FAIR Explained

FAIR EXPLAINED
Webinar FAIR Italy – February 27 2019
Luiz Bonino

FAIR PRINCIPLES: MANY WAYS TO LOOK AT THEM

FAIR PRINCIPLES
Findable:
F1. (meta)data are assigned a globally unique and persistent
identifier;
F2. data are described with rich metadata;
F3. metadata clearly and explicitly include the identifier of the
data it describes;
F4. (meta)data are registered or indexed in a searchable
resource;
Accessible:
A1. (meta)data are retrievable by their identifier using a
standardized communications protocol;
A1.1 the protocol is open, free, and universally
implementable;
A1.2. the protocol allows for an authentication and
authorization procedure, where necessary;
A2. metadata are accessible, even when the data are no longer
available;
Interoperable:
I1. (meta)data use a formal, accessible, shared, and broadly
applicable language for knowledge representation.
I2. (meta)data use vocabularies that follow FAIR principles;
I3. (meta)data include qualified references to other
(meta)data;
Reusable:
R1. (meta)data are richly described with a plurality of accurate and
relevant attributes;
R1.1. (meta)data are released with a clear and accessible data
usage license;
R1.2. (meta)data are associated with detailed provenance;
R1.3. (meta)data meet domain-relevant community
standards;
https://www.nature.com/articles/sdata201618

FAIR DATA PRINCIPLES - METADATA
Findable:
F1. metadata are assigned a globally unique and persistent
identifier;
data it describes;
F4. metadata are registered or indexed in a searchable
resource;
Accessible:
A1. metadata are retrievable by their identifier using a
implementable;
available;
Interoperable:
I1. metadata use a formal, accessible, shared, and broadly
I2. metadata use vocabularies that follow FAIR principles;
I3. metadata include qualified references to other metadata;
Reusable:
R1. metadata are richly described with a plurality of accurate and
R1.1. metadata are released with a clear and accessible data
usage license;
R1.2. metadata are associated with detailed provenance;
R1.3. metadata meet domain-relevant community standards;

FAIR DATA PRINCIPLES – DATA/DIGITAL RESOURCES
Findable:
F1. data are assigned a globally unique and persistent
identifier;
data it describes;
F4. metadata are registered or indexed in a searchable
resource;
Accessible:
A1. metadata are retrievable by their identifier using a
implementable;
available;
Interoperable:
I1. metadata use a formal, accessible, shared, and broadly
I2. metadata use vocabularies that follow FAIR principles;
I3. metadata include qualified references to other (meta)data;
Reusable:
R1. metadata are richly described with a plurality of accurate and
R1.1. metadata are released with a clear and accessible data
usage license;
R1.2. metadata are associated with detailed provenance;
R1.3. metadata meet domain-relevant community standards;

FAIR DATA PRINCIPLES – SUPPORTING ELEMENTS
Findable:
F1. (meta)data are assigned a globally unique and
persistent identifier;
F3. metadata clearly and explicitly include the identifier
of the data it describes;
resource;
Accessible:
A1.1. the protocol is open, free, and universally
implementable;
A2. metadata are accessible, even when the data are no
longer available;
Interoperable:
I1. (meta)data use a formal, accessible, shared, and
broadly applicable language for knowledge
representation;
I2. (meta)data use vocabularies that follow FAIR
principles;
(meta)data;
Reusable:
R1. (meta)data are richly described with a plurality of
accurate and relevant attributes;
R1.1. (meta)data are released with a clear and
accessible data usage license;
R1.2. (meta)data are associated with detailed
provenance;
standards;

REPOSITORIES SUPPORTING USERS TO ACHIEVE FAIR
Findable:
F1. (meta)data are assigned a globally unique and persistent
identifier;
data it describes;
resource;
Accessible:
implementable;
available;
Interoperable:
I1. (meta)data use a formal, accessible, shared, and broadly
I2. (meta)data use vocabularies that follow FAIR principles;
(meta)data;
Reusable:
R1.1. (meta)data are released with a clear and accessible data
usage license;
standards;

FAIR PRINCIPLES
 What does it mean?
 We need an identification mechanism, e.g., PID, PURL, DOI, …
 This mechanism needs to guarantee global uniqueness of the issued identifier, i.e., every time a given
identifier is called, the same resource is points to is retrieved
 This mechanism needs to guarantee persistency of the issued identifier, i.e., what happens when the
identifier scheme is changed?
 What do we need to fulfill this principle?
 How to describe the used identification mechanism?
 How to properly identify the identifier service? I.e., what is the commonly agreed vocabulary that can
represent that a given piece of information is the identifier of a digital resource?
 What is the uniqueness policy?
 How to represent the policy in a computer-actionable way?
 What is the required content of the policy, e.g., uniqueness mechanism?
 What is the persistency policy?
 How to represent the persistency policy in a computer-actionable way?
 What is the required content of the policy, e.g., persistency over updates of the mechanism?
 What is resolved by sending a request to the identifier, the actual digital resource, its metadata, etc.? I.e,
what is the protocol for getting the actual digital resource from its identifier?
F1. (meta)data are assigned a globally unique and persistent identifier

