|
Re: false positives...: msg#00016mail.spam.razor.user
Agreed, revocation is the way of dealing with bad submissions. In this case, it is likely that someone has a spamtrap address that auto-reports everything that comes in. Someone else may have maliciously subscribed that spamtrap to this mailing list to get it reported, or it may have been done auto-magicaly by a worm or virus (there have been reports by some of the SpamAssassin developers that some virii are submitting email addresses to list subscription addresses and that they've been recently finding some of their spamtrap addresses subscribed to lists.) In any event, this particular message when I ran it through razor-check -d only had a cf of 14, so it's not all that well trusted as being spam. If a handful of people accurately revoke it, it should drop off the list. The relevant debug output follows: Feb 11 14:53:40.889525 check[29515]: [ 2] Razor-Agents v2.22 starting razor-check d <snip> Feb 11 14:53:41.232537 check[29515]: [ 4] truth.cloudmark.com << 96 Feb 11 14:53:41.232641 check[29515]: [ 6] -a=c&e=2&s=j70-rFNVBuyVQjpLKEp3HLljSzQA a=c&e=4&ep4=7542-10&s=xn62Xy-z3j16_VK0VOrshdb1jo8A Feb 11 14:53:41.399614 check[29515]: [ 4] truth.cloudmark.com >> 20 Feb 11 14:53:41.399746 check[29515]: [ 6] response to sent.1 -p=0 cf=14&p=1 Feb 11 14:53:41.400293 check[29515]: [ 6] mail 1.0 e=2 sig=j70-rFNVBuyVQjpLKEp3HLljSzQA: sig not found. Feb 11 14:53:41.400734 check[29515]: [ 3] mail 1.0 e=4 sig=xn62Xy-z3j16_VK0VOrshdb1jo8A: Is spam: cf 14 >= min_cf 1 Also note that since e2 does not match, but e4 does, it shows that this *exact* message is not listed in the razor database, but instead a very similar with some significant subset of the text being identical. In general I find that a lot of the low-cf matches in razor are garbage hits. Although this run uses default settings, my production run uses a min_cf of 11 to try to quiet some of the bad hits down. At 01:21 PM 2/11/2003 -0500, Ed Hennis wrote: On 11 Feb 2003, 5:34pm (-0000), Dirk Koopman wrote: ------------------------------------------------------- This SF.NET email is sponsored by: SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See! http://www.vasoftware.com |
|
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| Previous by Date: | Re: false positives...: 00016, Ed Hennis |
|---|---|
| Next by Date: | Too many false positives...goodbye, razor: 00016, John J. Stimson-III |
| Previous by Thread: | Re: false positives...i: 00016, Ed Hennis |
| Next by Thread: | Re: false positives...: 00016, Fox |
| Indexes: | [Date] [Thread] [Top] [All Lists] |
| News | FAQ | advertise |