Problem description:
In my organization, we develop a new dataset by thoroughly processing, filtering, cleaning and combining several pre-existing datasets published on HF or other public repositories with a permissive license. We publish our obtained dataset on HF with a permissive license.
At some point, the owner of one of the original datasets changes their licensing schema and decides to withdraw the permissive license and forbid the use/re-publication of the data.
Since we have a DOI assigned to our published dataset, we cannot delete it or change its status to private.
So, we are incurring legal issues in case the original data owner finds out that we have published a dataset containing their data.
How should we handle these cases?
Is there a way to delete or hide a published dataset?
If you are on HF, @+username is the fastest way. Though I called her above.
If you’re sending an email… this is the only email address I can find, but I’m pretty sure it’s not… legal@huggingface.co
You may also be able to use the email function on the forum, but this is a little harder to notice because you have to come to the forum to see that it has been received.
Discord is apparently the fastest way to do this if you are in a hurry.
Now that the invite link to Discord is new, it’s fine, but it would be nice to have the ability to notify Discord of Notifications on the Forum and Hub. I don’t know if this is technically possible. (If it’s a separate account, you’d have to set it up yourself. Or maybe a shared staff issue posting space?)
Since this forum also seems to be a pre-existing well-known program, or is it possible that there are good built-in features lying around?
One of the reasons for the unusual dis-communication between users and staff within HF is probably the disconnect between Discord and the Hub.
While there are inconsistencies in communication that I see in HF, I will also play the duct tape role of humanly intervening as much as possible, but I can’t help you with the details of the LLM or the problems inside the server. Even in the past few days, I have been unable to respond to at least five cases. It’s just a temporary stall by a mob character like me.
If HF can fix this situation, they better fix it, and maybe implementing a help function to Flag function would be a good idea.
Good morning,
I have solved the issue by sending an email to website@huggingface.co.
They removed the DOI from the dataset and now I’m able to remove or change the status of the dataset.