New Study uses Timestamps, Patterns for Antispam

Dennis Faas's picture

Researchers at Northwestern University and New York's Yahoo! Research have shown that they can catch spammers by the timestamps of their emails, paving the way for smarter advertisements, better spam filters, and more convenient social networking. (Source:

The new technique can differentiate people by using only the timestamps in their Sent folders with a snapshot of what they're doing. You can get meaningful information by knowing what time people send their emails.

80 to 90 Percent of the World's Email is Spam

Yahoo! is interested in finding a better way to catch spammers. Between 80 and 90 percent of the world's email is spam, which uses bandwidth, storage space and time. Spam may end up costing $42 billion in the United States and $130 billion worldwide. (Source:

Spammers are constantly changing their domains and IP addresses. Spammers are limited, however, because they need a botnet to send their millions of emails.

Technique Tested on American and European Students

The new technique was tested by focusing on how frequently two groups of college students were sending emails and when the email sessions began and ended. One group consisted of European students from a few years ago at a time when Internet access at home there was rare. The study also focused in on American students, although Internet access for them has been fairly common for some time.

The groups fell into two categories: 'day laborers' sending the bulk of their emails during the working day and 'emailoholics' who sent emails all day long.

Researchers found that email behavior was consistent for each person. Fewer than 20 percent of American students deviated from their email category over a two year period. That stability may allow an email service to recognize when a user's account has been commandeered by spambots, theoretically alerting users or freezing accounts.

Patterns Could Prove Useful

Detailed descriptions of activity patterns could prove useful for sites with heavy traffic patterns, such as Twitter, who could optimize how their servers allocate resources.  The new idea could also dramatically improve things for Internet services like Aardvark that depend on real-time interactions. (Source:

More information can be found in the "Characterizing Individual Communication Patterns" report by R. Dean Malmgren, Jake M. Hofman, Luis A.N. Amaral, and Duncan J. Watts.

Visit Bill's Links and More for more great tips, just like this one!

Rate this article: 
No votes yet