Saturday 11 October 2008

Introducing the Recommendation Engine

One of the problems with the internet has been that there is so much information out there, and so much being added every second, that its not possible to keep up. So how to you see through all this clutter to the information that matters to you?

Well, the answer is that you use social bookmarking sites like ma.gnolia or del.icio.us or content sharing and discovery site like stumbleupon, digg or mixx. These work by allowing you to see what information other people are looking at, or finding interesting. In effect its like a giant popularity contest the more people recommending a link the more likely it is to be displayed on the main page. Recently Yahoo introduced 'buzz' combining their search analytics with user recommendation.


In recent years there have been all sorts of interesting improvements to social discovery, in terms of the ways that information is displayed digglabs coming up with the stack, swarm, bigspy and arc and most recently pics options. While these are all really fun, they are still more entertaining than labour saving. Sure splitting the information into groups is useful, but its still not accurate enough to make it really useful.

I spend most of my day connected to the internet in one way or another, so its no surprise I have tried most of the social bookmarking and content sharing sites, and one of the things that i have always wished for, was the ability to submit a url to a site and have the site come back with recommendations on related information that i might find useful.


It would seem that this request was at last been answered, Digg.com just launched a Beta of a service they are calling the recommentaion engine. The recommendation engine works in a quite complex way cross referencing what it knows about you, with what it knows about people with similar tastes to you, and what it knows about the url you submitted to come up with links to relevant information.




0 comments: