Hello wonderful tbudl, Have anyone in the group had any studies on how the bayesian spamfiltering priciple is affected by errantly spelled emails and "hi-jacking" the email subjectlines?
I have lately seen how my spamfilter is getting less efficient. I wounder whither the filteringmodel will be affected negatively by the increasing amount of "unique" spelling in certain mails. (Including this one :P ) Also I've seen some spamers use the lists most common subjectlines. (for example: #Bugreport 123456789 bla bla bla) as it infact may be a spammail. I have tried to train my bayesian mailfilter made by Achim Winkler to sort such mails out but not been successful - yet... I fear I have to start all over again train that @"#%3!Ϥ filter again... Is there a shortcut getting back on track? -Oh! I allmost forgot. I have blended subscriptions to various discusionlists in different languages in subfolders. How have you seen this affect the filtering efficiency in your cases. -- greetings from a warm and sunny holiday sweden, qe3ee mailto:[EMAIL PROTECTED] --------------------------------------------------------------- E fructu arbor cognoscitur 19:11:04 den 24 juni 2005 --------------------------------------------------------------- The Bat! v2.12.04 Windows XP.5.1.2600 if this mail has a virus please throw it away and notify me :)
________________________________________________ Current version is 3.5.25 | 'Using TBUDL' information: http://www.silverstones.com/thebat/TBUDLInfo.html
