A license is not mandatory in a data package but open data needs one


#1

The Frictionless Data specification does not require a license property.

Open data needs a license.

I’m wondering why the license property isn’t required at either the package or resource levels?

How would it be viewed if the app I’m making to create data packages required a license in datapackage.json?


#2

Technically open data doesn’t always need a licence, as some open data is unencumbered by intellectual property rights (and therefore has no owner with legal standing to issue a licence).


#3

Thanks @owenboswarva

So my question is now, should the Frictionless Data specification explicitly support identifying that:

  • the work is licensed under an open licence
  • copyright has been waived
  • copyright has expired
  • the work is public domain?

Providing clarity on the licence, waiver, or other mechanisms that allow reuse would make a data package more useful.


#4

Hello,
I have made research of OGD licensing, maybe it could be useful: https://www.researchgate.net/publication/281777766_Open_Government_Data_Licensing_Framework
Best regards,
Martynas


#5

@Stephen among other answers:

  • Data Packages are designed for all types of data not just open data
  • Someone may want to create a package they don’t yet know the license status of

General point: the license field is indicative not binding (as stated in the spec) – though I guess in some cases the license field may be used as the explicit statement of license …


#6

So a user interface for collecting the licence should allow any of:

  1. a name (an Open Definition license ID) from which path and title could be derived.
  2. a path and optionally a title to a licence that is not assessed as an Open Definition conformant license. As the path can be a url-or-path the license could be a file within the data package.
  3. null

Perhaps it would be useful in the specification to offer guidance on the use of licenses, e.g.:

  • prefer an open license
  • If not an open license, prefer the license to be at a URL
  • If the licence is not at a URL but included in the data package, prefer .txt or .md file formats over other file formats (.doc or .pdf)
  • If the license status is unknown, encourage a comment in the README.md file
  • If there is no copyright in the data or copyright has expired, encourage:
  • If copyright has been waived, encourage:
    • the use of CC0 or,
    • provide a URL to another waiver. or
    • include the waiver in the Data Package
  • Where copyright in the data is held by multiple parties, explain this in the README.md file
  • Include other copyright and attribution information in the README.md file, e.g.:
    • preferred attribution text or attribution URL
    • copyright notice, year

I wonder if this is worth writing up as a Frictionless Data Pattern?

Edit: I’ve drafted a guide at https://github.com/frictionlessdata/website/issues/157


#7

A post was split to a new topic: Open Data Business Models