[Spambayes] Guidance re pickles versus DB for Outlook

Jeremy Hylton jeremy@alum.mit.edu
Tue Nov 26 17:10:42 2002


>>>>> "TS" == Tim Stone <- Four Stones Expressions <tim@fourstonesExpressions.com>> writes:

  TS> Update probabilities is no longer necessary, as WordInfo now
  TS> calculates probabilities on demand.  Simply remove the call... 

The problem I had was with the learn() method, which is documented to
take three arguments but actually only takes two.  It would be really
helpful if there were some developer docs or at least consistent
docstrings.

I changed learn() to only take two arguments, modified by WordIndo
class, and removed the update_probabilities() call and eliminated
tracebacks.

But I also destroyed the spam filtering process.  Every message I've
received since making those changes scored a 1.000.  For example,
here's the scoring detail for your message:

[59102 refs]
Score: 0.99999745451

Clues
-----
*H* 5.07285667561e-06
*S* 0.999999981876
hylton 0.0223511806365
jeremy 0.0226462499923
>jeremy 0.06749672346
skip:" 50 0.155172413793
haven't 0.228515259315
subject:for 0.267754749015
header:Received:5 0.271990252442
header:In-Reply-To:1 0.30024968789
sorry 0.30024968789
header:Errors-To:1 0.30781519964
got 0.310423809387
update 0.333024118738
content-type:text/plain 0.387298828087
simply 0.604185271179
working 0.617911285794
how 0.630464812873
longer 0.638534738834
takes 0.646878824969
exactly 0.6540436457
skip:m 10 0.689618426887
url:org 0.691518467852
header:Reply-To:1 0.711509992491
skip:s 20 0.716962524655
read 0.73131168503
every 0.745634117591
skip:p 10 0.756747597975
skip:w 20 0.763982841115
now 0.809312390184
mailing 0.810022433084
arguments 0.844827586207
changes, 0.844827586207
default 0.844827586207
documented 0.844827586207
skip:> 40 0.844827586207
this? 0.844827586207
url:python 0.844827586207
email addr:python.org 0.908163265306
job. 0.908163265306
remove 0.921389083674
quickly 0.934782608696
restart 0.934782608696
skip:u 20 0.934782608696
list 0.941063400602
calls 0.949438202247
skip:s 30 0.949438202247
subject:: [ 0.949438202247
url:mail 0.949438202247
proposed 0.95871559633
url:mailman 0.95871559633
url:listinfo 0.96511627907
am, 0.973372781065
messages 0.973372781065
suggest 0.973372781065
skip:_ 40 0.97619047619
there's 0.97619047619
here's 0.978468899522
subject:]  0.983271375465
tried 0.984429065744
nothing 0.991493383743

Jeremy




More information about the Spambayes mailing list