I was thinking about this too. These bots have a lot of easily recognisable repetitive replies, so could be found and filtered. Until they get smarter of course.
It's an evolutionary battle that may ultimately need a PoW to fix.
Discussion
I don't know if it works.. The question is what does the client when you mark something as spam.. Is this only hidden in the UI or is that actually sent to the network... What would be needed for this approach...