DCCDR

Code for paper [Disentangled Contrastive Learning for Cross-Domain Recommendation].

Accepted by DASFAA 2023.

Requirement

• Python 3.6
• PyTorch 1.10.2
• Numpy

Files in the folder

• OurModel.py: our implementation of model
• run.py: model training and testing
• data/: four cross-domain recommendation tasks based on two widely used datasets Amazon and Douban
  • Amazon/Cell_Elec/
  • Amazon/Movie_Music/
  • Douban/Movie_Book/
  • Douban/Music_Book/
• utils/
  • load_data.py: auxiliary functions constructing training set and testing set for cross-domain scenario
  • parser.py: some parameters concerned with the model
  • helpers.py: functions to save the model

Running the code

1. default: python run.py (Task: Amazon-Movie-Music)

   If you want to choose a certain task: python run.py --dataset=[chosen dataset] --domain_1=[chosen domain 1] --domain_2=[chosen domain 2]