Bryan.C.Breen.1 at gsfc.nasa.gov
Tue Jan 13 12:57:28 PST 2004
At 19:57 1/12/04 -0800, Bob Snyder wrote:
>Bryan Breen wrote:
>Quote of spam message snipped here..
>>Just as the chatter from the last one of these dies down... bang. That's
>>just to funny.
>Somebody correct me if I'm wrong but isn't it a bad idea to quote a spam
>message in a 'real' message? Won't this tend to screw up the training of
>people's bayesian filters?
I only quoted the first dozen or so words. The nasty meat that should kick
off your filters was all that gibberish beyond that.
Really short messages are not very helpful to a bayesian filter, there's
far to little info to key off of (to call it good or bad). That small
snippet I used will have such an insignificant effect (not zero, just
insignificant) on the cumulative "logic" of the filters that it will not
"foul" them if they were created with a decently sized learning pool.
Just looking at what SA reported; the original message came in ranked +1.2,
where my small quote generated a -6.1, and the message of yours that I'm
replying to (with no quote of the original) generated a -6.3. Minimal
difference between quoting a small part or not quoting it at all. Yet major
difference between them and the original.
| Any opinion expressed in |
| this message is my own |
| and not those of NASA. |
More information about the lfs-support