News You Can Bruise for 2003 October 17

Hoo, if I may, hah. I've got my nitro-burning NewsBruiser Bayesian spam detector working and into CVS, and it is sweet. Now that I've got it working, I can implement comments in NewsBruiser with a relatively clear conscience. It's still a way for people to automatically write on your website, but now there's a way to automatically stop them from writing crap. There might even be comments on NYCB in the relatively near future (shock, gasp, etc.).

Other reasons why you should look at that screenshot: the first fake weblog name is really good, good enough for someone to use[0], and while making up fake spam for the sample Bayesian network I came up with the phrase "the set of all sluts".

I think the lack of comments is the leading cause of people who might use NewsBruiser not using NewsBruiser. Every time people complain about Blogger or Moveable Type, I (along with every other penny-ante weblog software author) see an opportunity to plug our own software. I'm too shy to actually do it, though. I'm stuck in the naive hacker's world where if I make something better everyone will automatically use it.

I still need weblog spam from you. I want to provide a starter Bayesian network with NewsBruiser, and I need me some data. The trackbacks and comments will share a Bayesian network, BTW; the datasets are undoubtedly similar, and that will cut down on training time.

PS: I made Reverend work with Python 1.5, because I live in a cave. If you're interested, it's here.

[0] Another weblog name I made up recently: "Commodius Vicus". You're welcome.