After whatever period of time was needed for the system to be populated with data, the Bayes filter has started appearing in the scoring calculus as BAYES_00 and others. This has been throwing quite large negative scores on messages that would otherwise have scored above thresholds that I am interested in having quarantined or blocked. Therefore I have a few questions.
1) How to disable Bayes from being included in the scoring totals? I see "Use Bayesian filter" under Configuration > Spam Detector > Options > Use Bayesian filter. Is this the correct location to just disable it all until I can figure out the following questions?
2) How do I manually, correctly flush the data from the Bayes system?
3) What is the process for manually populating the Bayes system with emails that I am already 100% confident are unwanted for various reasons: spam, phishing, 419, etc?
4) Is it possible for the Bayes scoring based on my manual population process to be used in the scoring without whatever automatic process has been populating the data for me? Basically, can I just periodically populate manually and prevent any automatic processes?
5) Is there a way to have multiple, separate Bayes scoring systems? The reason I ask is that would it not be better to have pools of emails "phishing", "marketing", "sales emails", 419, etc? And then have each separate Bayes system returning a separate score?
6) Can Bayes be configured to only provide positive numbers and drop negative numbers?
1) How to disable Bayes from being included in the scoring totals? I see "Use Bayesian filter" under Configuration > Spam Detector > Options > Use Bayesian filter. Is this the correct location to just disable it all until I can figure out the following questions?
2) How do I manually, correctly flush the data from the Bayes system?
3) What is the process for manually populating the Bayes system with emails that I am already 100% confident are unwanted for various reasons: spam, phishing, 419, etc?
4) Is it possible for the Bayes scoring based on my manual population process to be used in the scoring without whatever automatic process has been populating the data for me? Basically, can I just periodically populate manually and prevent any automatic processes?
5) Is there a way to have multiple, separate Bayes scoring systems? The reason I ask is that would it not be better to have pools of emails "phishing", "marketing", "sales emails", 419, etc? And then have each separate Bayes system returning a separate score?
6) Can Bayes be configured to only provide positive numbers and drop negative numbers?
Last edited:
