The bilpcoin spam farm (at least 10 accounts, although not all are active spammers) has been a source of irritation for me over the last year. The repetition combined with the length is insane.
Another account, @aquarius.academy, seems to feed from a list of dubiously-soirced "inspirational" quotations and has taken to replying to old posts long past payout. I try to keep an eye on it and down ote any upvotes while alerting those voters to the scam. Can you filter accounts rotating through such lists?
Finally, while I see less of it now, the ol' "nice post" and "follow for follow?" stuff has declined, but still seems to crop up from new users unfamiliar with our netiquette or bad actors trying the laziest possible methods.
I have a list of accounts that I'm tagging as human or not human, and it is my intent to focus on "human" interaction, but yes, I will be publishing the list of accounts that I am tagging as non human in my next post on this topic.
There is a filter in my data set for this, as I've set it up as a flag in the table I'm building.
I am really keen to see how many posts are just "Thank You" and the number of these, and how many unique users do things like that. That will be my next group by command on my data set, I am just processing some more rows to get a bit of a longer term picture than the week I sampled here.
A week is just a snap shop, whereas a few months will be meaningful, and enable week on week tracking for trends, word clouds, key words and overall sentiment analysis and more.
There's so much that can be done.
It's nice we have an open system for that kind of analysis. Web2 social media is full of the same kind of nonsense, but transparency is antithetical to their model.