Algorithm for recommending content to user on the basis of tags

Question

I am looking for a good algorithm that can recommend content objects to user by calculating similarity between user and content object. To calculate it, we have the content object tags (meta data) and user's interest data.

We can learn about user's interest in two ways:

1. Explicitly asking him: Ask him to rate a particular content item. to rank a collection of items from least fav to most fav.
2. Implicit ways: Learn by observing what kind of content a user accesses over the time. I want to implement a bit of both.

Please suggest some articles or papers that shows analysis of some good approaches?

Answer 1

This is an active area of research, so there are lots of papers on the topic. Try for example "An efficient boosting algorithm for combining preferences" by Freund et al. The Journal of Machine Learning Research vol. 4 at http://jmlr.csail.mit.edu/papers/volume4/freund03a/freund03a.pdf

Answer 2

Book: "Collective Intelligence in Action" by Satnam Alag.

Answer 3

Check out the Netflix Prize Reference section in wikipedia.