MC-HNN: Learning Latent Structural Semantics and High-Rank Representations for Hypergraph Neural Networks

Hypergraph Neural Networks (HNNs) have emerged as powerful tools for modeling complex high-order correlations. Most existing HNNs adhere to a two-stage message passing paradigm, where node feature propagation is mediated by hyperedges. In this paper, we analyze two structural limitations of this paradigm, which we term rank collapse and hyperedge semantic dependency. To address these challenges, we propose the Multi-Channel Hypergraph Neural Network (MC-HNN). We design a multi-channel message passing mechanism to maintain high-rank representations, while simultaneously introducing a latent hyperedge type encoding mechanism to inject an independent degree of freedom into hyperedge representations. Our analysis and experiments suggest that MC-HNN alleviates these bottlenecks and achieves strong empirical performance.

Hypergraphs are a way to represent relationships that involve more than two objects at a time, such as a paper written by multiple authors or a group of items connected by a shared event. This paper studies how neural networks learn from hypergraphs and identifies two limitations in a common message-passing design: information can be compressed too strongly, and hyperedge representations may depend too heavily on the nodes they connect. We propose MC-HNN, a model that uses multiple information channels and latent hyperedge types to preserve richer representations. Experiments on several benchmark datasets show that this design improves hypergraph learning performance.