and enables citation and citation
metrics.
Services to create a persistent identifier are often offered by your affiliated institution or
the repository you are using to describe your data. Talk to your library service for more information.
The identifier (preferably a persistent identifier) needs to be clearly stated in the metadata
record describing the data collection, and also in any associated data files or metadata.
About metadata descriptions
Comprehensive metadata records will include descriptive content that facilitates discovery, access
and reuse of the data being described. While there is no 'one size fits all' list,
comprehensive metadata should include:
a globally unique persistent identifier e.g. a DOI
a title
related people, i.e. the data creator or custodian
how to access the data and file formats
a description of how the data were created and how to interpret the data
subject or keywords
citation information that clearly indicates how the data should be cited
a machine-readable data licence
provenance and contextual information such as:
links to related publications, projects, services and software
methodology and processes involved in data production
spatial and temporal coverage (if relevant)
object-level data description
Providing metadata in a standard schema allows it to be read and used by machines as well as humans.
A rich metadata description alone does not ensure a dataset’s ‘findability’ on the internet;
the dataset needs to be registered or indexed in a searchable resource, such as a generalist
(e.g. Figshare,
Dryad,
Zenodo
domain-specific (e.g.
PANGAEA,
ADA,
ICPSR), or institutional
(e.g. ANU Research Repository ,
Sydney eScholarship Repository,
data.gov.au) data repository or
registry. Ideally these repositories/registries are indexed by search engines such as Google
and/or Google Scholar.
About data accessibility
Not all data that is discoverable can be freely accessed. Often there are embargoes,
access controls, and access permissions associated with data due to a variety of issues
such as privacy and commercial interests.
Even with all these issues much sensitive data
can be shared. Many other issues that
could be perceived as blockers to sharing data may be
overcome.
About protocols
Ideally users would like to retrieve appropriate internet content directly and unhindered once
they have located it. Internet protocols (e.g. http and ftp) define rules and conventions for
communication between devices, and tools and services are available to facilitate this process,
e.g. APIs.
Protocols
HTTP (Hypertext Transfer Protocol) is the set of rules for transferring files (text, graphic
images, sound, video, and other multimedia files) on the
World Wide Web. As soon as a Web
user opens their Web browser, the user is indirectly making use of HTTP.
FTP (File Transfer Protocol) is a standard Internet
protocol for transmitting
files between computers on the Internet over TCP/IP connections.
Read more:
Web protocols
API (Application Programming Interface): When information is made available in any machine
readable format, it becomes possible to make that information directly available to programs
that request that information over the web. An API is the way this information is made
directly available to other machines.
Effort is often required to maintain data resources online which can often lead to it
being neglected especially over long periods of time. This often leads to broken links
between the metadata and the data. Having at least a description of the data in the form
of a metadata record means there is a record of the data's existence allowing the possibility
of someone tracking it down.
About file formats
When selecting file formats for archiving, the formats should ideally be:
Non-proprietary
Unencrypted
Uncompressed
In common usage by the research community
Adherent to an open, documented standard, such as described by the State of California (see
AB 1668, 2007)
Interoperable among diverse platforms and applications
Fully published and available royalty-free
Fully and independently implementable by multiple software providers on multiple platforms
without any intellectual property restrictions for necessary technology
Developed and maintained by an open standards organization with a well-defined
inclusive process for evolution of the standard.
Schema standards are schemas that have gone through a formal validation process by a standards
organisation, such as the International Standards Organisation (ISO) or an equivalent body such
as the Dublin Core Metadata Initiative (DCMI); or, commonly used and consistently applied metadata
schemas that are well documented, endorsed, and maintained can also become ‘de-facto standards’,
e.g. RIF-CS for describing data collections and services used in Research Data Australia.
Resolvable global identifiers - An identifier is any label used to name something uniquely
(whether online or offline). URLs are an example of an identifier. So are serial numbers, and
personal names. A persistent identifier is guaranteed to be managed and kept up to date over a
defined time period.
The goal is to create as many meaningful linkages as possible between (meta)data resources to
enrich the contextual knowledge about the data, balanced against the time/energy involved in
making a good data model.
Machine-readable (meta)data
(Meta)data in a format that can be automatically read and processed by a computer, such as
CSV,
JSON,
XML etc.
For more information:
Open Data Handbook
Linked Data
Linked Data (also known as Linking Data) can be applied to improve the exploitation of the
“Web of data.” The expression refers to the publishing of structured data in a way that typed
links are created between data from different sources to provide a higher level of usability.
By using Linked Data, it is possible to find other, related data. Structured data should meet
four requirements to be called Linked Data:
URIs should be assigned to all entities of the dataset.
HTTP URIs are required to ensure that all entities can be referenced and cited by users and
user agents.
Entities should be described using standard formats such as RDF/XML.
Links should be created to other, related entity URIs.
