LFM-1b : music recommendation

Data :

• LFM-1b Dataset : music listening events.
  source : http://www.cp.jku.at/datasets/LFM-1b/

Methods :

1. caser : caser_pytorch/ A CNN-based next-song prediction model.
   Reference : Personalized Top-N Sequential Recommendation via Convolutional Sequence Embedding, WSDM 2018.

Files in this folder:

1. music_data_process.py : process the LFM-LE data, split sequence to sub-sequences, and save to files.
2. music_data_examination.py : check the overlaping rate : how many testing items doesn't appear in trainset.