Ross Vandegrift said:
> Obviously an interim solution is to whitelist, but long term is probably
> harder. What kind of SCE, like MSDN newsletters, product updates, etc
> is in the corpus for the GA? Maybe seeding the corpus with a bigger set
> of these type of mails will have some interesting/useful results.
Yes, this is my position. That way, patterns which overmatch SCE (and
thereby cause FPs) will be penalized heavily.
As a result I'm now collecting a corpus of nonspam SCEs, using a few
common FP'ing SCE sources. BTW if anyone knows of more FP'ing SCE
sources, I'd appreciate if they posted them; the more the better.
I'm currently getting C|Net, ZDNet, LockerGnome, Cramsession and The
Guardian (uk newspaper). MSDN I tried, but Passport.com doesn't like
Mozilla it seems ;)
--j.
--
'Justin Mason' => { url => 'http://jmason.org/', blog => 'http://taint.org/' }
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
