« LinkedIn Groups |
Main
| Dogging: Smart Mobs Go Carnal »
March 18, 2004
Can social networks stop spam?
Posted by Clay Shirky
Interesting articcle,
Can Social Networking Stop Spam?, about the work of UCLA researchers on social clustering as a spam detector, using the latent social network as a filter:
"When you get an e-mail from Alice with a 'cc' to Bob, you put a link between Alice and Bob," Boykin explained.
Examining six weeks worth of e-mails from Bobs, Carols, Alices and others, Boykin and Roychowdhury were able to identify the "components" of their burgeoning e-mail network.
"A component is a set of nodes which can all reach each other in the network," Boykin said. "It turns out that spam components and non-spam components are easy to distinguish" in a large enough network by examining so-called "clustering coefficients."
"In social networks, if A knows B, and B knows C, A often knows C also," Boykin explained. "Clustering coefficients measure this relationship."
In comparison to random networks, Boykin said he and his co-worker discovered that "non-spam components have high clustering coefficients, and spam components have clustering coefficients equal to zero."
the catch seems to be a large, readable population of email users -- the article doesn't make it clear how many people need to be involved to get to the claimed accuracy.
Comments (3)
+ TrackBacks (0) | Category: social software
- RELATED ENTRIES
- › Spolsky on Blog Comments: Scale matters
- › "The internet's output is data, but its product is freedom"
- › Andrew Keen: Rescuing 'Luddite' from the Luddites
- › knowledge access as a public good
- › viewing American class divisions through Facebook and MySpace
- › Gorman, redux: The Siren Song of the Internet
- › Mis-understanding Fred Wilson's 'Age and Entrepreneurship' argument
- › The Future Belongs to Those Who Take The Present For Granted: A return to Fred Wilson's "age question"
1. Abe on March 18, 2004 8:23 PM writes...
All that really needs to be done is somehow graft the data cached in Friendster, Tribe and the like into an email program. Its pretty unlikely you'll ever get spam from someone within 2 or 3 degrees of you. And if the links are visible then its even less likely as instant peer pressure gets applied to any spammer. Set up your email program so that email from within your network goes to the inbox and outside goes to suspect box and you have pretty good filter system to add to the current mix.
the problem of course is all the data is cached in private companies...
Permalink to Comment2. Julian Bond on March 19, 2004 3:11 AM writes...
Do I need to remind everyone that most virus emails these days have spoofed from addresses and sent to: addresses taken from people's address books. Infected machines are then being used as spam sources. So spam/viruses are already taking advantage of the inherent socila networking contained in people's address books. In fact it's been suggested that if you studied Virus email flow, you could build a FOAF network from the traffic because the from and to are typically FOAF with 3 degree separation.
Then there's the problem with hard white and black lists that one of the huge strengths of email is that you can contact new people for the first time with no prior engagement.
So the short answer is that in general I think this anti-spam technique is deeply flawed. just like all the other techniques.
Permalink to Comment3. Abe on March 19, 2004 4:34 PM writes...
Julian, yep your right. As long as email can be spoofed, social networks won't help that much. But in any architecture where the sender can be verified its a whole other story. Friendster for one, with all its problems, so far seems spam proof.
Permalink to Comment