Collaborative Filtering and Personalization

The main difference between collaborative filtering and personalization is the method. Both aim at giving the user/consummer something that is adapted to his taste.

In collaborative filtering the main idea is this: Those who tend to have liked the same thing in the past will like the same thing in the future, therefore we can give recommendations to the user by looking at what other users with the same past tastes have also liked.
A good example of that can be found on last.fm where the sites creates a radio for you of titles you have previously not listened to, but that other users who listen to similar music have listened to. Another example is Amazon, that offers recommendations based on what users with similar purchases have also purchased.

Personalization is not based on anyone else than the user himself. It is based on his choices or his status (platform, location...) an example of that would be www.versiontracker.com. the site provides updates and software on several platforms, but it takes the user straight to the menu of the platform the user is using at the time, hence if you are on Mac Os X at the time you access the site, it will take you straight to the mac download platform. Another example is myspace.com that takes you straight to the french version of the site if you access it from France by detecting your country of origin through your IP address (although using a Proxy will render this method ineffective).

In effect collaborative filtering and personalization have the same goal: making the site more relevant to the user, but they are based on different methods. Personalization is based on positive data from the user himself, whether communicated voluntarily or not. Collaborative Filtering is based on Speculative data from other users that the system recognizes as having similar taste.