This is a great effort. I wrote about this at the end of January (https://research.satnam.co/2018/01/30/scammers-impersonating...) and was working on a script last month that did the same thing but I was searching the Streaming API for specific keywords that I knew were triggers for their tweets. I set-up a Twitter account to also identify and report these scammers accounts. The problem I was having was I did not account for the variety of currencies that were being utilized, so I had to write new regexes for the different address types. By then, they had switched up tactics and I hadn't followed up on it since.
