If you use an external tool or feature of your email client to filter spam (and who doesn’t?), you probably lose legit emails once in a while. Legit emails that are filtered by anti-spam software are called false positives. You probably check your spam folder from time to time to look for those false positives, but it gets more and more annoying the more spam you get. I got tired of checking thousands and thousands of spam emails every week and found a nice feature of Mozilla Thunderbird that can be used to detect wrongly classified emails.
The feature of Mozilla Thunderbird that I use to detect false positives is the saved search folder. I just created a search with common keywords that I receive in legit emails and saved it as a virtual email folder. Whenever I click on that folder, Thunderbird searches my spam emails for these keywords. Instead of checking thousands and thousands of spam emails, I only have to look at around 5% of spam emails every week.
You might object by saying that a good Bayesian spam filter should be smart enough to detect legit emails by learning from these keywords or that I can still lose false positives by using a search folder instead of looking at all my spam emails. It is true that a Bayesian spam filter works really good even without such tricks, but from time to time I still have wrongly classified emails, despite training the filter regularly. And yes, I can still lose legit emails, that’s correct. But the chance that I get a false positive that doesn’t include one of these keywords is pretty slim and I’m really glad to have more time working on our products and on our business instead of looking at penny stock ‘tips’, bank scam emails or warez offers for hours.
Detecting false positives with Thunderbird
If you use an external tool or feature of your email client to filter spam (and who doesn’t?), you probably lose legit emails once in a while. Legit emails that are filtered by anti-spam software are called false positives. You probably check your spam folder from time to time to look for those false positives, but it gets more and more annoying the more spam you get. I got tired of checking thousands and thousands of spam emails every week and found a nice feature of Mozilla Thunderbird that can be used to detect wrongly classified emails.
The feature of Mozilla Thunderbird that I use to detect false positives is the saved search folder. I just created a search with common keywords that I receive in legit emails and saved it as a virtual email folder. Whenever I click on that folder, Thunderbird searches my spam emails for these keywords. Instead of checking thousands and thousands of spam emails, I only have to look at around 5% of spam emails every week.
You might object by saying that a good Bayesian spam filter should be smart enough to detect legit emails by learning from these keywords or that I can still lose false positives by using a search folder instead of looking at all my spam emails. It is true that a Bayesian spam filter works really good even without such tricks, but from time to time I still have wrongly classified emails, despite training the filter regularly. And yes, I can still lose legit emails, that’s correct. But the chance that I get a false positive that doesn’t include one of these keywords is pretty slim and I’m really glad to have more time working on our products and on our business instead of looking at penny stock ‘tips’, bank scam emails or warez offers for hours.