Our project is to create a service to a recommender system for movies based on your personal tastes. Our project combines our interests in scalable web systems, data science, and information retrieval to help imitate modern systems.
Here are some of our goals:
- Develop a dashboard of user recommendations
- Display movies where users can rate and search through movies to obtain accurate recommendations. We'll need the Movie Database API to grab information on movies, such as their titles, posters, and more.
- Movie Recommendation Engine in Spark
- Utilize collaborative filtering and an Alternative Least Squares model approach via PySpark to recommend movies based on previous user preferences and ratings. We'll be using the MovieLens 25M dataset to train our movie data.
- Login/Authentication & User Management
- Provide a basic authentication and user management system that a user can update their information. It's important for users to have some of these features for reasons related to customization and security.
First, the required datasets will be downloaded. This includes:
- The MovieLens dataset
- Transformed ratings (to account for our like and dislike system)
You can run the following commands to download the scripts and place them into their correct destinations:
makeIf you would like to download the files manually, make sure they are stored in the proper directories:
- MovieLens:
./services/database/data/ml-25m - Ratings:
./services/recommender/data
Afterwards, change the working directory to the services folder and run:
docker-compose up --build -dThis ensures you have the latest build and you can interact with your terminal while our application is running in the background.
-
P1:
-
P2: