Sheaf4Rec: Sheaf Neural Networks for Graph-based Recommender Systems
Concetti Chiave
Sheaf4Rec introduces a novel Sheaf Neural Network model for graph-based recommender systems, outperforming existing models in performance and efficiency.
Sintesi
Recent advancements in Graph Neural Networks (GNN) have led to the adoption of Sheaf Neural Networks for recommendation systems.
Sheaf4Rec proposes a solution inspired by category theory, providing a more comprehensive representation of nodes and edges.
The model shows significant improvements in F1-Score@10 and NDCG@10 compared to state-of-the-art models like NGCF and KGTORe.
Sheaf4Rec also demonstrates efficiency gains in recommendation computation, with substantial runtime improvements.
The paper is structured into sections covering related works, methodology, implementation details, analysis of efficacy, findings, and future research avenues.