If you have time for just one book to get yourself up to speed with the latest and best in recommender systems, this is the book you want. Recommender systems handbook is a carefully edited book that covers a wide range of topics associated with recommender systems. It is neither a textbook nor a crash course on recommender systems. For those who do have an inkling of what recommender systems are, this is an excellent educational resource on the main techniques employed for making recommendations, as well as how to evaluate such recommendations. It also shares interesting case studies on real-life applications of recommender systems, such as their role in the success of the video streaming company Netflix.

Recommender systems are typically implemented either as (1) content-based recommender systems or (2) collaborative filtering-based systems. In the former, recommendations are made based on the content of the item to be recommended and how much the system believes a user is interested in the content. In the latter, the likes and dislikes of other users are considered. The assumption here is that people should generally share the same interests as others who are like them.

The first section of the book takes a look at these primary techniques. However, instead of repeating existing literature on basic implementations for these systems, this book dives into state-of-the-art specializations of both content-based and collaborative filtering systems. It talks about context-aware and constraint-based content-based recommender systems and discusses new advances in collaborative filtering, including neighborhood-based collaborative filtering. This is immediately succeeded by a section discussing evaluation methodologies and metrics that can be used to assess recommender systems.

In the third section, the book goes through several important applications of recommender systems. The diverse set of case studies and examples helps illustrate the impact that recommender systems can have. They are not the brainchild of reclusive scientists working from ivory towers. Instead, recommender systems have found use in many aspects of our lives. They generate recommendations for what we watch on video streaming services or listen to on music streaming services. They help suggest courses we might be interested in when we sign up on an online continuing education website. If you have ever wondered how social networks like Facebook are so creepily accurate at suggesting friends you might know, this book explains the science behind it.

The last section of this book talks about the less headline-grabbing aspects of recommender systems. These include the impact of such systems on human interaction, as well as new, upcoming research topics, such as the incorporation of active learning or taking into consideration multiple recommendation criteria. I find the discussion on building robust recommender systems particularly interesting. In this discussion, the book talks about possible attacks, or ways to game recommender systems, as well as the mitigating techniques that could be adopted. Given the increasing prominence and impact of recommender systems, I feel that this will be an increasingly important area of study and advancement.

In about 1,000 pages, this book covers a wide gamut of topics in recommender systems. As I noted at the start of this review, it is not a textbook. It is not the book that you would want to start with if you were new to recommender systems. However, it is definitely a book to read to get updated on the state of the art of recommender systems, and also to get a feel of the breadth of the research areas available in this area.

More reviews about this item: Amazon