Semantic challenges in sharing dataset metadata and creating federated dataset catalogs. The example of the CIARD RING.

Semantic challenges in sharing
dataset metadata and creating
federated dataset catalogs
The example of the CIARD RING
Valeria Pesce (Global Forum on Agricultural Research and Innovation)
Linked Open Data in Agriculture
MACS-G20 Workshop in Berlin, September 27th–28th, 2017

Semantics involved in describing datasets
Name
Owner
Type of data
Topic(s)
Data standards used
Data structure
Place of collection
Date of collection
Distribution(s)
[…]
Dataset
for describing
datasets, e.g. DCAT
or DataCube
Metadata
vocabulary or
ontology
Authority data
KOS / thesaurus
“Value vocabularies” / Knowledge Organization systems
Data of type Organization,
e.g. VIAF
“Descriptionvocabularies”
KOS / classification
Concepts suitable for
organizing by Topic, e.g.
AGROVOC
Concepts describing
Types of data
Dataset structure
Dimensions
Attributes
Measures
Value lists
for describing data
structures, e.g.
DataCube or STAT-
DCAY
Metadata
vocabulary or
ontology
Distribution
Protocol
URL
Format
Size
for describing
geospatial entities,
e.g. GML
Schema
No universal
agreed model or
vocabulary!
for describing
distributions, e.g.
DCAT or VOID
Metadata
vocabulary or
ontology

Name
Owner
Type of data
Topic(s)
Data standards used
Data structure
Place of collection
Date of collection
Distribution(s)
[…]
Dataset
for describing
datasets, e.g. DCAT
or DataCube
Metadata
vocabulary or
ontology
Authority data
KOS / thesaurus
“Value vocabularies”
e.g. VIAF
AGROVOC
Concepts describing
Types of data
Dataset structure
Dimensions
Attributes
Measures
Value lists
Metadata
vocabulary or
ontology
Distribution
Protocol
URL
Format
Size
for describing
e.g. GML
Schema
The
dataset
resource
Metadata
vocabulary or
ontology
for describing data
structures, e.g.
DataCube or STAT-
DCAY for describing
distributions, e.g.
DCAT or VOID

Name
Owner
Type of data
Topic(s)
Data standards used
Data structure
Place of collection
Date of collection
Distribution(s)
[…]
Dataset
Metadata
vocabulary or
ontology
Authority data
KOS / thesaurus
e.g. VIAF
AGROVOC
Concepts describing
Types of data
Dataset structure
Dimensions
Attributes
Measures
Value lists
for describing data
structures, e.g.
DataCube or
STAT-DCAT
Metadata
vocabulary or
ontology
Distribution
Protocol
URL
Format
Size
for describing
e.g. GML
Schema
The
dataset
structure
Metadata
vocabulary or
ontology
for describing
datasets, e.g. DCAT
or DataCube
for describing
distributions, e.g.
DCAT or VOID

Name
Owner
Type of data
Topic(s)
Data standards used
Data structure
Place of collection
Date of collection
Distribution(s)
[…]
Dataset
Metadata
vocabulary or
ontology
Authority data
KOS / thesaurus
e.g. VIAF
AGROVOC
Concepts describing
Types of data
Dataset structure
Dimensions
Attributes
Measures
Value lists
Metadata
vocabulary or
ontology
Distribution
Protocol
URL
Format
Size
for describing
e.g. GML
Schema
Dataset
serialization
for describing
datasets, e.g. DCAT
or DataCube
for describing
distributions, e.g.
DCAT or VOID
Metadata
vocabulary or
ontology
for describing data
structures, e.g.
DataCube or STAT-
DCAY

