[1/4] from: robert::muench::robertmuench::de at: 23-Sep-2003 16:42
On Sun, 21 Sep 2003 12:43:26 +0200, Dide <[didec--tiscali--fr]> wrote:
> I made a Rebol script to delete spam mails directly on the server > without loading them. > > http://www.agora-dev.org/forums/view.php?bn=rebol_prjnvxprod&key=1061826280 > > (This tool will be improve soon (I hope) with new features). > > I'm investigating the way to automaticaly select spams in the list.
Hi Dide, thanks a lot for this tool. Very handy! I'm now using it to filter out those damnd 150KB messages about the MS Update. WRT to automatically select spam mails, I want to suggest a very simple but most effective approach: Let people mark messages as spam and collect a MD5 checksum on a central server. Than your tool can perform a check against the server to see if others already reported the mail as spam. You can get more information about such a concept from http://www.cloudmark.com IMO people are the best spam recognizer, much better than any algorithm. Of course I would like to see a Rebol based version ;-). If the filtering engine would work standalone and transparent on the server, where the mail-server is running as well, this would be perfect :-)) -- Robert M. Münch Management & IT Freelancer Mobile: +49 (177) 245 2802 http://www.robertmuench.de
[2/4] from: jvargas:whywire at: 23-Sep-2003 12:44
Hi Robert, MD5 checksums will not work for slightly mutating content. We need some sort of fuzzy signature to classify the spam. Also there should be a very fast an efficient protocol to consult the central database for this "signatures". IMHO I think it would be best to have the centralized server use some bayesian methods trained by a collective set of users. Also I think the best place to stop the spam is at the receiving smtp server, and it shouldn't stop at just classifying and stopping; it should also try to waste the spammer CPU and BW resources, i.e.. like responding very slow if the server is an black list or rejecting the spam before delivering it to POP. So what it would be great is to create rebol smtp-proxy to do the job in coordination with the central server and training by the local admin. My two cents, Jaime On Tuesday, September 23, 2003, at 10:42 AM, Robert M. Münch wrote:
> On Sun, 21 Sep 2003 12:43:26 +0200, Dide <[didec--tiscali--fr]> wrote: >> I made a Rebol script to delete spam mails directly on the server
<<quoted lines omitted: 29>>> To unsubscribe from this list, just send an email to > [rebol-request--rebol--com] with unsubscribe as the subject.
Cheers, Jaime -- The best way to predict the future is to invent it -- Steve Jobs
[3/4] from: g:santilli:tiscalinet:it at: 24-Sep-2003 10:07
Hi Jaime, On Tuesday, September 23, 2003, 6:44:51 PM, you wrote: JV> Also I think the JV> best place to stop the spam is at the receiving smtp server, and it Indeed, however on a busy mail server this is simply too much work. The client usually has much more spare power available, and that's why most big SMTP servers don't do spam filtering. (You also have to consider that a user could want to receive the emails you are filtering; it's always better to let the user decide.) Regards, Gabriele. -- Gabriele Santilli <[g--santilli--tiscalinet--it]> -- REBOL Programmer Amiga Group Italia sez. L'Aquila --- SOON: http://www.rebol.it/
[4/4] from: robert:muench:robertmuench at: 27-Sep-2003 17:34
On Tue, 23 Sep 2003 16:42:07 +0200, Robert M. Münch <[robert--muench--robertmuench--de]> wrote: Hi Dide, I mad a quick patch to the program to fix a bug that happens if the msg/from is 'none that can happens sometimes. ; populate colomn blocks insert/only b-msg reduce [ any [msg/subject " - no subject -"] either none? msg/from ["none"][first msg/from] any [all [msg/date to-string msg/date/date] ""] any [all [msg/date to-string msg/date/time] ""] size mesg-num msg ] Hope this helps.... -- Robert M. Münch Management & IT Freelancer Mobile: +49 (177) 245 2802 http://www.robertmuench.de
- Quoted lines have been omitted from some messages.
View the message alone to see the lines that have been omitted