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:

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).
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.

Name		Name	Last commit message	Last commit date
Latest commit History 8 Commits
datasets		datasets
Param_Settings		Param_Settings
README.md		README.md
preprocess.py		preprocess.py
svd_gcn_b.py		svd_gcn_b.py
svd_gcn_i.py		svd_gcn_i.py
svd_gcn_m.py		svd_gcn_m.py
svd_gcn_s.py		svd_gcn_s.py
svd_gcn_u.py		svd_gcn_u.py