I'm using 3.4, and have the junk mail filters pretty much standard(non-standard bayesian checked), and I go through the process of declassifying mail that's marked incorrectly, then marking it with it's correct distinction(i.e.~junk mail/good mail), but it doesn't seem to be learning very well.
I get topic reply notifications from one of my sites, and no matter how many times I unmark them as junk and mark them as good, the next one to come into the box gets marked as junk, and they're identical to the last one. This has probably happened 50 times now.
I've searched the forum and found threads with instructions like " declassify 15 times, then classify 15 times...." but this makes absolutely no sense to me.
Is there a simple way to help the filter system learn better? Right now, it's useless, as I'm still manually marking junk mail and unmarking good mail more often than not, but I don't want to install a third party app, because they seem to wreak havoc on PM, creating blank emails, etc...
thanks for any light you can shine on this

thanks,
json
Edit Eric: Moved to Junk Mail Filtering