Collaborative Filtering is a recommendation system approach that predicts user preferences based on similarities between users or items. It requires past interaction data and is widely used in platforms like Netflix and Amazon. Both user-based and item-based techniques exist, each having its pros and cons.