All email in last check marked legit

[cheriem]

Every single email - and there were over 100 pieces - gathered in my last check (done manually) was considered legit and highlighted in green. The problem is that I didn't mark any of these senders as legitimate using the Anti-junk tool and now my "show legit only" view is rather useless as the truly legit emails are lost in a sea of false "legit" emails.

Not sure if this is a clue, but when I did the mail check, I was in "legit only" email view.

Any idea how I can correct this situation? Some of these false legit emails don't rise to the level of "legit" - a status I customarily reserve for work and family emails - but they don't sink to the level of "junk".

My version is v4.5.2.0


Thank you.

[spc3rd]

Hello cheriem and welcome to the Esumsoft Forums!

To get started, please change your Antijunk message View to "SHOW ALL", then perform another manual mail check and post back with the results.

Also, you can right-click on any message > select Antijunk > Show evaluation. This will provide information as to why a message was evaluated the way it was.

Other members of the The Esumsoft Team will be reviewing the issue you are haviing and provide additional feedback.

In the interim, as you are our newest member, we request that you please review the following Sticky topic:

Information for new users and forum members

The article contains important, helpful information which all new members should be aware of. Feel free to ,let us know if you have any questions or comments.

Thank you & welcome aboard!

[cheriem]

Thank you for your response. I switched my junk view to "Show All" and was again met with a sea of green in my email check. Out of the 100 or so different senders in this batch - most are newsletter or promotional emails, such as FTD Flowers, Groupon etc, I would estimate at least 80-90, which had previously been categorized in the "unknown" category are now categorized as legit.

Attached is a screenshot of a single sender. It shows the typical category identification, which is unknown, with an occasional junk identification, to suddenly in the last two days since this issue began, suddenly turning to green.

I did a sample message evaluation in each category for this single sender, shown in the screenshot, and the info appears below:

GREEN

Score: -1

Bayesian: score: -1 0% (conf: 609 / 875 = 69.60%)

------------------

UNKNOWN

Score: 0

Bayesian: score: 0 Low confidence


---------------

JUNK

Score: 1

Rules: score: 1
Watches: (1)

Bayesian: score: 0 Low confidence



------------------

I appreciate your time and assistance in resolving this issue:)

[Jeff]

This is the normal and expected behavior. Basically, you've trained enough messages (ie. marking messages as Junk/Legit) that the Bayesian system now has enough data to start evaluating messages on its own. This doesn't mean you need to stop training, though.

If you were to re-evaluate one of your older (blue/unknown) messages, it would probably assign an evaluation to that, too. ie. right-click on the message and select AntiJunk -> Re-evaluate message.

Based on what you're saying, it sounds like you would prefer that the Bayesian not be given any weight at all. So here's what you can do:
1) From POP Peeper's main menu, select Tools / AntiJunk -> Settings
2) Enable "Advanced settings" on the bottom-left of the right window
3) Select the "Bayesian" page on the left
4) Set the 3 values below the "weight" column to 0 (see screenshot below)
5) Press OK

To test that this worked as expected, right-click on one of your green messages and select AntiJunk -> Re-evaluate message.
If it worked as expected (blue/unknown), you can select multiple messages and then right-click one of the selected messages and AntiJunk -> Re-evaluate.

This will allow Bayesian to continue learning your messages as you train them in case you want to try out Bayesian again in the future. Another approach may be to leave the "Junk" weight at 1, so that Junk messages are still evaluated even if you have newsletters or such that you'd prefer to remain "unknown."

Also, there may be other approaches that are better for your expectations. For example, if you only want to use your white/black lists as the sole condition for messages appearing as legit/unknown/junk, then that is possible, too, with some tweaks.

Set Bayesian to 0 weight

[cheriem]

Thank you. Following your suggestions worked and has allowed me to halt the sea of green:)