/ResSysSpotify

Code for the final solution submitted to the RecSys Challenge 2018.

Primary LanguageJupyter NotebookApache License 2.0Apache-2.0

RecSysSpotify

Los playlistos Team: Anna Bach, Jose Mena, Guillem Pascual, Oriol Pujol, Jordi Vitrià and Santi Seguí Univerisy of Barcelona. you can send us an email at santi.segui@ub.edu.

REQUIREMENTS:

  • Python >=3.5
  • gensim
  • anaconda
  • tensorflow >=1.5
  • whoosh

STEP 1: Obtaining The Data

  1. Download Spotify's official dataset and place the 'data' folder into the root folder of the project.
  2. save "challenge_set.json" into "utils_spotify" folder

STEP 2: CREATE DICTIONARY AND MODELS used in the final recommender function

RUN:

  • 1_MPD_id_trackuri_maker.ipynb
  • 2_common_titles_and_tracks_maker.ipynb
  • 3_MPD_line_sentences_maker.ipynb
  • 4_word2vec_tracks.ipynb
  • 5_word2vec_titles.ipynb
  • 6_word2vec_artists.ipynb

STEP 3:

  1. RUN all notebook "create_final_submission.ipynb" to create submission

License

Usage of the Million Playlist Dataset is subject to these license terms