Re: Reporting SPAM Filter Issues

Tech-Archive recommends: Fix windows errors by optimizing your registry

"Diane Poremsky [MVP]" <outlookmvp@xxxxxxx> wrote in message news:%23t74B%23AEGHA.2036@xxxxxxxxxxxxxxxxxxxxxxx
They probably use feedback from user Hotmail users hitting the Spam button. Don't expect them to bother with feedback on how to improve their Bayesian filter. Find better anti-spam solutions. Spampal is free (www.spampal.org).

No probably about it... they have a select group of people involved in spam reporting and it gives them all the spam they have time to review. If it doesn't meet your needs, try a filter you can train - there are several other free decent ones besides spampal.

The Bayesian filter in SpamPal is far more trainable than the one in OL2003 (which comes with a preset list which does NOT represent a particular users experience with spam but is a generic list from many users). You can add words. You can change weighting. You can configure how new words are weighted. You can configure a noise level after which the expired words are removed. You can easily change a word from good to bad or visa versa. You can import a list of good mails to pre-train the database as to which mails are good, and the same for import a set of bad mails to pre-train for spam. Does OL2003 provide pre-training import? SpamBayes is another Bayesian filter but I haven't used it to know how configurable it is. Bayesian should also NOT be the only method by which you rank the spamminess of mails.

Basically the OP said that Hotmail's spam filtering was unsatisfactory. I only gave *one* example of a far better alternative to eliminate spam. There are several. I prefer one that has multiple schemes of detecting spam. The user needs to trial several solutions to see what is best for them. However, none provide the magic bullet against spam (and neither does the challenge-response, especially due to its "challenge spam" spewed at innocents and because of the ability to turn a C-R user into a spammer).

.

Quantcast