Bayesian Filtering for Comments

Bayesian filtering for MoveableType? I didn’t even know such a thing existed! I found a plugin written by James Seng. At the time, I was actually looking for one of those little images that aren’t OCR readable and the user has to type whatever’s in the image (I’ve since discovered that’s called captcha and is implemented using the SCode plugin). I knew that this method introduced usability problems for the visually impaired and people who use text only web browsers, but I felt that it was a fair trade off in order to stop SPAM. But, in the process of looking for this, I figured I may as well look at other solutions, too (especially considering that GD isn’t working on my machine right now). Wouldn’t you know it, I found one I liked even better. I’ve liked the theory behind Bayesian filtering every since I’ve read Paul Graham’s A Plan For Spam. Mozilla Thunderbird uses this method of filtering, and while at first it’s quite inaccurate, it gets better with time (actually, with a larger sample of both good and bad stuff). Anyway, I’ve now implemented this method for comments on this site. I should now be able to re-open all my entries to comments, undoing the change I made back in November 2003.

One Reply to “Bayesian Filtering for Comments”

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.