Blacklist problem, as well as other filters

[Hoov]

Occasionally one of the filters fails to mark an e-mail as spam and for deletion. This has happened with the blacklist as well.

The status shows blacklist, or the name of the filter, but there is no effect.

Any idea's on the problem?

[Sidewinder]

Occasionally one of the filters fails to mark an e-mail as spam and for deletion. This has happened with the blacklist as well.

The status shows blacklist, or the name of the filter, but there is no effect.

Any idea's on the problem?

I suggest that you right click on the message header in the Inbox and select Show evaluation details. That will cause the side panel to pop out and reveal each of the tools numerical values assigned to the message. The mathematical sum of all of the tools determines the final rating. If you post the numbers for each I may be able to help further with your settings.

[Hoov]

Thanks for the offer.

The learning Filter has given it a score of 61, and then blacklist filter has given it a score of -50 . All the rest are blank. I would have thought that a blacklist entry would have overridden then learning filter.

[Sidewinder]

Thanks for the offer.

The learning Filter has given it a score of 61, and then blacklist filter has given it a score of -50 . All the rest are blank. I would have thought that a blacklist entry would have overridden then learning filter.

You can balance the evaluation process by tuning the values assigned to each of the tools except the Bayesian evaluation. Go to Settings, Spam Tools, Spam Ratings and move the sliders to influence the outcome. I have my friends set to +200, Blacklist to -200, Origin of Spam -150 and First Alert -150.

I suggest that you wait until you get your Bayesian training further along but once you have the training fairly complete I would also change Settings, Spam Tools,
Learning Settings, Bayesian Sensitivity to Low.

[stan_qaz]

No, they all work together to come up with a total score to use for classifying the message.

The problem with trusting the blacklist too much is that it is that spammers are getting good at tricking folks into blacklisting good addresses, if you are sure that won't happen to you you could crank the blacklist up to -200 with no problem.

Personally I don't blacklist individual addresses only wildcarded major domains like *@*.ru and keep the points down fairly low so if I do get a good message from someone there it will be rescued from deletion by one of the "good" tools.

[Sidewinder]

I would have thought that a blacklist entry would have overridden then learning filter.

The spammers these days have become extremely sophisticated in the structuring of their messages and they tend to make a single tool far less reliable as compared to the new methods in MWP2010. The use of the blacklist has diminished in value since the spammers tend to use a once and done address and you wind up filling up your list with addresses that may never show up again.

The Bayesian evaluation, Friends List, Blacklist, Origin of Spam and First Alert are not designated as filters. There is a separate tool included to allow you to design and use Filters to improve the outcome as well as assisting you in sorting and classifying messages.

[Hoov]

Thanks for the pointers Sidewinder and stan_qaz. Hopefully that will fix up the issues I have been seeing.