False Positives Are A Fact Of Life In Spam Filtering

from the type-I-and-type-II-errors... dept

Spam filtering companies are now saying that the "collateral damage" of falsely tagging legitimate email as spam is "inevitable", and it's just something we need to deal with. It's a simple trade-off. The fewer false positives you end up with, the more spam is going to get through as well. Still, it appears that some of the filtering systems are getting pretty good. My own combination of filters is at the point where, out of about 2500 spam emails per week, I'm finding maybe 2 to 3 that are false positives (though, there's a chance that some are just getting deleted), and about 5 to 10 actually slip through the filters and make it to my in-box. Those are pretty good percentages. Still, not everyone is setting up filters, and as someone in the article points out, the real solution needs to be a combination of many factors, from technology, education and government.


Reader Comments (rss)

(Flattened / Threaded)

  1.  
    identicon
    Richard, Feb 20th, 2004 @ 1:01pm

    Spambayes - no false positives

    I've been using spambayes.sf.net for 6 months now, and not a single false-positive. It doesn't take a "black or white" approach, but rather if a message falls inside an "unsure" range, then it's flagged as such. That way those messages, about one or two a week, may be filtered into a different folder - usually to be subsequently tagged as spam.

     

    reply to this | link to this | view in thread ]

  2.  
    identicon
    Anonymous Coward, Feb 20th, 2004 @ 2:03pm

    No Subject Given

    Sound like pretty bad percentages to me. Scale that up to an enterprise getting a couple hundred thousands spams a week, and that's a helluva lot more false positiives, each of which has to be dealt with by someone. Loss of productivity would be enormous.

     

    reply to this | link to this | view in thread ]


Add Your Comment

Have a Techdirt Account? Sign in now. Want one? Register here
Get Techdirt’s Daily Email
Save me a cookie
  • Note: A CRLF will be replaced by a break tag (<br>), all other allowable HTML will remain intact
  • Allowed HTML Tags: <b> <i> <a> <em> <br> <strong> <blockquote> <hr> <tt>
Follow Techdirt
A word from our sponsors...
Essential Reading
Techdirt Reading List
Techdirt Insider Chat
A word from our sponsors...
Recent Stories
A word from our sponsors...

Close

Email This