Anti-spam performance question

[Alycat]

When I moved to PocoMail I imported a lot of emails from my previous email client, including about 3,000 junk emails. I used these, plus about 4,000 good emails to do the initial training with Poco. I left the Poco spam filter options at default. My spam stat's started at about 55% and has steadily fallen to about 40%. All junk that is not caught is marked as junk, not just deleted. The junk filter runs on incoming emails at filter position #2, the first being if Avast has marked them as infected I move them to Junk. I have only had 3 or 4 good items marked as junk.

Should I expect better than this at this stage (a bit over two weeks), or is it a case of too early to tell?

Generally I have always run spam filtering on email clients at default and so far Poco has certainly been WAY below any others.

[Pete]

Some of us only train Pocomail's Bayesian filter when the BF flags a false negative or a false positive.

By running a huge group of emails through the BF as you did, you might have trained the BF on several emails that the BF would have caught without such training. By doing this, the BF thinks that it didn't catch these emails and counts them as false negatives and lowers the accuracy percentage.

At this point, one option is to reset the statistics in "Tools > Junkmail filtering". I suspect that your percentage will increase dramatically over the next few weeks.

[Alycat]

I reset my stat's a week ago, and they have consistently continued to decline - as an example of my anti-spam experience, yesterday I received 102 emails, 6 were re-directed to folders using filters, the remaining 96 all ended up in my In box, they all turned out to be spam.

I have renamed the two DB files and am just now marking bad emails as junk to see if that helps. We'll see what happens...