3

Is it okay to use data (for machine learning purposes) that specify that they can only be used for "research-only" purposes (or "non-commercial") in an open source product?

If so, is it okay to take donations from my open source product, or would that be considered as a "commercial" purpose?

For example: using face recognition "research-only" data for an open source face recognition model.

migrated from law.stackexchange.com Mar 6 '17 at 11:36

This question came from our site for legal professionals, students, and others with experience or interest in law.

  • What do you mean by "use"? Are you going to reproduce the data in the open source product? In that case, the software cannot be open source (at least the data part must be under a research-only / non-commercial license). – Zimm i48 Mar 7 '17 at 6:53
  • The data would be used for training a model. What is meant by that is, I learn patterns/features from the data and then encode those patterns in some "object" for predictive purposes (e.g. for face recognition: who's face does this belong to?). The object does not explicitly maintain a copy of the data or distribute it, but in some sense it has encoded the data (though you cannot "decode" to get it back). – miguel.martin Mar 7 '17 at 7:06
  • Really I was just curious to know whether or not an "open-source" product would be ok to contain "research-only" components. My assumption is that I would have to distribute my open source product under some license that states "research only purposes, non-commercial use" or at least the "object"/model that utilises that research-only data. – miguel.martin Mar 7 '17 at 7:08
  • 3
    When in doubt, ask the people who released the data. – Philipp Mar 7 '17 at 10:46
6

Given the details you give in comments, we can consider that your software would most likely be a derivative of the original research-only / non-commercial data.

Then the question is what the license has to say about derivatives? Does it even allow them? If it does not explicitly allow them, you are most likely not authorized from publishing any software based on this data, under any license.

If derivatives are allowed, under what conditions? In the case of a CC-BY-SA-NC license, you would have to keep the license for any derivative that you distribute, and this is not an open source license. But it could also be more lax, in which case maybe an open source license would be allowed.

In conclusion: it all boils down to exactly what is written about derivative works in the license.

0

Most likely you may not use the data. Open Source software explicitly allows usage of the software for ANY purpose(including commercial, do not mistake it for closed software). To give others a right to use something for any purpose, you have to have this right in the first place.

The "only for research" clause explicitly restricts your usage to only research use, not for all purpose. Thus, almost certainly you may not use it in an Open Source project.

A separate issue is if the license gave you the right to use the data for any purpose. Even then I am not sure if it would also allow for modification and redistribution (keep in mind GPL has all of these activities mentioned)

Your Answer

By clicking "Post Your Answer", you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.