Book recommendation system using TF-IDF and cosine similarity

The number of new books that appear on the market makes it difficult for readers to find books that meet their needs and desires. There are many new books every year. Sometimes there are productions with the same or similar titles but with different authors or publishers. The content may be other, even though the title is identical. It may also happen that a similarly titled book is a sequel to an earlier one. Therefore, this paper will focus on a content-based book recommendation system. The book content used in the system is the book title. The system provides book recommendations with book titles that are the same or similar to the book selected by the user. The way the system works is to extract the book title feature with Term Frequency-Inverse Document Frequency (TF-IDF). Then, all book titles are compared with cosine similarity. There are two kinds of results from this book recommendation system. First, 20 books are similar to the book selected by the user, where the 20 books have a similarity level above 50%. Second, 20 books have a similarity level below 50%. Both types of results are offered to users to be selected according to their preferences.