All data that fulfil these requirements and are released for the public are called
Linked Open Data (LOD).
http://www.lesliesikos.com/semantic-web-machine-readable-structured-data-with-meaningful-annotations/
Read more under the heading (meta)data include qualified references to other (meta)data in the following resource:
https://www.dtls.nl/fair-data/fair-principles-explained/
For more information on Linked Data standards -
RDF - https://www.w3.org/RDF/
Other related linked data standards:
OWL - https://www.w3.org/OWL/ , SKOS - https://www.w3.org/2004/02/skos/ , SPARQL - https://www.w3.org/TR/rdf-sparql-query/
About licenses
If data is not licensed no-one else can use it. In Australia, no licence is regarded as the same
as 'all rights reserved', confining any reuse to very limited circumstances. Applying a Creative
Commons licence to your data is a simple way to ensure that your data can be reused. The less
restrictive the licence, the more that can be done with the data.
To make it easy for the Web to know when a work is available under a particular license, a
“machine readable” version of the license provides a summary of the key freedoms and obligations
written into a format that software systems, search engines, and other kinds of technology can
understand.
<a rel="license" href="http://creativecommons.org/licenses/by/4.0/">
<img alt="Creative Commons License" style="border-width:0"
src="https://i.creativecommons.org/l/by/4.0/88x31.png"/>
</a><br />This work is licensed under a
<a rel="license" href="http://creativecommons.org/licenses/by/4.0/">
Creative Commons Attribution 4.0 International License</a>.
Data provenance is used to document where a piece of data comes from and the
process and methodology by which it is produced. It is important to confirm
the authenticity of data enabling trust, credibility and reproducibility.
This is becoming increasingly important, especially in the eScience community
where research is data intensive and often involves complex data transformations
and procedures.
Read more…
Provenance vocabularies such as
Provenir Ontology (PROV-O) and others
About Findable Score
The score was derived as follows:
Does the dataset have any identifiers assigned:
You answered which gave you out of a
possible 8 points.
Is the dataset identifier included in all mdetadata records:
You answered which gave you out of a
possible 1 point.
How is the data described with metadata:
You answered which gave you out of a
possible 4 points.
What type of repository or registry is the metadata record in:
You answered which gave you out of a
possible 4 points.
This gives you out of a possible 17 points.
About Findable Score
The score was derived as follows:
How accessible is the data:
You answered which gave you out of a
possible 5 points.
Is the data available online without requiring specialised protocols or tools once access has
been approved:
You answered which gave you out of a
possible 4 point.
Will the metadata record be available even if the data is no longer available:
You answered which gave you out of a
possible 1 points.
This gives you out of a possible 10 points.
About Findable Score
The score was derived as follows:
What (file) format(s) is the data available in:
You answered which gave you out of a
possible 2 points.
What best describes the types of vocabularies/ontologies/tagging schemas used to define
the data elements:
You answered which gave you out of a
possible 3 point.
How is the metadata linked to other data and metadata to provide context around the data:
You answered which gave you out of a
possible 3 points.
This gives you out of a possible 8 points.
About Findable Score
The score was derived as follows:
Does the dataset have any identifiers assigned?
You answered which gave you out of a
possible 4 points.
Is the dataset identifier included in all mdetadata records:
You answered which gave you out of a
possible 2 points.
This gives you out of a possible 6 points.
Let op, de tool is bedoeld voor educatieve en informatieve doeleinden. De scores die deze tool oplevert, zijn alleen bedoeld om het denken en de discussie op gang te brengen over mogelijke manieren om data gegevens in je project of organisatie meer FAIR te maken. Let op, de scores zijn gebaseerd op de interpretatie van ARDC van de FAIR Data Principes. Er bestaan ook andere interpretaties van de principes.
Bernabé, César, Tiago Prince Sales, Erik Schultes, Niek Van Ulzen, Annika Jacobsen, Luiz Olavo Bonino Da Silva Santos, Barend Mons, and Marco Roos. 2023. “A Goal-Oriented Method for FAIRification Planning.” Preprint. In Review. https://doi.org/10.21203/rs.3.rs-3092538/v1.
Data together. n.d. “GO FAIR Initiative: Make Your Data & Services FAIR.”GO FAIR. Accessed January 15, 2024. https://www.go-fair.org/.
Directorate-General for Research and Innovation, European Commission. 2018. Turning FAIR into Reality: Final Report and Action Plan from the European Commission Expert Group on FAIR Data. LU: Publications Office of the European Union. https://data.europa.eu/doi/10.2777/1524.
Welter, Danielle, Navtej Juty, Philippe Rocca-Serra, Fuqi Xu, David Henderson, Wei Gu, Jolanda Strubel, et al. 2023. “FAIR in Action - a Flexible Framework to Guide FAIRification.”Scientific Data 10 (May). https://doi.org/10.1038/s41597-023-02167-2.
Wilkinson, Mark D., Michel Dumontier, IJsbrand Jan Aalbersberg, Gabrielle Appleton, Myles Axton, Arie Baak, Niklas Blomberg, et al. 2016. “The FAIR Guiding Principles for Scientific Data Management and Stewardship.”Scientific Data 3 (1): 160018. https://doi.org/10.1038/sdata.2016.18.
Figuur 3.1: De stappen van het FAIRification template. Bij elke stap wordt aangegeven of het hier gaat om de inhoud (data), representatie en formaat (metadata) of hoe de data wordt gehost (infrastructuur). Bron: Welter et al. (2023), beschikbaar onder de CC BY 4.0 Deed Attribution 4.0 International licentie.