Semantics needed to describe datasets
Name
Owner
Type of data
Topic(s)
Data standards used
Data structure
Place of collection
Date of collection
Distribution(s)
[…]
Dataset
Metadata
vocabulary or
ontology
Authority data
KOS / thesaurus
e.g. VIAF
AGROVOC
Concepts describing
Types of data
Dataset structure
Dimensions
Attributes
Measures
Value lists
Metadata
vocabulary or
ontology
Distribution
Protocol
URL
Format
Size
for describing
e.g. GML
Schema
Reference value vocabularies
for describing
distributions, e.g.
DCAT or VOID
Metadata
vocabulary or
ontology
for describing
datasets, e.g. DCAT
or DataCube
for describing data
structures, e.g.
DataCube or STAT-
DCAY

Semantics of the values
• Standardization of the values, e.g. for “thematic
coverage” or “dimensions” of datasets, “format”
or “protocol used” of distributions etc.
• The value should be standardized, possibly a URI
• The value should be part of an authority list /
code list
RDF dataset vocabularies normally treat these
values as resources, so identifiable by URIs, BUT…
a) Often strings are used
b) Often a local concept URI is used
c) THERE AREN’T AGREED KOSs FOR EVERYTHING!
Authority data
KOS / thesaurus
Examples:
- VIAF registry
- Library of Congress
- ORCID
Examples:
- AGROVOC
- CABI thesaurus
Publisher metadata
Thematic metadata
Code list
Examples:
- ICASA variables
- CF conventions RDF
Dimensions
Authority data
Examples:
- GeoNames
- FAO Geopol Ontology
Geographic metadata

Examples of relevant value vocabularies
• Domain
• Agricultural concepts, topics: AGROVOC > GACS (or agreed subsets)
• Crop names: AGROVOC, Crop Ontology
• Soil types: USDA Soil Taxonomy, INSPIRE Registry
• Dimensions / variables: ICASA variables ( RDF?), CF conventions RDF
• Cross-domain
• Authority lists of organizations, projects: VIAF, CERIF, ORCID?
• Geospatial / geopolitical data: GeoNames, FAO Geopolitical Ontology
• Data formats / data standards? AgriSemantics Map of Standards
• File formats: IANA types ( RDF?), W3C formats
• Agreed list of types of data?
• Units of measure?
• Authority list of licenses (OpenDefinition list?)
Not for everything we would need!

The CIARD RING
The CIARD RING is a federated and curated catalog of agri-food datasets
and data services
http://ring.ciard.net
• a primary catalog (providers can catalog individual data services and
datasets directly in the RING) exposing all metadata as RDF
• a federated catalog (it harvests dataset metadata from other catalogs)
Federated catalogs so far
Dataverse
catalog
datasets services

Semantics in the RING dataset hub
• Dataset description vocabularies
The RING uses a combination of the DCAT-AP model + the VOID
vocabulary and the DataCube vocabulary
 a “RING DCAT profile” will be published
• Value vocabularies
• Domains: local RING Domains SKOS, based on FAO and USDA top-level
classifications of domains
• Types of data: local RING ”Types of data” SKOS, aligned with GODAN Ag Sector
Package types of data
• Topics: AGROVOC
• Countries: FAO Geopolitical Ontology
• Data formats / data standards: AgriSemantics Map of data standards
• File formats: “mapped” to IANA types and W3C formats when applicable

Examples of semantics in federated datasets - 1
• IFPRI dataset in Datahub (DCAT RDF)
dcat:keyword:
strings
dct:format: string
dcat:mediaType: string (IANA syntax)

Examples of semantics in federated datasets - 2
• IFPRI dataset in Dataverse (OAI-PMH XML response)
Description metadata: no published vocabulary
Geographic scope: string
Local keywords

Examples of semantics in federated datasets – 3a
• EuroStat dataset in EU Data Portal (DCAT RDF) (1)
dct:subject: URI of EUROVOC thesaurus
concept URI from EU KOS
additional property for dataset type
concept URI from EU KOS of licenses

RING: enriching and linking semantics
Match or partial match with
synonym in local KOS >>
Becomes a RING resource of type
skos:Concept and dc:FileFormat
with local URI
Becomes dct:conformsTo as in DCAT
with resource of type dc:Standard
linked to URI of same standard in
AgriSemantics Map of Data Standards

