chapter one

1 Recommendations, recommender systems, and components

 

This chapter covers

  • What is a recommendation
  • What is a recommender system, and some examples
  • What components are often contained in a recommender system
  • What data is used

Recommender systems are considered among the most important applications of AI, acting as the lens through which billions of users filter information on the internet. Whether you are searching, reading the news, or buying things on an e-commerce site, the content shown to you will often have been selected by recommender systems.

In this chapter, you’ll learn what a recommendation is, explore the data that powers recommendations, and understand the architecture underlying most production systems. We’ll examine concrete examples from companies like Netflix and Amazon, as well as other platforms, showing you not just what recommender systems do, but how they do it.

1.1 Starting on Recommender Systems

1.2 What is a Recommendation

1.2.1 User-to-item recommendations (U2I)

1.2.2 Item-to-item recommendations (I2I)

1.3 Building Recommenders with Different Data Types

1.3.1 Content-based recommendations with metadata

1.3.2 Collaborative filtering with Behavioral data

1.3.3 Contextual data

1.3.4 Available Datasets

1.4 What is a Recommender System

1.4.1 The four-stage recommender system

1.4.2 The four-stage architecture across domains

1.5 Further reading

1.6 Summary