SVD-GCN

Codes for CIKM 2022 paper SVD-GCN: A Simplified Graph Convolution Paradigm for Recommendation

Environment

The algorithm is implemented in Python 3.8.5, with the following libraries additionally needed to be installed:

• Pytorch+GPU==1.8.0
  
• Numpy==1.19.2
  
• Pandas==1.1.4
  

Run the Algorithm

1. Run preprocess.py to get the required number of singular vectors/values. To make the calculated singular value/vectors more accurate, q is expected to set (slightly) larger than req_vec (K in the paper).
2. Run SVD-GCN variants. SVD-GCN-B/U/I/M needs training, while SVD-GCN-S does not require any optimiziation. Param_Settings includes parameter settings for datasets used in this work.