/Graph_Music_Recommendation

This is the repository of our MLNS Project, trying to test some music recommendation with graphical methods.

Primary LanguageJupyter Notebook

cover

Graph Music Recommendation 🎶

This is the repository of our Project for the Machine Learning for Network Science (MLNS) course from CentraleSupélec by Chloé Daems, Amir Mahmoudi & Anne-Claire Laisney.

Motivations ✍️

We want to create a Recommendation System applied to music by using the notions seen in course. Our work was inspired by the Katarya, R., Verma, O.P. Efficient music recommender system using context graph and particle swarm. Multimed Tools Appl 77, 2673–2687 (2018). paper which showed some great results. We are using the same dataset as them i.e. data fetched with the user.getRecentTracks request at the Last.fm API. We have two main tsv files that you can download here :

  • The userid_profile.tsv which regroups informations on the user (userid \t gender \t age \t country \t date of registration);
  • The userid_ ... _logs.tsv which regroups informations on the log (userid \t timestamp \t artist-id \t artist-name \t track-id \t track-name)

The dataset regroups the whole listening habits (Jan, 27th 2008 till May, 5th 2009) of nearly 1,000 users, regrouping 19,150,868 logs.

VIDEO PRESENTATION 💻

Here you can find the link to our project presentation.