 |
 SpamCombat - part II
by Andrew Cooper
Working with the program
After I completed all preliminary steps, the main window of the program opened up. It was neat and accurate. I liked its interface, as it seemed both comprehensive and simple to me.
My editor gave me an email account that, in his words, was practically littered with spam. He used to complain that he would spend half an hour sifting through his mail and deleting unwanted correspondence. When I checked the email account with SpamCombat, I was even more emotional; 286 suspected spam messages out of a total 339 is by all accounts outrageous! At 24 megabytes of rubbish, it was totally unacceptable.
SpamCombat flagged some legitimate messages as spam, as other programs of its type often do, but mostly, the program made the right judgments. Here are some stats that will help us evaluate the program's effectiveness:
Total messages in inbox: 339
Messages considered to be spam: by Cooper (me) 307 / by SpamCombat (program) 286
Messages I recognized as spam that the program regarded as legitimate (spam not caught): 38
SpamCombat's catch rate: 88.79% ((339-38)/339).
Messages the program recognized as spam that I regarded as legitimate (falsely considered spam): 17
SpamCombat's false-negatives rate: 5.01% (17/339*100%).
The above data shows that the program is at least potent. The comparatively low ratio of false positives combined with quite a strong showing of caught spam emphasize the effectiveness of the program.
Added to that, the ability to catch viruses and worms, colorful and revealing statistical graphs, and powerful learning features make the program an interesting candidate for a home antispam solution.
Among the negatives that surfaced while I was using the program were some interface imperfections, difficulties in selecting the messages targeted for deletion from the mailbox server, errors in terminating the program during the deletion process, and "stop" button transformations. Also, I didn't like the way the messages were separated by spam filter groups or the color (though customizable) of highlighted message lines.
PS: When the review was ready and published on the website, the authors of the program contacted me again and warned to be careful with using "spam trap" feature that can be configured while setting mail accounts at the program's first run. According to G-Lock, "spam trap" should only be applied when the account only receives 100% of spam. In this case, spam trap will automatically fully clean your email box, and will not analyze its contents for spam. This is done to help users instantly get rid of all messages arriving to a particular mailbox when the user knows that all of them are spam. Please take this matter into your attention.
Verdict
The program is promising, both in the way it carries out its basic task of protecting the user from spam and in its desire for self-learning. It is definitely worth trying. Just be careful with auto-deletion, early (uneducated) Bayesian filter permissions, and the process of selecting mails to be purged from the server.
Developer's web site: www.glocksoft.com
Download link: www.glocksoft.com/download.htm
Back to Part I
|
|
