CONTENT-BASED FILTERING

 

As established in Chapter 2, content-based filtering methods provide recommendations based on similar item attributes and the profile of an individual user’s preferences. The content-based filtering system then attempts to recommend items similar to those that a user has liked or browsed in the past. After purchasing a book about “machine learning,” for example, Amazon’s content-based filtering is likely to serve you other books:

- from the same author,

- from the same category, e.g., data science, and

- have similar title keywords, e.g., “machine learning.”

As expected, content-based filtering relies heavily on a description of the item’s characteristics and the profiling of individual user preferences. Under this model, items such as documents, online posts, images, and videos need to be adequately described in the form of keywords/tagging/metadata or through more sophisticated...