FAIR PRINCIPLES
 If we don’t have the identifier, the digital resource should be described with rich enough metadata that we
can find it through the combination of the items in this metadata
 For different types of digital resources, what are the minimal metadata elements that provide this richness?
 How to describe the metadata in a commonly agreed and computer-actionable way? By using a common way
to represent metadata, tools can be made that are able to interpret metadata from any kind of digital
resource.
F2. data are described with rich metadata

FAIR PRINCIPLES
 The discovery of a digital resource should be possible from its metadata. For this to happen, the metadata
must explicitly contain the identifier for the digital resource it describes.
 How to differentiate the information about the digital resource’s identifier and the one about its metadata
identifier? I.e., a metadata record contains two identifiers, of itself (the metadata record) and of the data
they describe. What is the vocabulary that contains concepts to describe a metadata identifier and digital
resource they describe
F3. metadata clearly and explicitly include the identifier of the data it
describes

FAIR PRINCIPLES
 Most people use a search engine to initiate a search for a particular digital resource. If the resource or its
metadata are not index in a searchable resource, the capability for an individual to find it is substantially
reduced.
 For the data part, a full indexing is equivalent to allowing complete and direct querying on the data, which
may not be feasible every time. An intermediate step would be to select a number of relevant parts of the
data to be highlighted by their metadata, which would be indexed. E.g., in a dataset describing gene
information, it may be relevant to allow the indexing of the unique genes that the dataset has information
about.
 Search engines benefit from common interfaces (or at least interfaces that are described in a commonly
agreed way) to allow the harvesting of the elements (metadata and/or data) to be indexed.
 A common representation format for the metadata also improves the possibility of different searchable
resources to index the metadata records.
F4. (meta)data are registered or indexed in a searchable resource

FAIR PRINCIPLES
 In order to access a digital resource, the requestor needs to be able to implement the used communication
protocol. Therefore, this protocol should be open, free and universally implementable. Moreover, the
protocol should also describe whether authentication and authorization mechanisms are required.
 How to describe the communication (accessibility) protocol?
 What are the elements required in the description of the communication (accessibility) protocol, including
the authentication and authorization procedure?
 How to demonstrate that the protocol is open, free and universally implementable?
A1. (meta)data are retrievable by their identifier using a standardized
communications protocol;
A1.1 the protocol is open, free, and universally implementable;
A1.2. the protocol allows for an authentication and authorization
procedure, where necessary;

FAIR PRINCIPLES
 Cross-reference to data from third-party’s FAIR data and metadata will naturally degrade over time.
Therefore, it is important for FAIR providers to continue to provide descriptors of what the data was to
assist in the continued interpretation of those third-party data.
 How to guarantee long-term persistency of the metadata?
 How to describe that the data (digital resource) referred by the metadata are no longer accessible? Is it
necessary to inform why?
 How to harmonize the persistency of the metadata with the GDPR’s “right to be forgotten”?
available

FAIR PRINCIPLES
 The digital resource is described using a formal, accessible, shared and broadly applicable language

 How to inform the language used to represent the digital object?
 How to provide this information for the metadata? In a meta-metadata?
 How to demonstrate the formality (BNF), accessibility (resolution of the language description document),
shareability and broad applicability of the language (IANA media-type?)?
I1. (meta)data use a formal, accessible, shared, and broadly applicable
language for knowledge representation

FAIR PRINCIPLES
 The metadata values and qualified relations should themselves be FAIR
 Inform which vocabularies are used
 What is the minimal FAIRness for these vocabularies to be considered to follow FAIR principles?
I2. (meta)data use vocabularies that follow FAIR principles

FAIR PRINCIPLES
 Relationships within digital resources and between local and third-party data, have explicit and “useful”
semantic meaning
 Qualify (provide proper semantics) the references to other digital resources
 As per I2, these references (and their qualifiers) should also be FAIR
I3. (meta)data include qualified references to other (meta)data

FAIR PRINCIPLES
 Digital resources should inform who has which rights under which circumstances (license), what is their
provenance and use relevant standards adopt by the community in which the resource has been
created/used
 Inform the usage license:
 What representation format can be used for a computer-actionable license description?
 What are the required concerns that should be present in this description (rights, conditions, …)?
R1.1. (meta)data are released with a clear and accessible data usage
license;
R1.3. (meta)data meet domain-relevant community standards;

FAIR PRINCIPLES
 Inform the digital resource’s provenance information:
 What are the core provenance information?
 What are the community-specific provenance information?
 How to represent provenance? Which vocabularies to use?
 Inform the relevant community standards used by the digital resource (certification):
 How to describe which standards are used?
 How to describe compliance to these standards?
 How to demonstrate that the standards are accepted by a given community?
R1.1. (meta)data are released with a clear and accessible data usage
license;
R1.3. (meta)data meet domain-relevant community standards;

CURRENT SITUATION
Retrieve
non-FAIR
data
Analyse and
prepare
datasets
Combine with
other data
Query combined
data

