Chapter 5: Training Recommender Systems
In this book, so far we have worked through recipes to train deep learning with fastai for a variety of datasets. In this chapter, we will go through recipes that take advantage of fastai's support for recommender systems, also known as collaborative filtering systems. Recommender systems combine the characteristics of tabular data models introduced in Chapter 3, Training Models with Tabular Data, with characteristics of text data models introduced in Chapter 4, Training Models with Text Data.
Recommender systems cover a narrow, but well-established, use case: given a set of users and their ratings of a set of items, a recommender system predicts the rating that a user will give for an item that the user has not rated yet. For example, given a set of books and a set of readers' assessments of these books, recommender systems can make predictions about a given reader's assessment of a book they haven't read yet.
In this...
