ted-open-data-examples

Curated SPARQL examples for querying the TED Open Data Service — the EU's public procurement data published as RDF.

This is a knowledge product consumed by multiple TED tools. Each consumer reads its own index file at the repo root; the .sparql files themselves are a shared pool.

Consumer	Index file	Purpose
TED Open Data Service (OP-TED/ted-open-data)	web-library.yaml	Populates the editor's "Query Library" tab
TED Open Data Assistant	web-library.yaml (today), evolving to llm-knowledge.yaml as it gets curated	Example-driven knowledge for natural-language SPARQL generation

A query may appear in either index, both, or neither — the indexes are independent curated views over the same pool.

The assistant currently indexes the same examples that drive the web app (web-library.yaml) — this gets the assistant grounded in the curated examples immediately. As llm-knowledge.yaml gets populated with assistant-specific intent metadata and any LLM-only examples, the assistant will switch to it as its primary source.

Layout

.
├── README.md
├── LICENSE                  # CC BY 4.0
├── CONTRIBUTING.md          # how to add or fix a query
├── web-library.yaml         # web app's Query Library tab + (today) assistant's RAG source
├── llm-knowledge.yaml       # assistant-specific overlay — planned/curated, not yet indexed
├── queries/                 # the .sparql pool — organise as you wish
│   └── *.sparql
├── ci/
│   └── reindex.py           # refreshes the assistant's RAG index from web-library.yaml
└── .github/workflows/
    └── reindex-queries.yml  # runs ci/reindex.py on every push to main

The data is modelled with the eProcurement Ontology (ePO) v4.x. The SPARQL endpoint is:

https://publications.europa.eu/webapi/rdf/sparql

Consuming this from your tool

Pick the index that matches your use case and read it via the GitHub raw URL. Each entry's sparql: field is a path relative to the repo root.

https://raw.githubusercontent.com/OP-TED/ted-open-data-examples/main/web-library.yaml
https://raw.githubusercontent.com/OP-TED/ted-open-data-examples/main/llm-knowledge.yaml
https://raw.githubusercontent.com/OP-TED/ted-open-data-examples/main/<sparql-path>

A consumer typically:

1. Fetches its index file
2. For each entry, fetches the file at sparql:
3. Indexes / displays / processes them as needed

web-library.yaml schema

- category: Notices
  title: Information about notices published on a specific date
  description: This query retrieves information ... published on a specific date.
  sparql: queries/notices-per-day.sparql

llm-knowledge.yaml schema

See the header comment in llm-knowledge.yaml. Each entry describes the intent the example answers, the ePO terms used, and any gotchas worth grounding the assistant on.

Contributing

See CONTRIBUTING.md. In short: open a PR against develop; entries are published when develop merges to main.

When develop merges to main, a GitHub Action (reindex-queries.yml) automatically refreshes the TED Open Data Assistant's RAG index from web-library.yaml so newly added queries become available to the assistant. No manual reindexing step is needed.

License

The query examples and documentation in this repository are licensed under Creative Commons Attribution 4.0 (CC BY 4.0) — see LICENSE.