Quote:
|
Originally Posted by cookie_365
Problem is, no computer algorithm can ever be as good at deciding what's spam & whats not. And how many false positives is acceptable? 1%? 5%? 0%? It all depends on the person, the value of the content of the messages, the chances the content is similar to spam messages blah blah blah.
If you think about how long it takes for your own Outlook/Mailwasher/whatever rules to sort out what's (probably) spam & what isn't, then its probably only a few seconds a day for most people. <Snip>
|
I have used Spampal for several months now, I have about 6 addresses in the whitelist, the rest of the filtering is done on the default rules, blacklists and the regexp filter. In all of that time I have only had one e-mail accidentally classified as spam (an e-mail newletter from Game) so that got whitelisted and only around 5 spam messages that were not recognised as spam. Addresses that I send mail to are automatically added to the whitelist so I don`t have to worry about those.
On the basis of your percentages, my spam has ben classified correctly with as good as 100% accuracy with probably about 10 minutes work on my part at the very beginning.
If Spampal can do it this well then there is no reason (theoretically of course) that NTL couldn`t do it.