C-Command Software Forum

How Sensitive Is Spamsieve?

My question is this:

I subscribe to a news site - However, occasionally this same site puts out adverts containing no news, under the same banner.

I want to read one but not the other.

If I mark the news releases as “good” and the adverts as “spam”, will SS be able to differentiate?

I think I know the answer - but the more I know about this program, the better I like it.

Yes, if you train SpamSieve with examples of both types of messages, it should learn to tell them apart.

Wow!
Indeed Wow!

I did not expect that answer. Great product - In three days, already my Inbox is darn near as clean as it was when I first opened it.

Even now, before it is fully trained, it is operating at 87% effectiveness. I have corrected the few mistakes it has made which should make it even more efficient down the road.

Again, Great Product, Michael! Thanks.

Anything Special After 1000
Welp, 12 days into the SpamSieve experience and I am quite pleased with the results.

I am using an iMac, round bottom OS X, version 10.4.11, 2 gigs DDR SDRAM, 1.25 GHz PowerPC G4. MAIL PROGRAM.

283 Good Messages
678 Spam messages (71%)
49 SpM Messages per day

46 False Positives
20 False negatives

93.1% Correct

Corpus
272 Good Messages
575 Spam Messages (68%)

64,476 total words

Rules
468 Blocklist Rules
418 Whitelist rules

I’ve been fairly diligent in marking all incoming messages - Also I have changed several from spam back into good and vice versa.

Questions (Whew)

. After 1000 incoming - will anything change or will something pre-programmed “kick in” that I should be aware of?

. Do you see anything out of kilter with the figures I’ve provided?

Thanks Michael,
Vince

No. It doesn’t switch into a different mode, or anything.

They look fine. If you use the Set Date button to change the date to yesterday, you can get a sense of what the recent accuracy is. The overall accuracy is not that informative at this point, since it includes the period when SpamSieve wasn’t trained.

By the way, there’s a button to copy the statistics to the clipboard so that you don’t have to re-type them.