Join our Programming community?

Richie Hindle richie at entrian.com
Mon Jan 13 09:48:48 EST 2003


[Terry] 
> Spam Alert! Standard spam intro.  Translation: "Spam follows".
> Probability .9, as estimated by internal Baysian spam filter based on
> +-1000 spam messages and >10000 real messages.

My Spambayes classifier (trained on 800 spam and 600 ham) says:

Spam probability: 0.424847267756 [on a scale of 0-1]

Clues:

message-id:@python.org    0.00328707085464 
header:Organization:1     0.0204792575513 
header:X-Complaints-To:1  0.0240739915861 
community.                0.0652173913043 
programmer                0.0918367346939 
url:forums                0.0918367346939 
subject:Programming       0.155172413793 
[lots of boring stuff in the middle]
join                      0.782260460822 
today                     0.818211859201 
subject:Join              0.836925617583 
alot                      0.844827586207 (are alot of spammers illiterate? 8-)
emailing                  0.908163265306 
spam,                     0.908163265306
email                     0.912218244033 
subject:our               0.949438202247

So our classifier is more cautious than yours.  8-)  Mind you, ours doesn't
work on phrases yet.
 
-- 
Richie Hindle
richie at entrian.com




More information about the Python-list mailing list