Influence of fake news in Twitter during the 2016 US presidential election [pdf]
arxiv.org
arxiv.org
Reminds me of previous fake papers... although that may be the point. If not... whew. I bet you couldn't pay them for the "data" from step3, since it does not exist. I'm getting a kick out of the "source notes" field.
I have thought quite a bit how to actually do this. It's more than a hobby of mine to track how disinformation works and spreads. You would need to pick specific testable predictions and claims from a broad sample of historical news, the older the better, and start there. Things that can be measured with numbers. Predictit and other betting site questions would be a good foundation.
lol, just noticed they disabled issues.