Dataset licensing

Every project that's going to produce a dataset other people might use needs a license — the legal terms under which they can use, modify, and redistribute it.

Most dataset owners haven't read the Creative Commons family in detail, so AfriAnnotate has a guided decision-tree wizard that walks through 5–8 plain-English questions and recommends one of nine license outcomes.

This is AfriAnnotate-specific — useful for academic / NGO datasets where the license decision is often deferred or made wrongly because the team didn't have time to research CC terms.

Where it lives

Project → Settings → Licensing.

Visible to Owner and Admin. Manager sees a read-only summary of the chosen license. Reviewer / Annotator don't see the tab — but the license badge is visible to everyone in the project header so annotators know what terms their work is being collected under.

The decision tree

The wizard asks (in order, with branching):

1. Is this dataset going to be public? If no → internal team only → Proprietary; specific external parties with gated access → Restricted-use.
2. Can commercial use be allowed? Filters CC vs CC-NC family.
3. Must derivative works be allowed? Filters CC-ND vs the rest.
4. Must downstream users credit you? Filters CC0 vs CC-BY.
5. Must derivatives use the same license? Filters CC-BY-SA vs plain CC-BY.

The branches map to these nine outcomes:

Outcome	SPDX identifier	Plain English
CC0 1.0	CC0-1.0	Public domain. Anyone can do anything, no credit required.
CC-BY 4.0	CC-BY-4.0	Use freely, including commercially, with credit.
CC-BY-SA 4.0	CC-BY-SA-4.0	As CC-BY, but derivative works must use the same license.
CC-BY-ND 4.0	CC-BY-ND-4.0	Use freely with credit, but no modifications (or share modifications privately).
CC-BY-NC 4.0	CC-BY-NC-4.0	Non-commercial use only, with credit.
CC-BY-NC-SA 4.0	CC-BY-NC-SA-4.0	Non-commercial, same-license derivatives, credit.
CC-BY-NC-ND 4.0	CC-BY-NC-ND-4.0	Non-commercial, no derivatives, credit. The most restrictive Creative Commons option.
Proprietary (no public release)	proprietary	Internal-team use only. Not redistributed outside the project's organisation.
Restricted-use (gated access)	restricted-use	Released only to specific approved parties under a custom data-use agreement.

What the wizard persists

The chosen license is stored as a JSON blob on Project.license:

{
  "identifier": "CC-BY-SA-4.0",
  "name": "Creative Commons Attribution-ShareAlike 4.0",
  "url": "https://creativecommons.org/licenses/by-sa/4.0/",
  "summary": "Use freely, including commercially, with credit. Derivative works must use the same license.",
  "restricts_commercial": false,
  "restricts_derivatives": false,
  "requires_attribution": true,
  "requires_share_alike": true,
  "chosen_at": "2026-03-15T14:32:11Z",
  "chosen_by": 42,
  "reasoning": [
    {"question": "Will this dataset be public?", "answer": "yes"},
    {"question": "Can commercial use be allowed?", "answer": "yes"},
    {"question": "Must derivatives be allowed?", "answer": "yes"},
    {"question": "Must downstream users credit you?", "answer": "yes"},
    {"question": "Must derivatives use the same license?", "answer": "yes"}
  ]
}

The reasoning trail is the killer feature — six months later when someone asks "why did we pick CC-BY-SA?", the answer is recorded in the project itself.

Changing the license later

Owners can re-run the wizard at any time. Picking a different outcome does not retroactively change the license under which existing exports were distributed — recipients of the v1 export have v1 license terms. The new license applies to subsequent exports.

For datasets where you need the license to flow downstream correctly, include the SPDX identifier in your export manifest (exports/manifest.json carries license.identifier and the URL).

Where the license shows up

Once chosen, the license:

• Appears in the project header as a badge next to the name
• Is included in the data export manifest (the JSON sidecar that ships alongside any export)
• Surfaces in the organisation-wide dataset listing for org admins doing inventory ("which of our 47 projects use CC-BY-SA?")
• Is included in the annotation export JSON so individual task-level exports carry their license context

Why a tree at all

Most dataset owners haven't read the CC family in detail. They pick "CC-BY-SA" because it sounds permissive, not realising it requires derivatives to use the same license — which is a non-starter for many downstream consumers.

The wizard:

• Translates the questions into plain English ("can people use this for-profit?" rather than "do you allow commercial use?")
• Prevents impossible combinations (e.g. you can't have ShareAlike and No Derivatives at the same time)
• Forces a deliberate choice for restrictive options (the wizard shows a warning when the tree lands on CC-BY-NC-ND because it's the most restrictive)

Quick reference

If you already know what you want:

You want…	Pick…
Public domain — no strings attached	CC0
Maximum reuse, just want credit	CC-BY
Like CC-BY but force the work to stay open	CC-BY-SA
Credit required, no remixing	CC-BY-ND
Academic / research only, no for-profit	CC-BY-NC
Same but with share-alike	CC-BY-NC-SA
Most restrictive: no commercial, no remixing	CC-BY-NC-ND
Internal team only, no release	Proprietary (proprietary)
Gated release to named parties under a data-use agreement	Restricted-use (restricted-use)

What's next

• Project setup → — the full project create flow
• Data security → — per-project access controls (different from licensing — security controls who can see the data, licensing controls what they can do with it)
• Consent library → — annotator-side consent for data use (different again — consent is the annotator's agreement to participate, license is the downstream consumer's terms)