Why would SpamSieve treat a USPS email as Spam? It does have my email address in the “From” address - uspostalservice_at_usps.com_sequitor@duck.com is that the reason?
“You trained this message as Good from your mail client.
Prediction: SpamSieve had predicted this message to be Spam, and you corrected the mistake. SpamSieve’s Bayesian classifier predicted this message to be Spam based on a statistical analysis of its content.
Score: 73 (0 is least spammy; 100 is most spammy)”
Should I change my Score value?
Thanks
Chris
Airmail 26.0.1, Sequoia 15.6.1
That’s a strange looking address, but it doesn’t seem to be the reason. What does the Words section of the log entry say?
Hi, Michael:
Summary: SpamSieve’s Bayesian classifier predicted this message to be Spam based on a statistical analysis of its content.
Score: 73 (0 is least spammy; 100 is most spammy)
Words: S:United(1.000), 9background(0.999), 9background(0.999), O:f5fafd(0.999), U:Eqww(0.999), U:Eqww(0.999), ^iw-55(0.999), ^iw-55(0.999), mobile®(0.001), mobile®(0.001), T:Threads(0.002), smalltext(0.002), U:B7g(0.002), U:B7g(0.002), U:G99(0.002)
Accuracy: Mistake (False Positive, i.e. SpamSieve classified as Spam, but you trained as Good)
Thanks, and Happy New Year
Chris
Thanks. It looks like this message contained multiple words (and other characteristics) that had previously appeared in messages that you trained as spam but not in ones trained as good. Perhaps this is just bad luck. However, it might also be worth reviewing the spam messages shown in the Corpus window to make sure they are all actually spam.
Will do, thanks. And Happy New Year!
A post was split to a new topic: Good Message From Forum in Junk