FAIRIFICATION WORKFLOW
Retrieve
non-FAIR
data
Download/locate file
Identify API call
Identify data access protocol

Retrieve
non-FAIR
data
Analyse and
prepare
datasets
Standard format (XML, RDF, relational DB API, VCF, DICOM, etc.)?
What is the content?
Column/field names?
Relations?
Data domain and range?
Understand the data
Data munging
…

Retrieve
non-FAIR
data
Define
semantic
model
What are the concepts involved?
Relations among concepts?
Existing vocabularies for the concepts and instances?
Analyse and
prepare
datasets
Interoperability
Reusability

Retrieve
non-FAIR
data
Define
semantic
model
Make
data
linkable
Apply the semantic model on the
original data to make it linkable
Analyse and
prepare
datasets
Interoperability
Reusability

Retrieve
non-FAIR
data
Define
semantic
model
Make
data
linkable
Assign
license
Who can access/reuse the data?
Under which conditions?
Analyse and
prepare
datasets
Accessibility
Reusability

Retrieve
non-FAIR
data
Define
semantic
model
Make
data
linkable
Assign
license
Define
metadata
Authors
Version
Distributions
Provenance
…
Analyse and
prepare
datasets
Findability
Accessibility
Interoperability
Reusability

Retrieve
non-FAIR
data
Define
semantic
model
Make
data
linkable
Assign
license
Define
metadata
Deploy
FAIR data
resource
How to make the data and me
tadata available in a FAIR way?
Analyse and
prepare
datasets
Findability
Accessibility
Interoperability
Reusability

Retrieve
non-FAIR
data
Define
semantic
model
Make
data
linkable
Assign
license
Define
metadata
Deploy
FAIR data
resource
Combine with
other FAIR data
Query combined
data
Analyse and
prepare
datasets

FAIRNESS ASSESSMENT CHALLENGES

WHY TO ASSESS?
 Because everybody is talking about FAIR and my resources should be
seen as FAIR, whatever this means?
 To satisfy funders requirements?
 To serve as a guideline for achieving higher levels of interoperability
and reuse with clarity on the concrete benefits (help improve)?

WHAT TO ASSESS?
 Metadata and data?
 Only metadata?
 Only data?
 What do you mean by data?
 In the FAIR principles, data refers to a variety of different resources, e.g.,
“traditional” data, services, software, APIs, vocabularies, ontologies, articles, etc.

HOW TO ASSESS?
 Manual
 Takes advantage of human understandable artifacts, which are currently
prevalent
 May lead to subjective assessments and, therefore, harder to compare
resources
 Harder to scale
 Harder to evaluate FAIR for machines, which is the main goal of the FAIR
principles
 Automatic
 Requires more rigor on the assessed resources
 More likely to produce objective assessments
 Easier to scale
 Able to check if machines can, in fact, “work” with the (meta)data

HOW TO “READ” THE ASSESSMENTS?
 Need for a scoring system
 One score for the 4 aspects of FAIR? Does not seem useful.
 One score per aspect (F, A, I and R)?
 One score per principle? What about the sub-principles?
 Is there a hierarchy among the principles? Is there an order of precedence? Or different
weights?
 Is there an acceptable minimal FAIR level? Should it be across domains and applications
or domain/community-dependent?
 Do we use a pass/fail approach or introduce intermediary compliance levels in
each/some evaluation?
 Need for a visual representation of the scores
 To facilitate quick perception of the FAIRness level, a visual representation of the FAIR
scores is required, e.g., stars, bars, etc.

GENERAL CHALLENGES
 Clarify that nobody has been asked to be 100% FAIR. Many times a lower
FAIRness level is perfectly adequate.
 How to deal with the conflicting forces that, from one side want to push
the communities towards a better (and FAIRer) data landscape and,
from the other side, want to preserve the status quo (existing
“kingdoms”) but labeling themselves FAIR?
 Who will define the assessment criteria?
 Who will execute the assessments based on the defined criteria?
 Should we have a unique set of assessment criteria? Or a core set for
general comparison and domain-specific sets on top of the core for the
specific needs of a given domain/application?

OTHER CHALLENGES
 FAIR should be used as a guideline for achieving higher levels of interoperability and
reuse with clarity on the concrete benefits.
 Improvements on FAIRness can be done incrementally, e.g., first deal with metadata
then go for data.
 We need a combination of (FAIR) infrastructure and (FAIR) community practices.
 How to deal with other aspects not covered by the FAIR principles, e.g., openness,
quality, …?

Q&A – CONTACT INFO
Luiz Bonino
International Technology Coordinator – GO FAIR
Associate Professor BioSemantics – LUMC
E-mail: luiz.bonino@go-fair.org
Skype: luizolavobonino
Web: www.go-fair.org

FAIR Explained

More Related Content

What's hot (8)

Similar to FAIR Explained (20)

More from Luiz Olavo Bonino da Silva Santos (7)

Recently uploaded (20)

FAIR Explained