Bayesian Classifiers and other algorithms
Attention - newbie programmers/PHP script kiddies leave now 
.
Hi all,
I'm fed up with article marketing. The way to make money is to build an awesome site. So let's talk algorithms!
I've built a membership site and I'd like to do a number of things:
1. Keep spammers away.
2. Keep user generated content clean (i.e. AdSense safe).
3. Classify members into different categories (scammer, troll, nice person etc. etc.).
4. Do some kind of thing like Amazon recommendations ("if you chatted to this member, then you might also like to chat to these members").
5. Do some photo processing and look for duplicate profiles with same photo, photos not of people etc. etc.
I wondered if any of you hard core programmer types were doing anything similar on your own sites. I've tested a few off the shelf membership sites, but their code is very basic (e.g. they just look for known lists of swear words).
First of all I'm not a mathematician and I never studied computer science :confused:. But the other day I found a couple of C# Naive Bayesian Classifier modules (here's an anti-spam one).
My site doesn't have enough data yet, but I can see that with 10,000+ members this stuff would work a treat. I've also found another one that will allow me to classify my members into different types. That impressed me a lot
.
However, I did a bit of Googling last night and realised that there are all kinds of other artificial intelligence/machine learning algorithms out there. So I wondered what sort of stuff I should look at to make my site the smartest membership site on the planet
. This has a practical use because being smarter than the average site is a great USP. It also looks rather awesome on my resume and could lead to consulting gigs for other site owners .
.Hi all,
I'm fed up with article marketing. The way to make money is to build an awesome site. So let's talk algorithms!
I've built a membership site and I'd like to do a number of things:
1. Keep spammers away.
2. Keep user generated content clean (i.e. AdSense safe).
3. Classify members into different categories (scammer, troll, nice person etc. etc.).
4. Do some kind of thing like Amazon recommendations ("if you chatted to this member, then you might also like to chat to these members").
5. Do some photo processing and look for duplicate profiles with same photo, photos not of people etc. etc.
I wondered if any of you hard core programmer types were doing anything similar on your own sites. I've tested a few off the shelf membership sites, but their code is very basic (e.g. they just look for known lists of swear words).
First of all I'm not a mathematician and I never studied computer science :confused:. But the other day I found a couple of C# Naive Bayesian Classifier modules (here's an anti-spam one).
My site doesn't have enough data yet, but I can see that with 10,000+ members this stuff would work a treat. I've also found another one that will allow me to classify my members into different types. That impressed me a lot
.However, I did a bit of Googling last night and realised that there are all kinds of other artificial intelligence/machine learning algorithms out there. So I wondered what sort of stuff I should look at to make my site the smartest membership site on the planet
. This has a practical use because being smarter than the average site is a great USP. It also looks rather awesome on my resume and could lead to consulting gigs for other site owners .
:)