We use Gnip to connect to the Twitter firehose. They have a Compliance Stream that “Enables Gnip’s customers to ensure their stored Twitter data complies with Twitter’s terms of service.” (http://support.gnip.com/customer/portal/articles/500113-stream-info#Twitter). Where are the specific policies for stored Twitter data?
I have only found a hint of it in the API terms of service (https://dev.twitter.com/terms/api-terms):
“You should only surface activity that is organically displayed on Twitter.
For example, your Service should execute the unfavorite and delete actions by removing all relevant messaging and Twitter Content, not by publicly displaying to other end users that the Tweet was unfavorited or deleted.”
Specific questions that I’m trying to answer:
- Do we have to immediately remove all information discovered from the Compliance Stream?
- Does it have to be a hard delete in our data store, or can we just flag it as “deleted”?
- When removing a user, do we need to cascade that to remove all of their tweets?
- Do we have to remove any information at all, if we don’t actually display individual users or tweets (i.e. if we only display aggregate calculations and analytics to our users)?