Licensing datasets
This page is for managers and editors who must state under which terms each dataset may be re-used. It explains what makes a license open, how to record the license on a dataset page in Exponential, and the questions to settle with your legal counsel before you choose. It does not recommend a particular license: that choice belongs to your organization.
Note This page describes features and common practice; it is not legal advice. Check the law that applies to your organization with your legal counsel.
Contents
- What an open license must allow
- Questions to settle before choosing
- Recording the license in Exponential
- Showing the license to people and machines
- Data that cannot be open
What an open license must allow
The Open Definition 2.1 lists what a license must permit to be "open": use, redistribution, modification, separation of parts, compilation with other works, non-discrimination against persons, groups or fields of use, and propagation of the same rights to everyone who receives the work. It allows a few conditions, such as requiring attribution or that changed versions stay under the same terms.
In the EU, Directive (EU) 2019/1024 addresses licenses for the re-use of public sector documents. Your national law and policy may name the licenses public bodies should use.
Questions to settle before choosing
| Question | Why it matters |
|---|---|
| Does the body hold all rights in the data? | Data bought from or produced with third parties may carry their terms |
| Does national policy prescribe a license? | Many jurisdictions name a default for public data |
| Is attribution wanted? | Decides between licenses with and without an attribution condition |
| Does the data contain personal data? | Personal data generally is not published as open data; anonymize or aggregate first (see "Privacy checks before publishing data") |
| Are there security, commercial or third-party confidentiality reasons to hold back? | Some datasets, or columns, may need to stay unpublished |
| Who signs off? | Name the role that approves the license of each dataset |
Recording the license in Exponential
Give the dataset class a required License field of the datatype Selection (ezselection), with one option per
license your body approved. A selection keeps the wording identical on every page and makes it possible to list
datasets by license.
- Setup > Classes, group Content, the class
dataset, Edit. - Under Add attribute choose Selection, press Add, name it
License, identifierlicense, tick Required, untick Translatable (the license is the same in every language). - Add one option per license, with its official short name, for example the name and version as the license steward publishes them.
- Press OK.
If you need the license text or its address on the page, add a second field of the datatype URL (ezurl) named
license_url, or keep one page per license and relate datasets to it with an object relation.
For the steps of the class editor see Content classes for websites.
Showing the license to people and machines
- On the page: show the license name and a link to its text near the download links, not in the footer.
- In files: where the format allows, add the license and source in a header or a separate
READMEfile in the distribution. - In feeds: an RSS export can map one attribute to the description or category of each item
(RSS feeds); a catalog that needs
dct:licensein DCAT has to be written by your own template (see "DCAT and national portals"). - In the API: the REST API returns every field of the object, the license field included (REST API v1 and v2).
Data that cannot be open
Not every dataset may be open. Typical reasons: personal data, security, third-party rights, or law that limits re-use. You can still publish a description of such a dataset with "Not available for re-use" as its license option, so people know it exists and whom to ask.
Related pages
- Designing a dataset class
- Choosing datatypes
- Licensing explained for buyers: the license of the software, which is a different question
- Free documentation: this page is free to copy, translate and improve under the GNU FDL 1.3.