- Collaborative Filtering is a standard data mining technique used
for web sites.
- It's not particularly new (~1989), but is really commonly used.
- Here is a paper on
Eigentaste: A Constant Time Collaborative Filtering Algorithm.
- The idea is that you want to use one user's preferences (e.g.
internet clicks) to rank things for other users.
- One thing to note is that it uses proof to show it
is O(1) (ok it doesn't show the proof, but ...) for lookup.
- Secondly, it shows simulations on data sets.
- Note it doesn't use clicks but user rankings.
- Solid methodology (proof and simulation), highly cited.