We have come a long way in this chapter. We first learned about document vectors and gained a brief introduction to the cosine similarity score. Next, we built a recommender that identified movies with similar plot descriptions. We then proceeded to build a more advanced model that leveraged the power of other metadata, such as genres, keywords, and credits. Finally, we discussed a few methods by which we could improve our existing system.
With this, we formally come to an end of our tour of content-based recommendation system. In the next chapters, we will cover what is arguably the most popular recommendation model in the industry today: collaborative filtering.
