Chapter 5. Putting It All Together: Content-Based Recommender

Throughout this part of the book, we’ve introduced some of the most basic components in a recommendation system. In this chapter, we’ll get hands-on. We’re going to design and implement a recommendation system for images from Pinterest. This chapter, along with the book’s other “Putting It All Together” chapters, will show you how to work with datasets by using open source tools. The material for this kind of chapter refers to code hosted on GitHub that you will need to download and play with in order to properly experience the content.

Since this is the first practical hands-on chapter, here are some extra setup instructions for the development environment. We developed this code on Windows running in a Windows Subsystem for Linux (WSL) Ubuntu virtual machine. The code should run fine on Linux machines, with more technical adaptation for macOS and a lot more for Windows, in which case it would be better to run it on a WSL2 Ubuntu virtual machine. You can look at the setup for WSL in the Microsoft documentation for Windows. We picked Ubuntu for the image. You will also need NVIDIA CUDA and cuDNN if you have an NVIDIA GPU and want to use it.

We will be using the Shop the Look (STL) dataset from “Complete the Look: Scene-Based Complementary Product Recommendation” by Wang-Cheng Kang et al.

In this chapter, we will show you how to build a content-based recommender. Recall that a content-based recommender uses indirect, ...

Get Building Recommendation Systems in Python and JAX now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.