The Flemish Government in Belgium has an interoperability program called Open Standards for Linked Organizations (OSLO), which focuses on both technical and semantical interoperability of data and systems used for (digital) government service delivery. OSLO bridges between the geographical data and transactional e-government services. On the semantical level, information is aligned with European standards (ISA² Core Vocabularies and INSPIRE), enriched by data extensions to comply with the local context. On the technical level, we developed RESTFul APIs which build upon the principles of Linked Data (Buyle, 2017).
Because of its extensibility and since it is already a standard for data interchange on the web, Flemish Administrations have chosen the Resource Description Framework (RDF) to facilitate the creation and reuse of machine-readable data. Linked Data is often perceived as complex and verbose. JSON is a popular lightweight data-interchange format, easy for humans to read and write, but a major disadvantage is that the data is not self-describing. JSON-LD, however, is a lightweight Linked Data format. It is easy for humans to read and write and based on the already successful JSON format (json.org). We discovered that JSON-LD was quickly embraced by the developers in designing the APIs. JSON-LD has no formal standardized method for data validation. In data-integration projects, however, there is always a need to validate if data satisfy a set of conditions. We use the Shapes Constraint Language, a W3C Recommendation for describing and validating RDF graphs, to define classes together with constraints on their properties (Knublauch & Kontokostas, 2017).
To unlock the full potential of interoperable services, we are designing hypermedia-driven Web APIs. Therefore the Flemish Government is applying a shared vocabulary, which describes common concepts such as paging and collections, paving the way towards interoperable Web APIs (Lanthaler & Gütl, 2013). This vocabulary, which is understood by both the server and the client, allows us to map the API interactions to low-level HTTP interactions (Lanthaler, 2014). The combination of JSON-LD and re-usable machine readable API contracts allows developers to create generic APIs. This approach can avoid expensive tailor-made connectors.
We have applied this strategy for raising semantic and technical interoperability using JSON-LD both for providing access to the data in our Base Registries as in the high impact project “the citizen’s profile ” that will allow 6.4 million citizens to better access and use public services.
3. E-government driven by Semantical and Technical interoperability
Open Standards for Linked Organisations
@rafke
4. Building upon existing interoperability initiatives
INSPIRE Conference 2018 - titelsessie zelf toevoegen via diamodel
4
OSLO
PERSON
VOCABULARY
OSLO
BUSINESS
VOCABULARY
OSLO
ROAD
VOCABULARY
OSLO
ORGANISATION
VOCABULARY
OSLO
DCAT-AP VL
VOCABULARY
W3C
International
Standards
EU ISA CORE
Vocabularies
and INSPIRE
Local OSLO
Extensions
Public Working Groups
with:
> EU - ISA²
> Federal Government
> Regional Government
> Local Government
> Industry
> Academia
… and more
5. https://data.vlaanderen.be/ns/
INSPIRE Conference 2018 - titelsessie zelf toevoegen via diamodel
5
> Person
> Organisation
> Address
> Building
> Road
> Public Service
> Transactional Services
> Civic issue tracking
> Notification
> Consent (GDPR)
> Contact preferences
> Government Grants
> Public Domain
> Local Decisions
> Political Mandate
6. > Governance Body empowered by Parliament
> A process and method for developing data
standards (ISA²/W3C…)
> Triple Helix approach: Government, Private
Sector, Academia
Governance model
INSPIRE Conference 2018 - titelsessie zelf toevoegen via diamodel
6
Governance
1 16 130 1000+
Domain
models
Contributors Definitions
Collaborative
development
Public review of a
candidate standard
Build reference
implementations
Acknowledge + publish
as standard
8. Bridging the gap between models and implementation
Two distinct use cases
INSPIRE Conference 2018 - titelsessie zelf toevoegen via diamodel
8
Base registries
• Unique identification
of objects
• Exposed via data
standards
Business oriented
web services
• Querying, editing and
transforming data
• From different sources
• To meet business use
cases
9. 1. Embed semantics into data formats
2. Validation of data resources based on semantic models
3. Building a generic hypermedia client
How can we lower costs for data and service integration?
