Part 1. Recommendations

This first part of the book, including chapters 2 through 6, explores one of the three pillars of Apache Mahout’s machine learning implementations: collaborative filtering and recommendation. With these techniques, you can understand a person’s tastes and find new, desirable content for them automatically. This part is also a warm-up for the rest of the book, which will depend heavily on the Apache Hadoop distributed computing framework. You’ll meet machine learning in Apache Mahout in simple Java first, and then in Hadoop.

Chapter 2 introduces recommender engines, as implemented in Mahout, and covers evaluating performance in the context of a runnable example. Chapter 3 discusses representing recommender data efficiently in Mahout. Chapter 4 catalogs the various recommender engine implementations available in Mahout and their varying features and attributes.

Chapter 5 presents a case study based on data from a dating site, which shows how you can adapt the approaches in Mahout to cope with real-world data and produce a production-ready recommender. Finally, chapter 6 provides a first look at Mahout’s use of Apache Hadoop to implement a large-scale distributed recommender engine.