Multiple Enhanced Contrastive Learning for Collaborative Filtering

Collaborative filtering is a widely used technique for recommender systems, but it suffers from data sparsity, which leads to poor model robustness and generalization. In this paper, we propose a multiple-enhanced contrastive learning model that can balance the collaborative and augmented information in the recommendation model. Our model consists of two branches: one branch uses the original bipartite graph to learn the collaborative information, and the other branch uses an augmented graph to learn the augmented information. We use a contrastive fusion layer to merge the two branches and optimize the contrastive learning objective. The contrastive fusion layer can effectively enhance the robustness and generalization of the model by fusing the augmented graph representation and the embedding representation optimization information while retaining more high-order interaction information. We evaluate our model on three public datasets and compare it with several baseline methods. The experimental results show that our model can significantly improve the recommendation accuracy.