Followers tweeting nonsense or just tweeting sentences that just don't quite fit with reality is exactly what I'm hoping to identify. It's easy enough to find a known spammer and block them, but my hope is to identify a spam account before they ever actually send any links. There are certainly some categories of spammers that this approach will not work on, although they are easy enough to detect using other means. I have observed a trend of bots that attempt to appear human and those are the ones that I hope to identify. I will have something completed by about two weeks from now so I'll post with how effective it is.
@johnsheehan I just followed you, I'll keep an eye on your followers for spammers. For others, I'm @jeff_tucker on twitter so please if you find an interesting spam account that is trying to "look human" let me know so I can grab their account history before twitter bans them, I'm primarily interested in their initial tweets prior to sending out links. I've noticed that if I start using keywords on trending topics then I get hits, although not nearly as many as I am hoping to get (imagine that, complaining about not getting enough spam). I'll keep trying and I think an algorithm that attempts to create the optimal post to attract the most spammers would be an interesting project as well, I may expand my research at some point in the future. Thanks everyone, appreciate the help. -Jeff On Apr 3, 9:55 pm, "M. Edward (Ed) Borasky" <zn...@borasky- research.net> wrote: > On Sun, 3 Apr 2011 21:34:54 -0700 (PDT), John Sheehan > > <johnshee...@gmail.com> wrote: > > You can use my account as an example. I'm currently getting between > > 50 > > and 150 follow spams per day for the last 3 weeks. Here's a graph > > that > > demonstrates the 'attack'http://screencast.com/t/xl7zcgdYI > > > If you have any other questions, I'm @johnsheehan and can be reached > > via email same user name at gmail. > > > John > > Interesting - most of the followers I've been getting are real humans, > not that they're all that interesting humans. ;-) There was a period > when I was getting a bunch of followers that were tweeting nonsense, > sometimes not even real sentences. Eventually one or two them started to > tweet links. Apparently the way they work is to build a network - if one > of them follows you and you don't block them, then they copy each other. > I suppose it's possible to collect data via the API and do graph theory > analysis on these accounts to isolate the clusters, but it hardly seems > worth the effort when there are so many accounts just spamming multiple > trending topics apparently with little interference from Twitter. > > -- > http://twitter.com/znmebhttp://borasky-research.net > > "A mathematician is a device for turning coffee into theorems." -- Paul > Erdős -- Twitter developer documentation and resources: http://dev.twitter.com/doc API updates via Twitter: http://twitter.com/twitterapi Issues/Enhancements Tracker: http://code.google.com/p/twitter-api/issues/list Change your membership to this group: http://groups.google.com/group/twitter-development-talk