Question re: training & Bayesian classifier

I have a corpus that was started on SpamSieve v2, and I notice that there are a lot of spam messages in “good” that I can’t do anything about - so I’m starting over, and I want to set this up the most efficient way possible. I’m reading this in the app’s help:

Normally, the [Train SpamSieve allowlist] settings are enabled. This means that training a single message as good will make SpamSieve classify all future messages from that message’s sender as good. Likewise, training a message as spam will make SpamSieve classify all future messages from that message’s sender as spam. Normally, the [Train Bayesian classifier] preference is also enabled, which will let SpamSieve learn from the message’s contents so that it can recognize future messages that are similar even if the sender is different.

So let’s say I have a contact form that always comes from I want the CONTENTS to be scanned, with good messages routed to me, and bad messages routed to spam.

Am I seeing correctly that if I train it as good that’s going to whitelist my entire contact form, but if I train it as spam it will always route it to spam?

How do I tell SpamSieve that, for a given sender, I ONLY want the Bayesian classifier running? Do I have to completely disable the “train allowlist” features?

For messages trained with SpamSieve 3, you can fix any incorrect trainings from the Corpus window or Log window. For messages trained with SpamSieve 2, if you still have the message in your mail client, you can (re)train it there to update its category in SpamSieve. If you no longer have access to a message trained with SpamSieve 2, I’m not sure how you would be seeing in the app that it’s in the wrong category, but you would need to reset the corpus and re-train to change it.

Yes, but only if you only train one message.

No. This will happen automatically if you train some messages from that address as good and others as spam. SpamSieve will disable the corresponding blocklist and allowlist rules for those messages so that they don’t match any messages. You can also manually disable rules in the Blocklist and Allowlist rules to achieve the same effect.

1 Like