Good to know PHP - Spam mod

I just wrote a small spam mod for this blog software. Lets see if it works.

eleven comments, already:

Testing if comments still work.. yes! :)
niko - 21 08 05 - 07:49

Yep it works ;)
How’s the work on the .NET wrapper?
Raedwulf - 21 08 05 - 10:51

“Have you got anything without spam?”
“Well, there’s spam egg sausage and spam, that’s not got much spam in it.”
Factory - 21 08 05 - 11:13

buy your own eastern european wife online right now! 20% off, only this week!
(if this gets on, you failed)
Armen138 (link) - 21 08 05 - 14:30

better luck next time :P
Armen138 - 21 08 05 - 14:31

‘small’ spam mod doesn’t suggest it works with sophisticated AI to detect spam messages but does suggest it does something ie. if you noticed all the spam messages had the email field filled with random chars. That might help in detecting the spam-still this concept does approach ai – sorry :D
Raedwulf - 21 08 05 - 21:10

you dont neccesarily need a good AI routine to do this, three tables of ‘good’, ‘bad’ and ‘ignored’ words in mySQL, then just anaylse each word and create a ‘score’. Then all you have to do is set a baseline score, anything above it is good, below it is spam and deleted/highlighted.
John C - 21 08 05 - 23:00

That’s simple :D
Raedwulf - 22 08 05 - 08:12

Wouldn’t a baysian approche do just that but instead also make a smart assumption about the score. You act on the notion that all bad words only come out in bad content. But baysian for sorting out bad and good comments in PHP sounds a little overkill. I just wanted so state that in all the papers I have read that notion of counting bad words have been proven to fail. I saw one with counting good and bad (somewhat a naive baysian classifier) can’t say anything about the result.
Mikael B - 22 08 05 - 13:29

The way which i suggested (i.e. detect emails with words in them as ‘good’ and random chars as ‘bad’ ) might work but of course if you had an email like or it would be caught because they don’t make sense to a word detector :P
Raedwulf - 22 08 05 - 16:49

hehe :) my current anti-spam strategy is just a list with bad words combinations like ‘super poker’, because I assume no one will post something about that in this blog :)
niko - 22 08 05 - 19:59

Displayed name:  
Remember personal info?
Website (optional):
Enter "layered" (antispam):
Comment:Emoticons / Textile


Small print: All html tags except <b> and <i> will be removed from your comment. You can make links by just typing the url or mail-address.
Note: If you type in your email adress above, it will be visible to other visitors, although it will be hidden for bots using javaScript.