Neural Tree Collaborative Filtering: Rethinking Graph Collaborative Filtering as Tree Collaborative Filtering with Curvature-Aware Propagation Depth

2026-08-10

AI Summary

This work addresses the limitation of traditional graph collaborative filtering methods that employ a uniform propagation depth, thereby neglecting structural heterogeneity across nodes and often causing oversmoothing at peripheral nodes while under-exploring central ones. To overcome this, we propose a curvature-aware adaptive propagation mechanism that reconstructs the user–item interaction neighborhood as a rooted tree and dynamically assigns node-specific propagation depths based on discrete Ricci curvature—a proxy for local degree imbalance. Our approach strictly generalizes NGCF, preserving deep discriminative information in positively curved regions, and offers both theoretical guarantees and architectural generality. Experiments demonstrate that the proposed model outperforms state-of-the-art GCF backbones on three public benchmarks and functions effectively as a plug-and-play module to enhance advanced self-supervised recommendation systems.

Abstract

Graph Collaborative Filtering (GCF) has become the dominant paradigm in modern recommender systems by modeling user-item interactions as a bipartite graph and propagating embeddings through a fixed number of message-passing layers. However, applying a uniform propagation depth to every node ignores a fundamental property of real interaction graphs: nodes differ substantially in their local connectivity, so peripheral nodes quickly suffer from over-smoothing while hub-like nodes remain under-explored beyond their immediate neighborhood. In this paper, we revisit GCF from a tree-structured perspective and propose Neural Tree Collaborative Filtering (NTCF), a framework that re-interprets each node's local neighborhood as a rooted tree and assigns a node-specific propagation depth based on a closed-form local-degree-imbalance score that serves as a discrete Ricci-curvature proxy. We provide a theoretical analysis showing that (i) NTCF strictly generalizes NGCF, degenerating to NGCF when all curvature-induced depth adjustments vanish (a lower bound on its representation power), and (ii) the curvature-aware schedule retains strictly more discriminative information at deep layers on positively-curved (peripheral) nodes than uniform-depth propagation. NTCF can achieve higher performance than most widely used GCF backbone models and can be integrated into existing advanced self-supervised models as a backbone, replacing their original backbone to achieve enhanced performance. Extensive experiments on three public datasets demonstrate the superiority of NTCF.

Problem

• Graph Collaborative Filtering
• over-smoothing
• propagation depth
• node connectivity
• Ricci curvature

Innovation

• Neural Tree Collaborative Filtering
• Curvature-Aware Propagation
• Graph Collaborative Filtering
• Ricci Curvature
• Node-Specific Depth