Reproduce “Geometric Structural Knowledge Graph Foundation Model” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Ling Xin, Mojtaba Nayyeri, Zahra Makki Nayeri, Steffen Staab

Gamma enables zero-shot missing-link prediction across graphs with unfamiliar entities and relations. It uses relation-graph message passing, then parallel entity-graph branches whose outputs are weighted by the query; its best-performing configuration uses complex and split-complex branches. Across 53 evaluation graphs, average MRR is 0.376 versus ULTRA’s 0.360, a 4.4% relative gain, while transductive-category average MRR remains at 0.310.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.