Just like probably anyone else I'm fighting with the spam disease :(
Even though the filters in place (Outlook's own, Mailenable server based) are doing a reasonable good job, sometimes a false positive occurs. Now that wouldn't be a big issue if that spam folder was only a few messages per day. As matter of fact though it's growing almost daily and I now get in excess of 4000 spam messages every single day, thus it became practically impossible to identify false positives by sifting through the list.
So I looked into the Outlook Views to provide me with some help there and I found out that you can filter views using special SQL / DASL filter described at Andrew Delin's blog.
I've now created a filtered view of the Junk folder that only displays items with certain keywords in subject and text. This reduces the number of messages displayed to a about 100, which is now a feasible number to sift through.
To use SQL to filter open "Customize View" and then press filter. Note the SQL tab and open it. Click the "Edit these ..." checkbox on the bottom and the SQL filter box becomes active. Now you can enter simple conditions like
"urn:schemas:httpmail:subject" LIKE '%nexusdb%' or "urn:schemas:httpmail:subject" LIKE '%ADO%' or "urn:schemas:httpmail:subject" LIKE '%ODBC%' or "urn:schemas:httpmail:subject" LIKE '%MANAGER%' or "urn:schemas:httpmail:textdescription" LIKE '%nexusdb%' or "urn:schemas:httpmail:textdescription" LIKE '%ADO%' or "urn:schemas:httpmail:textdescription" LIKE '%ODBC%' or "urn:schemas:httpmail:textdescription" LIKE '%Manager%'
(The string comparisons above are case insensitive). For a full list of possible query fields check out urn:content-classes:message on MSDN.
Sure it's not avoiding any false positives not displayed, but it makes an impossible task manageable, and I think it's better to catch a few positives than missing all due to frustration when trying to find them in thousands of messages.
Long time ago, searching for
Long time ago, searching for an antispam solution, I went with SpamBayes, mainly due to user reports that they could get it to train with minimum or no false positives.
After some short initial training, I started tracking the Junk folder, and there really was no false positive for a few months, that's when I stopped tracking Junk folder completely.
With bayes filtering results vary by the specifics of the training they get /they are not uniform in all cases. Anyway I'm glad I went for SpamBayes, with all the others I was getting false positives. That was few years ago though.