Question About Redistributing Public GitHub WatchEvent Data for Academic Research #213341
Replies: 1 comment 1 reply
|
This is a good question to ask before publishing. The short answer is: using GH Archive data for open-access research is supported; publicly redistributing a derived event dataset is the part that needs permission. Those are two different things under GitHub's terms. What the policies actually sayTwo clauses in GitHub's Acceptable Use Policies are relevant:
Section 7 is what lets you do the study. Section 6 is what makes shipping a dataset of raw-ish events a different question: a downloadable file that someone else can query is much closer to "reproduce/duplicate" than a paper describing your findings. Section H of the Terms of Service (API Terms) also says all API use is subject to the ToS and Privacy Statement. GH Archive collected the data through GitHub's APIs, so their MIT-licensed tooling does not give you a licence over the underlying event data — it covers the archive project's code, not the events themselves. What I would doPublish the paper open access. That is an explicit condition in Section 7, and it is the easiest way to stay inside the research carve-out. Publish the method, not the dataset. This is the cleanest route and it is fully reproducible:
Anyone can then rebuild your dataset from If you really need to ship the dataset, ask for written permission. Section 6 requires express written permission for reproduction, and only GitHub can give you that. Use the support/contact route for research partnerships rather than a community discussion — no one here can grant it. Say exactly what you want to publish, which event types and date range, and that it is non-commercial and open access. On pseudonymisationReplacing
"Non-personal" in Section 7 is doing real work here. A pseudonymous interaction graph may still be personal data under GDPR if re-identification is reasonably possible, so the safeguards above matter for the research-carve-out argument as much as for ethics review. Licensing the derived workYou can license the code you write (your extraction and pseudonymisation scripts) however you like — MIT is fine. For the data, you cannot grant rights you don't have, so a Useful primary sources: GitHub Acceptable Use Policies §6–7, GitHub Terms of Service §H, and GH Archive. I'm not a lawyer, so treat this as practical guidance rather than legal advice — for a public dataset release, get the written permission the policy asks for. If you want to see more of my write-ups on data and tooling, I'm on GitHub at https://github.com/P-r-e-m-i-u-m |
Uh oh!
There was an error while loading. Please reload this page.
🏷️ Discussion Type
Question
💬 Feature/Topic Area
Other
Body
Hi GitHub Community,
I'm exploring the possibility of using publicly available GitHub WatchEvent data collected through GH Archive to construct a dataset for non-commercial academic research.
The dataset would contain historical user–repository interactions, with user identifiers pseudonymized and repository identifiers and timestamps retained.
Before proceeding with any public release, I'd like to clarify a few questions regarding GitHub's data usage and redistribution policies:
Data Redistribution: Does GitHub permit researchers to publicly redistribute processed datasets derived from public WatchEvent records for non-commercial academic purposes?
Privacy: Would replacing original user identifiers with pseudonymous IDs be sufficient, or are additional safeguards required when publishing user–repository interaction sequences?
Licensing: Is there a recommended license for such derived datasets? My understanding is that GH Archive's MIT license does not automatically cover the underlying event data.
Alternative Approaches: If direct redistribution is restricted, would publishing scripts and instructions that allow researchers to reconstruct the dataset from GH Archive be permitted?
I understand that publicly accessible data is not necessarily freely redistributable, and I'd like to ensure compliance with GitHub's Terms of Service and privacy policies.
Any guidance or pointers to relevant official documentation would be greatly appreciated. If this requires clarification from GitHub's Legal or Policy Team, I'd also appreciate information on the appropriate contact channel.
Thanks in advance!
All reactions