Three licenses, one open project.
Qipu publishes three types of open output: the data pipeline code, the derived datasets, and the editorial content. Each is governed by the license best suited to its use.
What covers what
| Output | Scope | License |
|---|---|---|
| Pipeline code | Python pipeline (extraction, transformation), dbt models, export scripts, configurations | GNU AGPL-3.0 ↗ |
| Derived data | JSON/CSV files published under /public/data/ | ODbL 1.0 ↗ (or the source licence where it imposes no share-alike — see §04) |
| Editorial content | Articles, analyses, methodology, AI summaries | Creative Commons CC BY 4.0 ↗ |
Free pipeline, network-aware copyleft
The pipeline code (Python + dbt + export scripts) is published under the GNU Affero General Public License version 3.
You may
- Clone the repository, study the code, audit it freely.
- Modify it for your own needs (local government, non-profit, research).
- Redeploy it for your territory (multi-city, multi-region, national scale).
- Use it commercially (training, consulting, custom integration).
You must
- Preserve copyright notices and the license in any redistribution.
- If you run a modified version on a publicly accessible server (website, SaaS), publish your modified version under AGPL-3.0 as well.
- Allow users of your modified version to access your source code.
A commercial license is available
If your organisation wishes to integrate our code into a proprietary product without publishing its modifications (e.g. a consulting firm integrating our pipeline into a software suite sold to its clients), a commercial license alternative to AGPL can be negotiated. Contact us to discuss terms.
Derived data: ODbL 1.0
A derived database carries the licence of its most restrictive source. Paris Open Data and Recife publish under ODbL, which requires share-alike, so our JSON/CSV exports derived from them are ODbL too. Exports built solely from French State data (DGFiP, DECP, INSEE, DRIHL) remain under Etalab Open Licence 2.0, and those built from DataSF are public domain (PDDL).
You may
- Download files freely for any purpose, including commercial.
- Republish, modify, and cross with other datasets.
- Embed them in paid products (reports, studies, dashboards).
You must
- Attribute each source dataset in the form its producer requires. For the City of Paris: "Dataset name, Ville de Paris, sous licence ODbL".
- Share-alike (ODbL) — if you redistribute a database derived from our Paris or Recife exports, publicly or as a service, it must itself be under ODbL. This does not apply to exports built solely from French State data (Etalab Open Licence 2.0) or from DataSF (PDDL).
- Credit Qipu as the producer of the derived database: "Qipu · derived database, licensed ODbL".
- Keep the source and source_url fields present in our JSON files — they trace the origin row by row.
- Not imply that Qipu endorses your reuse.
Licence by source
| Source | Source licence | Obligations |
|---|---|---|
| Paris Open Data | ODbL 1.0 | Attribution + share-alike |
| Recife (Dados Abertos) | ODbL 1.0 | Attribution + share-alike |
| French State (DGFiP, DECP, INSEE, DRIHL) | Etalab Open Licence 2.0 | Attribution only |
| San Francisco (DataSF) | PDDL | Public domain — no obligation |
| Vancouver | Open Government Licence – Vancouver | Attribution only |
Articles & analyses: CC BY 4.0
Articles, analyses and methodology texts are under Creative Commons Attribution 4.0 — freely reusable with attribution.
You may
- Quote a paragraph or an entire article in another medium.
- Translate an article and publish the translation.
- Reuse an analysis in a book, a course, a report (including commercial ones).
You must
- Cite Qipu (or the named author when indicated) with a link to the original article.
- Indicate whether you have modified the text (translation, summary, edit).
When a text is AI-generated
Some short summaries published on this site (grant-purpose plain-language rewrites, automatic translations) are generated by language models (Gemini, Claude). They are released under CC BY 4.0 like other editorial content, but their AI provenance must be disclosed if you republish them, in line with Paris Open Data editorial guidance and the European AI Act of August 2024 on transparent labelling of synthetic content.
How to cite us
If you are a journalist, researcher, student, or just curious, here is the recommended citation format for your articles, theses and reports:
Based on qipu.org, a derived database licensed ODbL, built from Paris Open Data (Ville de Paris, sous licence ODbL), Etalab and INSEE (Etalab Open Licence 2.0). Full methodology: qipu.org/methode. Accessed [date].In doubt? Get in touch.
We respond within five business days to questions about license, attribution, citation, commercial integration or redistribution.