Search papers, labs, and topics across Lattice.
This paper introduces HeterSEED, a novel framework for heterogeneous graph learning designed to address the challenges posed by heterophily, where connected nodes have dissimilar labels. HeterSEED decouples representation learning into a heterogeneous semantic channel and a structure-aware heterophily channel, using pseudo-label-guided partitioning to separate and aggregate homophilic and heterophilic neighborhoods. Experiments on five real-world heterogeneous graphs demonstrate that HeterSEED outperforms existing methods, particularly in highly heterophilic scenarios, and the authors provide theoretical justification for its improved expressiveness and reduced prediction bias.
HeterSEED achieves state-of-the-art performance on heterophilic heterogeneous graphs by decoupling semantic and structural information, offering a more robust approach than relying on feature similarity alone.
Many real-world heterogeneous graphs exhibit pronounced heterophily, where connected nodes often have dissimilar labels or play different semantic roles. In such settings, standard heterogeneous graph neural networks that aggregate messages along metapaths or meta-relations primarily based on feature similarity can propagate misleading information, since feature similarity may be misaligned with underlying relational semantics. In this paper, we propose HeterSEED, a semantics-structure decoupling framework for heterogeneous graph learning under heterophily. HeterSEED decouples representation learning into a heterogeneous semantic channel that captures type- and relation-aware local semantics and a structure-aware heterophily channel that separates homophilic and heterophilic neighborhoods via pseudo-label-guided partitioning and aggregates them using metapath-based structural weights. A node-level adaptive fusion mechanism then combines the two channels to produce context-dependent node representations. Theoretically, we establish that, on heterogeneous graphs under heterophily, HeterSEED is strictly more expressive than standard heterogeneous graph neural networks that rely primarily on feature similarity and provably reduces the prediction bias introduced by heterophilic neighbors. Experiments on five real-world heterogeneous graphs, including two large-scale networks at the million-node and hundred-million-edge scale, demonstrate that HeterSEED consistently outperforms representative heterogeneous graph neural networks and recent heterophily-aware baselines, especially in strongly heterophilic regimes.
