Generate DOI for datasets

[arokem]

If the datasets are to be linked in papers, it would be important for the datasets to have a persistent identifier. You could programmatically generate these using Zenodo (https://zenodo.org/).

[frankong]

That’s a good idea. I tried to look into it, but it seems like you would have to upload the data to zenodo to get a doi. Do you know of a way to go around that?

[frankong]

The datasets right now have unique links, which will not be changed

[arokem]

What's the problem with uploading the data to Zenodo? It would be a mirror of the data on mridata.org. I think that you can upload up to 50 GB of data with no problem, and I am willing to bet that they would give you more space if you explained your use-case.

Another option is to mirror to an institutional repository. For example, I can tell you from personal experience that the Stanford SDR is a fantastic place to put data, and they create a PURL (which is as good as a DOI).

Another (and I think least favorable) option is to become a DOI provider. But that's expensive and complicated, so I would recommend against that.

[arokem]

And unique links are good, but won't have the same long-term guarantees that a DOI would have.

…

On Wed, Jun 20, 2018 at 11:53 AM Frank Ong ***@***.***> wrote: The datasets right now have unique links, which will not be changed — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#10 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAHPNrfeQ5Kg2w2rTOtnK14zxEwxHqcCks5t-hucgaJpZM4Uu3Mb> .