Thank-you everyone. I shall have Bayes forget emails, then retrain with a rule that counts linkedin invitations as spam. I have one I wrote earlier and implemented after Bayes had gone off track.
Cheers, S. -- as silly as fun simon@klunky / .co.uk / .org / .net pgp 4BA78604 Benny Pedersen <m...@junc.eu> wrote: >Simon Loewenthal skrev den 2013-05-16 15:17: > >> * have sa-learn exclude references to linkedin? > >basicly you need to know how bayes works, it does not just use the word > >linkedin as writed, but more or less split it to each letter, so >linkedin and other content is splited into atoms that being stored into > >bayes database, it does not matter if that atoms is ham or spam yet > >but if you sa-learn --spam then the atoms is counted as spam, and if >you use --ham is counted as ham, --forget is when you like bayes to >autolearn again on no data from this emails with linkedin in :) > >thats the basic, what was the question now ? :(
