The influence of the widespread adoption of JSON-LD based international standards.

By Noel Vizcaino

De jure or de facto
international
standards

  1. Set the stage for developement to happen. They are pre-requisites.
  2. They have staying power. e.g. Look at Dublin Core and ISO 19115.
  3. Besides other knowledge, they bring structure to the current environment.
  4. Already massive amounts of cross-pollinated compliant data that cannot be ignored.

Metadata serialisation

  1. XML standards played a role and still strong but...
  2. Replaced largely by JSON particularly for APIs.
  3. A downgrade in disguise, easier learning curve and thus broad adoption...
  4. but luckily something happened ...

W3C JSOND-LD: foundational standard

  • An RDF document that is a superset of JSON and thus compatible.
  • Major search engines agreed to make it the future of web metadata.
  • Web metadata: Both schema.org (recommended) and DCAT accepted. Other standards have settled for these two.
  • Google Knowledge Graphs emerges as well as others.
  • Along with JSON-LD context document, any JSON becomes semantic linked data.
  • Modularity and extension by addition favour specialist domains.
  • Most advance RDF serialisation yet AFAIK. (Note: YAML-LD* will follow)
  • A graph as the most flexible schema.
  • Perfectly suited for metadata.

JSON-LD momentum: a reality, today

  • Used by W3C in many standards.
  • FAIR by design.
  • Google Datasets. Knowledge Graph.
  • The whole USofA administration (via DCAT-US) at all levels.
  • CERN Opendataportal
  • Financial: Bloomberg professional services/ Hypermedia API
  • Geospatial: GeoJSON-> GeoJSON-LD -> OGC Earth Observation GeoJSON-LD
  • Smart cities: FIWARE/ETSI NGSI-LD
  • International Data Spaces. Data sovereignity (ISDA Data model). Brokered architecture standardisation effort.
  • Mappings: ISO 19115 (geospatial)<-> DCAT <-> Schema.org . None can be ignored.
  • Life sciences success: bioschemas.org

JSON-LD core ecosystem

  • More than just data or metadata.
  • The serialisation processing is also standardised. (JSON-LD processing)
  • Shaping JSON-LD serialisation data is standardised. W3C JSON-LD framing.
  • Many data views, independent of original data shape.
  • Ready for ingestion in many databases but RDF-aware ones preferred. (To offload processing and storage)

Software Engineering: a Copernican moment for practitioners

  • Where Clean/Hexagonal/Onion architecture meets with Domain-driven Design.
  • Hard dependencies used to go towards the lowest level of abstraction, at the bottom.
  • Now the core, domain model, is at the centre.
  • Presentation layer, Databases, Object stores, networking, go to the exterior ring (top).
  • The core is expected to be stable(r).
  • The idea is to shield the architecture from disruptive change, among other benefits.

Software Engineering and Ontologies

  • Ontologies are an integral part of scientific software development lifecycle
  • Drive complexity out of the domain core.
  • Also part of the functional end product.
  • Ontology standards (bringing the domain terminology) will affect the source code, the design (and the modularity), the queries, the metadata and data, etc.
  • True interdisciplinary collaboration becomes essential.
  • Unix philosophy could be applied to Ontology Engineering. Minimalism and conciseness.

JSON-LD as metadata

  • For domain intense inter-layer communication and eventual storage.
  • RDF-awared databases abound. As well as other Graph related technologies. JSON-LD is just the beginning.
  • It becomes relevant for the querying, communication and storage support.
  • Regulatory compliance will play a role. Each domain have different challenges.
  • ISO Graph Query standard being developed by same SQL committee. Beyond W3C SPARQL and Apache Gremlin.
  • REST and GraphQL friendly. In the case of the latter there are overlaps.
  • APIs should be simple and stable.

Some Links

JSON-LD 1.1
Web metadata. (Google) Dataset.
JSON-LD playground (validation/transformations)

Q&A