Display Relevant Articles Based on Browsing History (similar to Google Ads)

[GregL83]

I was pondering the idea of collecting the browsing history of users on my site.

 

Which articles they read.  What they search for.  Where they have come from.

 

I wanted to start displaying relevant articles on my index page or suggested articles based on their browsing history.  I really have no solid idea on how to implement this tech.

 

The main concept would be:

1) Collect data about the user

2) Use user data to relate to relevant content

3) Display relevant content to the user when they login

 

Have any of you done this for an ad firm?

[GregL83]

BUMP...

 

No ideas?  Anyone want to take a stab???

[lastkarrde]

One solution (probably not the most efficient as I have no experience in the field) would be to..

 

• Record what pages containing content the user visits.
  
• Tag/tokenize the pages to give you a list of 10-15 keywords about the page. You could either do this manually or with a simple word search (ignoring words such as a,and,then,the etc..)
  
• Record the tags of pages the user visits
  
• Serve up recommendations based on the tags of pages the user has visited in the past
  

[GregL83]

lastkarrde,

 

I was thinking along the same lines of what you've suggested.  I think its a great start  and would work.  BUT i'm wondering if anyone has already done something similar, because sometimes you can't always see problems or efficiency issues until the system is already coded.

 

I'll do some more research and see what solutions i can come up with.

 

Otherwise, if anyone has done any kind of algorithms for ad relevance, your advice would be greatly appreciated.