Spamassassin is running and tagging and doing everything it generally should be doing but the results are horrible. I've fed it over 60,000 spam messages to learn from but its only catching about 5-10% of the spam.

Is there any way to make sure that its really taking into consideration all of those spam messages I've fed it?

FYI, the training process went something like this:

/opt/zimbra/bin/zmlmtpinject -d /spam -r spam@ai.4ph.com -s spam@ai.4ph.com

zmtrainsa ai.4ph.com spam password spam Inbox

I ran the 60k messages in a couple batches and it took about 24 hours to run so I'd like to think it did *something*.

Would the dspam instructions on this forum help the accuracy significantly? Any advice would be greatly appreciated.

And just to clarify, SA is tagging the headers and running tests and it is identifying spam. Its just doing such a poor job thats its not really useful.

thanks,
altimage