[spambayes-bugs] [ spambayes-Bugs-1036135 ] detect faxes as spam
SourceForge.net
noreply at sourceforge.net
Tue Oct 5 16:12:08 CEST 2004
Bugs item #1036135, was opened at 2004-09-28 08:32
Message generated for change (Comment added) made by kpitt
You can respond by visiting:
https://sourceforge.net/tracker/?func=detail&atid=498103&aid=1036135&group_id=61702
Category: Outlook
Group: Binary 1.0rc2
>Status: Closed
Resolution: None
Priority: 5
Submitted By: zimcad (zimcad)
Assigned to: Nobody/Anonymous (nobody)
Summary: detect faxes as spam
Initial Comment:
the outlook plugin detects incoming faxes in outlook as
suspected spam.
Combined Score: 29% (0.288208)
Internal ham score (*H*): 0.85227
Internal spam score (*S*): 0.428687
# ham trained on: 327
# spam trained on: 573
3 Significant Tokens
token spamprob
#ham #spam
'from:none' 0.0180723
12 0
'x-mailer:none' 0.665707
120 419
'to:none' 0.721932
12 55
Message Stream
Seiten: 1
Seiten: 1
Die Mitteilung hat folgenden Anhang:
All Message Tokens
8 unique tokens
'cc:none'
'content-type:text/plain'
'from:none'
'message-id:invalid'
'reply-to:none'
'sender:none'
'to:none'
'x-mailer:none'
----------------------------------------------------------------------
>Comment By: Kenny Pitt (kpitt)
Date: 2004-10-05 10:12
Message:
Logged In: YES
user_id=859086
Because these messages have so few tokens, it probably will
take a little more training on these than it might for
normal e-mail messages.
Also, you may be able to adjust the Possible Spam cutoff on
the Filtering tab of SpamBayes Manager to help out a little.
It will depend on how your other unsure messages are
scoring, but there's a good chance you could raise the
cutoff from 15 to 20 without increasing your false negatives
(spam in your Inbox).
----------------------------------------------------------------------
Comment By: zimcad (zimcad)
Date: 2004-10-05 03:32
Message:
Logged In: YES
user_id=1055336
sure, it only takes a time to train spambayes (i was too
impatient) after i had 5 faxes in the inbox the score is at
24% - so i think we can erase my message
----------------------------------------------------------------------
Comment By: Kenny Pitt (kpitt)
Date: 2004-09-28 17:01
Message:
Logged In: YES
user_id=859086
The following tokens look like they would probably be pretty
consistent for incoming faxes, and the fact that none of
them appear in the "significant tokens" list seems to
indicate that you haven't trained any incoming fax messages.
'cc:none'
'content-type:text/plain'
'message-id:invalid'
'reply-to:none'
'sender:none'
29% isn't that high a score, and I don't think it would take
much training to drive that into the good message range. I
don't have the ability to receive faxes in my Inbox, so I've
never tried training one. Is SpamBayes preventing you from
training on these?
----------------------------------------------------------------------
You can respond by visiting:
https://sourceforge.net/tracker/?func=detail&atid=498103&aid=1036135&group_id=61702
More information about the Spambayes-bugs
mailing list