RING: linking semantics
“Uganda” matches a local concept of type
skos:Concept and dc:Location,
mapped to URI of Uganda country in
FAO Geopolitical Ontology
“Women” narrower of “Socio-
economic data” local concept,
mapped as closeMatch to URI of
“socioeconomic development”
concept in AGROVOC

Queries can leverage LOD mappings - 1
Example: To get all datasets with geographic coverage of “Uganda" using the
Geopolitical Ontology URI for “Uganda"
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX dc: <http://purl.org/dc/terms/>
PREFIX dcat: <http://www.w3.org/ns/dcat#>
PREFIX owl: <http://www.w3.org/2002/07/owl#>
PREFIX skos: <http://www.w3.org/2004/02/skos/core#>
DESCRIBE ?dataset ?distro WHERE {
?dataset rdf:type dcat:Dataset .
?dataset dcat:distribution ?distro .
?dataset dc:spatial ?topic .
?topic owl:sameAs <http://aims.fao.org/aos/geopolitical.owl#Uganda> .
}
URI of the “Uganda” in the
Geopolitical Ontology

Example: To get all datasets on topic "Livestock" using the AGROVOC URI for "Livestock"
PREFIX owl: <http://www.w3.org/2002/07/owl#>
?dataset dcat:theme ?topic .
?topic owl:sameAs <http://aims.fao.org/aos/agrovoc/c_4397> .
}
URI of the “Livestock” concept in
the AGROVOC thesaurus

Example: To get all datasets complying with the INSPIRE specification for soil
?distro dc:conformsTo ?standard .
?standard owl:sameAs <http://vest.agrisemantics.org/node/19915> .
}
URI that identifies the
INSPIRE specification for Soil

Conclusions
• The major semantic challenges when integrating (meta)data arise from the
lack of use of common value vocabularies, not so much from the use of
different description vocabularies / schemas / formats.
• In most cases the lack of good semantics in the (meta)data at the level of
value vocabularies is not due to ill will or lack of awareness, but to the
constraints posed by most dataset management tools.
• The machine-readable layer and the SPARQL endpoint of the RING are not
for the end users: we expect this layer to be used by developers to build
added-value services for the end users on top of the featured datasets.

Relevant vocabularies, catalog tools, catalogs
• DCAT: http://www.w3.org/TR/vocab-dcat/
• DCAT AP: https://joinup.ec.europa.eu/asset/dcat_application_profile/home
• STAT-DCAT: https://joinup.ec.europa.eu/asset/stat_dcat_application_profile/home
• DataCube: http://purl.org/linked-data/cube#
• VOID: http://rdfs.org/ns/void-guide
• DDI-RDF Discovery Vocabulary: http://rdf-vocabulary.ddialliance.org/discovery.html
• VIVO Datastar: http://sourceforge.net/projects/vivo/files/Datastar%20ontology/
• CERIF for datasets: https://cerif4datasets.wordpress.com/c4d-deliverables/
• CKAN: http://ckan.org/
• Dataverse: http://dataverse.org/
• Datahub: http://datahub.io/
• DataCite: http://search.datacite.org/ui?q=subject%3Aagriculture
• Re3data: http://www.re3data.org
• OpenAIRE: https://www.openaire.eu/
• CIARD RING: http://ring.ciard.info

Thank you
Semantic challenges in sharing dataset metadata
and creating federated dataset catalogs.
The example of the CIARD RING
Valeria Pesce (GFAR)
valeria.pesce@fao.org
Linked Open Data in Agriculture
MACS-G20 Workshop in Berlin, September 27th–28th, 2017

Semantic challenges in sharing dataset metadata and creating federated dataset catalogs. The example of the CIARD RING.

More Related Content

What's hot (20)

Viewers also liked (11)

Similar to Semantic challenges in sharing dataset metadata and creating federated dataset catalogs. The example of the CIARD RING. (20)

More from Valeria Pesce (16)

Recently uploaded (20)

Semantic challenges in sharing dataset metadata and creating federated dataset catalogs. The example of the CIARD RING.