http://www.graviton.de/source/
* a score for a spam probability; for each badword a number (I've read this somewhere else)
Some scores/cases (not all) may be save to delete, or put in a 'sure to delete' bin.
Initially a badword has a standard zero value to begin with. Not all badwords need to be weighted; colour addressing needn't be necessary, and perhaps a spam_regex notification can be replaced by the actual found word. I have defined a regex:string with several |-signes. Please see http://www.regular-expressions.info or http://zez.org/article/articleprint/11 for more information. Other regex related tools: 'V-Grep', 'Agent Ransack', 'RegexBuilder' (2 3) or perhaps 'Regex Assistant' (regex_tips.htm; could be useful for that 'txt markup' idea..)
* a timer/countdown may be of use; if you're about to hit the 'empty trash' button, it could happen, with a fast internet connection, the programme just did another mailcheck on the server, possibly deleting legitimate emails. Ghosting the 'empty trash' button around that time may not be required/mandatory. viewtopic t=238
And a few more, (also;) for discussion,
* when email is suspicious/verdächtig:
Message contains the following unknown words (dictionary check (Aspell), not found in dictionary) and
Message contains the following links (web and mail addresses) viewtopic t=358
* And, if suspicious; Before spammers turn to "insinuations" and "stories", messages with a weird layout could perhaps be relfected by an equally weird character show, as summed up (indicated/displayed) in Popman's rules column. This requires a simple (as in non-complex, and very much lighter than Flash/Silverlight) opensource (txt?-) animation package. /As instructed by the structure of the received message. Or something like this. Fun! If kept simple and light perhaps. Perhaps a hint then; do we have spam backfiring?! Replace "Mouse" with "badRat" ;)
* Message originates from outside; [domain], [domain country] (as found in the header, x-from/return-path, could be a victim mail address, but it identifies a suspectible message anyway)
viewtopic t=380 (Tommy's request; is it a good idea to include ".. and contains none of Good_Domains"? Perhaps it's good to count the rejected domains in a log as well, and keep them organised.)
* Message originates form a trusted domain, namely [domain] (from domain whitelist?) (Suggested being displayed in short: "orig: [domain]") The argument being that good or reasonally good domains won't let themselves being abused easily. And, some domains can't be abused at all because of their internal (Apache? webserver) usage restrictions?
This could be mentioned in the unfiltered message list Rules column (which as yet remains empty?)
* I also think the Rules column should show legitimate (known, whitelisted) senders, or mention that the sender is whitelisted
+ Ahem, with the risk of indeed overloading the internet (as it is already):
autoreply with: please include one of the following keyword in your message: [keyword1, keyword2]
- victims (should) get notified they're being abused (so take action, ask your internet provider for help)
- spammers may address/target you more specifically using this information, but I think they're also more easily dealt with that way. Anyone for comments on that?
(messages containing those keywords can be processed separately)
((suspect messages could be auto-replied with fake keywords)) -> Please discuss, or dismiss
+ A web repository storing legitimately email addresses, voluntairy, and to get a 1 to 5 score with 1 being victim 2 being unknown and 5 being strong, trustworthy/respected (and I would join/register with my email address, probably eventually getting a 3 or so)
(and why not, when you're already victim of spamming? Those email addresses are ours period.) (And yes, I know spammers use victims email addresses; please discuss, or dismiss. Please use arguments, the computer does too;) A 'No unsolicited advertising/werbung'-tag may not be required.
* I don't know if I'm right, but I like the fact your programme opts not to make use of a leaning mode, as it may (the reason being) too processor/processing demanding
=-^-=
Happy Spam analyzing, congrats, we could learn a lot from this (so thank you one hell of a lot, vielen Dank;) I know there are many priving eyes on the net keen on visiting this site in particular, and right they are.