On 2017-06-06 20:44, Anthony Youngman wrote:
On 06/06/17 18:20, Carlos E. R. wrote:
On 2017-06-06 19:18, Wols Lists wrote:
On 06/06/17 14:53, Carlos E. R. wrote:
And actually, I would like to classify some spam in TB based on the
*contents* of the email. It seems like you can't tell TB *how* to recognise spam, and you can't filter based on the *content* of the mail :-( Yes, you can. Thunderbird learns when you tag something as spam or not.
Thunderbird *learns*.
Which means, in other words, that I cannot *tell* it.
"Dear David" is spam. Why can I not tell thunderbird to put it in the spam bin!!!
The act of marking a post as spam with the spam flag in Thunderbird tells the program to learn about that post. So yes, you tell it.
So how does it know that the string "Dear David" means it is spam? The point is, if *I KNOW* that something is a clear marker of spam, there is no way I can *TELL* thunderbird about it!
All I can do is tell thunderbird "this is spam". There is *no way* that I can tell thunderbird *WHY* it is spam.
Ok, you can't, but it is the way most antispam filters work. I think it is called a bayessian filter, and after some training it becomes very effective. If you tag five emails and the five contain "Dear David", it will learn to consider that text as a clue. If you want to tag on a specific text in email you need to use a scoring filter such as spamassassin and add your own rules. -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)