INSPIRE Conference 2018 - titelsessie zelf toevoegen via diamodel
9
10. The Web as a Blueprint
INSPIRE Conference 2018 - titelsessie zelf toevoegen via diamodel
10
11. The Web as a Blueprint
INSPIRE Conference 2018 - titelsessie zelf toevoegen via diamodel
11
http://bit.ly/joinup_uri
Persistent
identifiers:
URIs
Links to
other
information
Dereference-
able HTTP
URIs
Standardised
information
(RDF)
http://data.vlaanderen.be/id/adres/3706808
Koningin Maria Hendrikaplein 70,
9000 Gent
12. The Web as a Blueprint
INSPIRE Conference 2018 - titelsessie zelf toevoegen via diamodel
12
Persistent
identifiers:
URIs
Links to
other
information
Dereference-
able HTTP
URIs
Standardised
information
(RDF)
http://data.vlaanderen.be/id/adres/3706808
Machine
friendlyHuman
Readable
13. The Web as a Blueprint
INSPIRE Conference 2018 - titelsessie zelf toevoegen via diamodel
13
Persistent
identifiers:
URIs
Links to
other
information
Dereference-
able HTTP
URIs
Standardised
information
(RDF)
https://data.vlaanderen.be/ns/adres
https://www.w3.org/ns/locn#adminUnitL2
14. The Web as a Blueprint
INSPIRE Conference 2018 - titelsessie zelf toevoegen via diamodel
14
Persistent
identifiers:
URIs
Links to
other
information
Dereference-
able HTTP
URIs
Standardised
information
(RDF)
http://data.vlaanderen.be/id/adres/3706808
data.vlaanderen.be/id/organisatie/OVO002949
Koningin Maria Hendrikaplein 70
9000 Gent
Agentschap Informatie Vlaanderen
15. There is more to publishing linked data than
data dumps and SPARQL endpoints
INSPIRE Conference 2018 - titelsessie zelf toevoegen via diamodel
15
Source: http://linkeddatafragments.org
high server efforthigh client effort
16. JSON Linked Data (JSON-LD)
A developer friendly Linked Data format, based on the successful JSON format
Machine interpretable semantics through “context file”
100% compatible with JSON and easy to retrofit
× No formal, standardised method for data validation
× Semantics about data are machine readable, service methods and operations are not
Syntactical interoperability and
embedding semantics into services.
INSPIRE Conference 2018 - titelsessie zelf toevoegen via diamodel
16
18. Shared context for new services
Automatically derived from RDF vocabularies
INSPIRE Conference 2018 - titelsessie zelf toevoegen via diamodel
18
{
"@context": "http://data.vlaanderen.be/context/infrastructuur.jsonld"
"@id": "https://example.org/graphs/picknicktable-1234",
"@type" : "Picnictable",
"geometrie" : {"type":"Point","coordinates":"POINT(3.712,51.0248)"},
}
19. Shapes Constraint Language
W3C Recommendation since 20 July 2017
A standard for validating RDF data against a set of constraints
(cardinalities, domain, range, code lists,…)
SHACL files available for all Vocabularies & Application Profiles
Data validation using SHACL
INSPIRE Conference 2018 - titelsessie zelf toevoegen via diamodel
19
20. > As-is: data is semantically annotated using web standards, but the operations
and methods for using the data through APIs are not
> Problem: still high integration costs due to the need for bespoke connectors
between API’s to exchange information in real time.
> To-be: standardisation of frequently used methods (pagination, querying,
CRUD,…) using Hypermedia and API standards/best practices.
Towards generic Hypermedia-driven API clients
INSPIRE Conference 2018 - titelsessie zelf toevoegen via diamodel
20
23. And reusable tooling
INSPIRE Conference 2018 - titelsessie zelf toevoegen via diamodel
23
OSLO² UML modelling using Smart Tags
containing labels, definitions and pointers to re-
used vocabularies (URIs).
Model
UML is transformed to RDF (Turtle) using
an EA-to-RDF tool (OSS)Transform
Generation of HTML spec (vocabulary and
application profile), JSON-LD context and SHACL
using the OSLO Specification Generator
Specs
Validation using Ontology Pitfall Scanner, Turtle
syntax validator and JSON-LD validator.
Published on Github and data.vlaanderen.be
Publish