SpamSieve Update

I’m almost a week into my evaluation of SpamSieve. I’ve been relying on it as my only spam filter for the last 4 days. It’s been going quite well, and accuracy is excellent for such a short time training the filter (and a limited initial corpus):

Filtered Mail
164 Good Messages
2389 Spam Messages (94%)
566 Spam Messages Per Day

SpamSieve Accuracy
1 False Positives
56 False Negatives (98%)
97.8% Correct

Corpus
1652 Good Messages
2073 Spam Messages (56%)
110834 Total Words

I’d say that’s pretty good. In fact most of the false negatives were on the first day. I haven’t had a single false identification in the last two days. I had a problem getting the Applescripts working properly in Apple’s Mail, but the developer (Michael Tsai) replied to my email instantly and we worked through the problem in just a few email exchanges. Doesn’t get much better from a support perspective, and I haven’t even purchased it yet! Performance wise, I’m quite pleased. I’ve had the program running for a few days straight now and it’s using 29MB of memory. It’s been on the CPU for 1m43s, which is basically nothing. I’ve also plugged it into Entourage 2004 on my PowerBook at work, and it’s been working flawlessly there too.

I’ll probably give it another week before I pony up the $25, but at this point I’d say run, don’t walk to the download page and give SpamSieve a try. It blows the socks off the junk mail filters supplied by Apple and Microsoft.