In this project, we created a movie recommendation system based on graph signal processing. The task is to recommend movies to a user given some ratings of that user. More details can be found in our final report.

In order to run the code, the following dependecies need to be dowloaded on your machine.

For a detailed way of installing most of libraries all at once follow these installation instructions. You will also need the following library.

• [Surpise] - Install Surprise library

  $ conda install -c conda-forge scikit-surprise

The data containing can be dowloaded here MovieLens 100k and should be moved inside the Data folder.

• Main.ipynb: Contains our recommendation system. To get the 10 most relevant recommendations for a given user you can run the get_recommendations() function. We also compare our model with a matrix-factorization based recommendation system. You can get the prediction given by this model using the baseline_recommendations() function.

• Deniz Ira
• Jonathan Labhard
• Daniil Dmitriev
• Paul Griesser