ITEM-BASED COLLABORATIVE FILTERING
As you’ll recall from Chapter 2, recommender systems are not a specific algorithm but rather an application of machine learning and data mining. The two primary architectures for recommending items are content-based filtering and collaborative filtering. This chapter examines collaborative filtering.
Collaborative filtering recommends items to a user based on analysis of similar users and their preferences, past purchases, ratings, or other general behavior. Collaborative filtering can also be split into two methods.
The first method is user-based collaborative filtering, which generates recommendations to a target user based on the historical preferences of similar users. Another way of expressing this is people similar to you who buy x also buy y.
In practice, like-minded users are first identified, and their ratings or preferences are then collected and grouped to produce a weighted average. The group’s general preferences...