Funny, I've seen requests like this but this thought never occurred to me till I read this post.
Most people who spam messages spam a set string over and over. So what one can do, in addition to blacklisted terms (a feature flyrocket's ODom proxy already has), is write a routine that tests to see if a person repeated the same text within a certain amount of time, say 30 seconds or so. If they have, simply ignore the message by prohibiting it from displaying on the screen. If a person has been flagged for repeating a message make sure to reset the timer so that if the person again repeats the same string in the next 30 seconds it will ignore that one too.
@
flyrocket, this feature doesn't seem like it would be too hard to code. It would be a nice addition to combat spam.