FLAIRS-39 · 2026 · Companion site

JSON-LD 1.2 and Beyond

Extensions for machine learning data exchange.

jsonld-ex is a reference implementation of proposed JSON-LD 1.2 extensions that address twelve gaps in the current specification for machine learning workflows: confidence tracking via Subjective Logic, provenance, security hardening, native validation, vector embeddings, temporal modeling, GDPR compliance, and FHIR R4 clinical interoperability.

This site lets you exercise the extensions live in your browser. Every demo runs the official @jsonld-ex/core package — the same TypeScript implementation that has 100% feature parity with the Python reference.

§ 1Install

python (pypi)

pip install jsonld-ex

javascript (npm)

npm install @jsonld-ex/core

§ 2Quick start

python

from jsonld_ex import annotate, validate_node, Opinion, cumulative_fuse

# Annotate a value with AI/ML provenance
name = annotate(
    "John Smith",
    confidence=0.95,
    source="https://ml-model.example.org/ner-v2",
    method="NER",
)

# Validate against a shape
shape = {
    "@type": "Person",
    "name": {"@required": True, "@type": "xsd:string"},
    "age":  {"@type": "xsd:integer", "@minimum": 0, "@maximum": 150},
}
result = validate_node({"@type": "Person", "name": "John", "age": 30}, shape)
assert result.valid

# Formal confidence algebra (Subjective Logic)
a = Opinion(belief=0.8, disbelief=0.1, uncertainty=0.1)
b = Opinion(belief=0.7, disbelief=0.05, uncertainty=0.25)
fused = cumulative_fuse(a, b)

§ 3Explore the extensions

Each section pairs a runnable demo with the equivalent Python call.

§ 4Cite

bibtex

@inproceedings{syed2026jsonldex,
  title     = {JSON-LD 1.2 and Beyond: Extensions for Machine Learning Data Exchange},
  author    = {Syed, Muntaser and Silaghi, Marius and Abujar, Sheikh
               and Alssadi, Rwaida and Khushbu, Sharun Akter},
  booktitle = {Proceedings of the 39th International Florida Artificial
               Intelligence Research Society Conference (FLAIRS-39)},
  year      = {2026},
}

§ 5Authors

Muntaser Syed, Marius Silaghi, Sheikh Abujar, Rwaida Alssadi, Sharun Akter Khushbu — Florida Institute of Technology · University of Alabama at Birmingham · Daffodil International University.