Collaborative filtering—a rating-based recommendation engine

By recommending only similar items or items from similar users, your users might get bored of the recommendations provided due to the lack of diversity and variety. Once a user starts interacting with a service, for example, watching videos on YouTube, reading and liking posts on Facebook, or rating movies on Netflix, we want to provide them with great personalized recommendations and relevant content to keep them happy and engaged. A great way to do so is to provide a good mix of similar content and new content to explore and discover.

Collaborative filtering is a popular approach for providing such diverse recommendations by comparing user-item interactions, finding other users who interact with similar items, and recommending items that those users also interacted with. It's almost as if you were to build many custom stereotypes and recommend other items consumed from the same stereotype. Figure 11.6 illustrates...