Reproduce “Bootstrapping Heterogeneous Graph Representation Learning via Large Language Models: A Generalized Approach” 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

Hang Gao, Chenhao Zhang, Fengge Wu, Junsuo Zhao, Changwen Zheng, Huaping Liu

GHGRL uses an LLM to estimate node-attribute categories and guide a GNN’s message passing without requiring a hand-specified schema. Its estimates need not match true types perfectly in tested settings. On the constructed IMDB-RIR benchmark at r=20%, GHGRL reports Macro-F1 75.15 ± 0.43 and Micro-F1 75.35 ± 0.77, versus 71.83 ± 0.23 and 72.18 ± 0.47 for PSHGCN + LLM.

Summary by alphaXiv.

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