English

Feature Propagation on Knowledge Graphs using Cellular Sheaves

Artificial Intelligence 2026-01-19 v2 Information Retrieval Social and Information Networks

Abstract

Many inference tasks on knowledge graphs, including relation prediction, operate on knowledge graph embeddings -- vector representations of the vertices (entities) and edges (relations) that preserve task-relevant structure encoded within the underlying combinatorial object. Such knowledge graph embeddings can be modeled as an approximate global section of a cellular sheaf, an algebraic structure over the graph. Using the diffusion dynamics encoded by the corresponding sheaf Laplacian, we optimally propagate known embeddings of a subgraph to inductively represent new entities introduced into the knowledge graph at inference time. We implement this algorithm via an efficient iterative scheme and show that on a number of large-scale knowledge graph embedding benchmarks, our method is competitive with -- and in some scenarios outperforms -- more complex models derived explicitly for inductive knowledge graph reasoning tasks.

Keywords

Cite

@article{arxiv.2309.03773,
  title  = {Feature Propagation on Knowledge Graphs using Cellular Sheaves},
  author = {John Cobb and Thomas Gebhart},
  journal= {arXiv preprint arXiv:2309.03773},
  year   = {2026}
}
R2 v1 2026-06-28T12:15:23.681Z