Chapter 3. Recommending relevant content

 

This chapter covers

  • Understanding recommendation engines based on users, items, and content
  • Finding recommendations about friends, articles, and news stories
  • Creating recommendations for sites similar to Netflix

In today’s world, we’re overwhelmed with choices, with a plethora of options available for nearly every aspect of our lives. We need to make decisions on a daily basis, from automobiles to home theatre systems, from finding Mr. or Ms. “Perfect” to selecting attorneys or accountants, from books and newspapers to wikis and blogs. In addition, we’re constantly being bombarded by information—and occasionally misinformation! Under these conditions, the ability to recommend a choice is valuable, and even more so if that choice is aligned with the person receiving the recommendation.

3.1. Setting the scene: an online movie store

3.2. Distance and similarity

3.3. How do recommendation engines work?

3.4. User-based collaborative filtering

3.5. Model-based recommendation using singular value decomposition

3.6. The Netflix Prize

3.7. Evaluating your recommender

3.8. Summary