Inspiration
Reading is an essential part of life; however, good books do not just come and find you. You have to find them. Our team realized that many of the current book recommendation lists out on the Internet today do not work so well, and you don't always get books you will like, so we decided to build our own.
What it does
Booklog allows you to enter books you recently read or enjoyed. Using modernized algorithms, it finds a list of books related to the books you entered, giving you more and more books to read and enjoy.
How we built it
We used a Django backend for authentication and database, with a RESTful API. We downloaded a huge dataset of books, ratings, descriptions, and more, and imported it into the database. We then developed algorithms to find related books and trained the algorithm on the dataset. Finally, we imported the connections into the backend and hooked everything up.
The frontend is build with Angular. It uses the backend's API to retrieve stored information and recommended books for the user.
Challenges we ran into
At first, we tried to use a machine learning algorithm consisting of neural networks to find book recommendations. However, we soon realized that it was not a good fit for the problem, and had to think up a new algorithm from scratch.
We also used many frameworks and libraries no one was familiar with, such as the Django REST Framework, ng-bootstrap, and more.
Accomplishments that we're proud of
We're proud of getting our whole project to work extremely well. The recommendations are mostly on-spot and very accurate.
What we learned
We learned that for an algorithm to work efficiently, it needs a lot of data to train on. If we only had a few ratings, or a few books, the result would have been a lot worse than this. The sheer amount of data we used greatly affected the final product. We also learned how to use a REST API to grab information from the backend on the frontend, whereas we used to use Firebase or other frameworks for our projects.
What's next for Booklog
We are very excited and think that this is an extremely viable product for today's market. We plan to continue developing it, making the algorithms even better, and adding more and more features to completely refine the app.
Log in or sign up for Devpost to join the conversation.