chapter four

4 Content-based recommendations

 

This chapter covers

  • How to design proper graph models for a content-based recommendation engine
  • How to import existing (not-graph) datasets in the graph models designed
  • How to implement working content-based recommendation engines

Suppose you would like to build a movie recommender system for your local video rental store.

Old-fashioned Blockbuster-style rental shops have largely been put out of business by the advent of new streaming platforms like Netflix, but some still exist here and there. There’s one in my town where back when I was at university (a long time ago) I used to go with my brother every Sunday to rent some action movies (keep this preference in mind; it will be useful later!). That’s besides, the important thing here is that this specific scenario inherently has a lot of peculiarities in common with real, more complex online recommender systems. These include:

4.1   Representing item features

4.2   User modeling

4.3   Providing recommendations

4.4   Advantages of the graph approach

4.5   Summary

4.6   References