Many people these days use, either individually (using Mozilla Thunderbird, for instance), company-wide, or in their email service (such as GMail, Hotmail, Yahoo! Mail, etc.) “learning” anti-spam filters (also called “bayesian filters”). But how many people use them correctly?
You see, the thing about those filters is that they have to be taught. And many people simply ignore (or are unaware of) that part, and the filters remain inefficient and untrustable for them. Some spam keeps arriving in their inbox, and they have to continuously check the “Spam” folder for false positives (legitimate mails wrongly tagged as spam).
Which is sad, because that’s an easily fixable situation.
To put it simply: you should ensure that every mail, when deleted, was previously correctly tagged, either as spam or as non-spam. That’s how the filters learn and improve.
For instance, let’s say that a spam mail slips through the filter and arrives in your inbox. What should you do? Many people simple delete it - thus giving the filter the message: “this was not spam, I’ve simply deleted a normal message.” In other words, the filter has just learned the wrong thing. What you should have done was to tag it as spam first (in GMail, for instance, the button is labeled “Report Spam”), and then go to the Spam folder, and then delete it there. By doing this, the filter learned what you wanted it to learn - that that mail is spam (and so are mails like it).
Another possibility is a false positive. A legitimate mail goes (wrongly) to your Spam folder. Should you read it and delete it? No! By doing that, you’re reinforcing the error, you’re telling the filter that that mail was, indeed, spam. You should remove the “Spam” tag from that mail before you delete it (for instance, in GMail, the button is called “Not Spam”).
Another common error: you subscribe to some site’s mailing list, and you find out that more than half of the mails you receive from there simply don’t interest you. It may be tempting to simply tag those as spam, and get rid of them forever that way. But that’s the wrong thing to do - they’re not really spam, since you subscribed to that list; they’re just not interesting to you. What you should do is simply unsubscribe to the list - or, if a part of those mails actually interest you, you merely delete those that don’t. By reporting them as spam, you’ll keep receiving them - they’ll just go to the Spam folder. All of them, even those that might interest you. You’ll also be “confusing” the filter; those mails are legitimate. It’s like punishing a dog for obeying you - you’ll only be sending mixed messages, and confusing the hell out of the poor thing.
If, instead, you do all of this correctly, then after a couple of months your filters will actually become trustworthy - you’ll almost never see any spam mail in your inbox, and you can be virtually sure that what is in your Spam folder is spam.


Recent Comments