@semanticvoid pointed out that one of the problems with using @spam to collect spam for research is that some classes of spammers may be underreported. What is there was an account that just copy tweets from famous accounts or a feed and inserted malware links once a day?
Any ideas on how to improve my data/experiment?