This repository is the code of our paper：
"Feature-decorrelation Adaptive Contrastive Learning for Knowledge-aware Recommendation"
main contributions:
-
We propose a feature decorrelation method for GNNs on KGs and IGs.
-
We design an adaptive knowledge purification method to extract effective knowledge semantics on the KGs and generate augmented views for downstream tasks.
-
We design a cross-view semantic alignment method using contrastive learning.
keywords: Recommender system, Graph neural networks, Contrastive learning, Knowledge graph
the framework of our model:
